PDF Key Technologies of Intelligentized Welding Manufacturing The Spectral Diagnosis Technology For Pulsed Gas Tungsten Arc Welding of Aluminum Alloys Yiming Huang Ebook Full Chapter

Key Technologies of Intelligentized
Welding Manufacturing The Spectral

Diagnosis Technology for Pulsed Gas
Tungsten Arc Welding of Aluminum
Alloys Yiming Huang
Visit to download the full and correct content document:
https://textbookfull.com/product/key-technologies-of-intelligentized-welding-manufactu
ring-the-spectral-diagnosis-technology-for-pulsed-gas-tungsten-arc-welding-of-alumin
um-alloys-yiming-huang/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Key Technologies of Intelligentized Welding

Manufacturing Visual Sensing of Weld Pool Dynamic
Characters and Defect Prediction of GTAW Process
Zongyao Chen
https://textbookfull.com/product/key-technologies-of-
intelligentized-welding-manufacturing-visual-sensing-of-weld-
pool-dynamic-characters-and-defect-prediction-of-gtaw-process-
zongyao-chen/
Friction Stir Welding : Dissimilar Aluminium Alloys

First Edition Noor Zaman Khan
https://textbookfull.com/product/friction-stir-welding-
dissimilar-aluminium-alloys-first-edition-noor-zaman-khan/
Advances in Welding Technologies for Process

Development Vishvesh J Badheka
https://textbookfull.com/product/advances-in-welding-
technologies-for-process-development-vishvesh-j-badheka/
Welding : principles and practices Bohnart
https://textbookfull.com/product/welding-principles-and-
practices-bohnart/
Welding stainless steel questions and answers a guide
for troubleshooting stainless steel welding related
problems First Edition. Edition Kotecki
https://textbookfull.com/product/welding-stainless-steel-
questions-and-answers-a-guide-for-troubleshooting-stainless-
steel-welding-related-problems-first-edition-edition-kotecki/
Key Technologies of Magnetically Coupled Resonant

Wireless Power Transfer 1st Edition Yiming Zhang
(Auth.)
https://textbookfull.com/product/key-technologies-of-
magnetically-coupled-resonant-wireless-power-transfer-1st-
edition-yiming-zhang-auth/
Machining of Light Alloys: Aluminum, Titanium, and

Magnesium 1st Edition Diego Carou (Editor)
https://textbookfull.com/product/machining-of-light-alloys-
aluminum-titanium-and-magnesium-1st-edition-diego-carou-editor/
Automation in the Welding Industry: Incorporating

Artificial Intelligence, Machine Learning and Other
Technologies 1st Edition Syed Quadir Moinuddin
https://textbookfull.com/product/automation-in-the-welding-
industry-incorporating-artificial-intelligence-machine-learning-
and-other-technologies-1st-edition-syed-quadir-moinuddin/
Encyclopedia of Aluminum and Its Alloys, Two-Volume Set

1st Edition George E. Totten
https://textbookfull.com/product/encyclopedia-of-aluminum-and-
its-alloys-two-volume-set-1st-edition-george-e-totten/
Yiming Huang
Shanben Chen
Key Technologies of
Intelligentized Welding
Manufacturing
The Spectral Diagnosis Technology for
Pulsed Gas Tungsten Arc Welding of
Aluminum Alloys
Key Technologies of Intelligentized Welding
Manufacturing
Yiming Huang Shanben Chen
•
Key Technologies
of Intelligentized Welding
Manufacturing
The Spectral Diagnosis Technology
for Pulsed Gas Tungsten Arc Welding
of Aluminum Alloys
123
Yiming Huang Shanben Chen
School of Materials Science and Engineering Shanghai Jiao Tong University
Tianjin University Shanghai, China
Tianjin, China
ISBN 978-981-13-7548-4 ISBN 978-981-13-7549-1 (eBook)

https://doi.org/10.1007/978-981-13-7549-1
Library of Congress Control Number: 2019935997
© Springer Nature Singapore Pte Ltd. 2020

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
As a cross-disciplinary study of arc physics, heat transfer, metallurgy, and

mechanics, welding is an indispensable material processing method in industrial
manufacturing. With the rapid development of social economy and industrial
technology, the characteristic of intelligence in manufacturing is becoming
increasingly apparent. Up to 2017, the penetration rate of digital design tools, the
numerical control rate of key processes, and the digitization rate of production
equipment in China have reached 63.3, 46.4, and 44.8%, respectively. A new round
of technology and industrial revolution characterized by digitalization, networking,
and intelligence in manufacturing is in the ascendant, and intelligent manufacturing
has become an important direction for the high-quality development of the indus-
trial economy. In order to improve the total factor productivity in the production
process to meet the needs of the development of green low carbon cycle, it is an
inevitable trend to upgrade the traditional welding technology.
In order to make the welding technology developed in a safe, reliable,
cost-effective, and green direction, it is generally necessary to start from the fol-
lowing aspects: First, it is to propose a new type of welding manufacturing tech-
nology; second, it is to change the postproduction inspection mode of the current
welding process. The new welding technology has always been the research focus
of scientific workers. However, there is still a long way to go from being practical
due to the limitation of the current technical level. Online detection and real-time
control of the welding process can effectively reduce the manpower and material
resources required for non-destructive testing after welding, which is expected to
realize the digital transmission of the whole life cycle of the manufacturing process,
such as design, fabrication, and repair. Thus, it has always been the research hot
spot in welding industry.
As the basis on which online monitoring relies, the quality of information on
welds is always the focus of attention. Therefore, the time and spatial resolution
of the weld information, the degree of consistency, reconfigurability, secondary
development, and the visual form of the information are made a claim. At present,
the acoustic emission, arc voltage, molten pool image, and arc emission spectrum
of the welding process obtained by using the sensor online have been proved to
v
vi Preface
effectively express the information of welding parameters, joint quality, product

service, and so on. Among them, the arc emission spectrum is the most abundant
signal source in the welding process. It not only contains the metal elements in the
welding consumables, but also contains gas components such as nitrogen, hydro-
gen, and oxygen. Furthermore, it has the unique function to describe the arc tem-
perature and the particle density of the arc column. Besides, it has the advantage of
not being in direct contact with the arc. Therefore, the arc spectrum has increasingly
become the focus of research in the field of welding.
The content of this book is mainly based on the research results on the moni-
toring and control of the welding process using spectral analysis performed by the
authors and their research team. The theory, experiment, and application of arc
spectral analysis in the welding process of aluminum alloys are summarized,
analyzed, and summarized.
The book is divided into seven chapters. Chapter 1 is a review which mainly
describes the research status of sensing technology in the welding process, focusing
on the application of spectral detection in the field of welding. Chapter 2 introduces
experimental instruments, mainly including the welding system and a self-developed
spectral sensing system. Chapter 3 introduces the theoretical basis of spectral
detection, including the selection of spectral lines and the influence of welding
process parameters on the spectral lines. Chapter 4 mainly describes the feature
extraction based on the spectral lines of interest and introduces the empirical mode
decomposition method and its application in arc spectral data. Furthermore, the
effectiveness of the method for detecting porosity defects by the spectral analysis is
proved by experiments. Chapter 5 focuses on the data reduction technique on the full
arc spectral band. First, the application of the traditional linear dimensionality
reduction and advanced manifold learning on arc spectral data is discussed,
respectively. Then, the orthogonal local linear discriminant projection algorithm
based on the maximum margin criterion is introduced. Chapter 6 proposes the
mechanism model and control method of the porosity defect. Based on the spectral
features proposed in the previous section, the real-time control of the welding
process is achieved by the intelligent control technology. This chapter content
includes the nucleation and growth model of the porosity, controller design, simu-
lation, and control test. Chapter 7 is the conclusion.
This book is written on the basis of the research results supported by the
National Natural Science Foundation of China under Grant No. 51575349 and
No. 51075268. I would like to express my gratitude to the robot welding intelligent
technology laboratory (Shanghai Jiao Tong University) that provided me with
learning opportunities and experimental conditions. I would also like to thank my
supervisor, Prof. Shanben Chen, for introducing me into the door of intelligent
welding and giving me meticulous guidance. Thanks to Dr. Tao Lin, Dr. Huabin
Chen, Dr. Yanling Xu, Dr. Na Lv, Dr. Zhifen Zhang, and other teachers of the
laboratory who taught me professional skills and introduced me to the frontier
knowledge of this field. I would like to express my gratitude to the colleagues in
welding institute (Tianjin University) for their care and support to me, especially
the teachers in our team, such as Dr. Shengsun Hu, Dr. Lijun Yang, Dr. Huan Li,
Preface vii
Dr. Zhijiang Wang, and Dr. Junqi Shen. I would like to thank my family for their
understanding of my work, especially my wife, Ms. Yinying Chen, for her
companionship.
Due to the limited knowledge level of the author and complex interdisciplinary
knowledge involved in the research of monitoring and control in welding manu-
facturing, we hope that experts, scholars, and readers who read this book will be
kind enough to comment on any improper or incorrect information contained in this
book.
Tianjin, China Yiming Huang

Shanghai, China Shanben Chen
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Welding Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 The Principle of the Welding Process . . . . . . . . . . . . . . . . 1
1.1.2 Gas Tungsten Arc Welding and Improvement Process . . . . 2
1.2 Welding Defects of Aluminum Alloys . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Application Range of Aluminum Alloys . . . . . . . . . . . . . . 3
1.2.2 Study on Weldability of Aluminum Alloys . . . . . . . . . . . . 4
1.3 Intelligent Welding Manufacturing Technology . . . . . . . . . . . . . . 5
1.3.1 Sensing Technology Based on Electrical Signals . . . . . . . . 5
1.3.2 Sensing Technology Based on Sound Signals . . . . . . . . . . 6
1.3.3 Sensing Technology Based on Visual Images . . . . . . . . . . 7
1.3.4 Sensing Technology Based on Arc Spectrum . . . . . . . . . . 8
1.4 Challenges of Spectral Sensing Technology . . . . . . . . . . . . . . . . . 17
1.4.1 Data Dimension Reduction . . . . . . . . . . . . . . . . . . . . . . . . 17
1.4.2 Welding Control Technology . . . . . . . . . . . . . . . . . . . . . . 19
1.5 Conclusion Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2 Welding Arc Spectrum Sensor and Welding System . . . . . . . . . . . . . 27
2.1 Welding System for Multi-information Acquisition . . . . . . . . . . . . 27
2.1.1 Selection of Welding Methods . . . . . . . . . . . . . . . . . . . . . 27
2.1.2 Welding System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.3 Spectral Information Acquisition System . . . . . . . . . . . . . . 31
2.2 Spectral Sensing System Based on Linear Array CCD . . . . . . . . . 32
2.2.1 Design of Optical System . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.2 Determination and Calibration of Parameters . . . . . . . . . . . 36
ix
x Contents
3 Basic Characteristics of Arc Spectrum in P-GTAW Process . . . . . . . 43

3.1 Basic Principle of Arc Spectrum Detection . . . . . . . . . . . . . . . . . . 43
3.1.1 Selection of Characteristic Spectral Lines . . . . . . . . . . . . . 44
3.1.2 The Variation Rule of Characteristic Spectral Lines . . . . . . 46
3.2 Spectral Line Selection Based on Improved Clustering
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2.1 K-Medoids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2.2 Determination of the Number of Categories . . . . . . . . . . . 49
3.2.3 Similarity Measure Function . . . . . . . . . . . . . . . . . . . . . . . 51
3.2.4 Initial Center Point Selection . . . . . . . . . . . . . . . . . . . . . . 53
3.2.5 Iterative Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.2.6 Validity Analysis of Clustering Results . . . . . . . . . . . . . . . 55
3.2.7 Cluster Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.3 Influence of Welding Process Parameters on Arc Spectrum . . . . . . 58
3.3.1 Wavelength Characteristics of Arc Spectral Information . . . 58
3.3.2 Influence of Welding Current on Arc Spectral
Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 59
3.3.3 Influence of Argon Flow Rate on Arc Spectral
Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.3.4 Influence of Arc Length on Arc Spectral Information . . . . 63
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4 Monitoring of Welding Process Based on Characteristic
Spectral Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.1 Calculation of Electronic Temperature . . . . . . . . . . . . . . . . . . . . . 67
4.1.1 Solving Te Based on Boltzman Method . . . . . . . . . . . . . . 68
4.1.2 The Spatial Distribution of Te . . . . . . . . . . . . . . . . . . . . . . 69
4.2 Feature Extraction Based on Hydrogen Spectral Lines . . . . . . . . . 74
4.2.1 Feature Extraction of Te Curve . . . . . . . . . . . . . . . . . . . . . 74
4.2.2 Feature Extraction Based on EMD . . . . . . . . . . . . . . . . . . 80
4.3 Prediction of Hydrogen Porosity Under Different Welding
Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 88
4.3.1 The Effect of Groove Angle . . . . . . . . . . . . . . . . . ...... 89
4.3.2 The Influence of the Thickness of Root Face
on the Pores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.3.3 The Influence of the Gap Width on the Porosity . . . . . . . . 99
4.3.4 Significance Analysis of Influencing Factors . . . . . . . . . . . 105
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Contents xi
5 Dimension Reduction of Arc Spectrum for Porosity Detection

in P-GTAW Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.1 Feature Extraction from Arc Spectrum Information . . . . . . . . . . . . 111
5.1.1 Comparative Study on Dimension Reduction Methods . . . . 114
5.1.2 Research on LLE Algorithm . . . . . . . . . . . . . . . . . . . . . . . 117
5.1.3 Relationship Between Characteristic Signals and Porosity
Defects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.2 Orthogonal Local Linear Discriminant Projection Algorithm
Based on MMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.2.1 Maximum Margin Criterion . . . . . . . . . . . . . . . . . . . . . . . 125
5.2.2 Improved LLE Algorithm Based on MMC . . . . . . . . . . . . 127
5.2.3 Orthogonalized Eigenvector . . . . . . . . . . . . . . . . . . . . . . . 129
5.3 Porosity Classification Based on Decision Fusion . . . . . . . . . . . . . 132
5.3.1 Statistical Analysis of H I Spectral Lines . . . . . . . . . . . . . 132
5.3.2 GA-Based SVM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6 Formation Mechanism and Control of Porosity in GTAW
of Al Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.1 Formation Mechanism of Hydrogen Porosity . . . . . . . . . . . . . . . . 137
6.1.1 Nucleation Process of Bubble . . . . . . . . . . . . . . . . . . . . . . 138
6.1.2 Bubble Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
6.2 Observation of Hydrogen Porosity in the Weld . . . . . . . . . . . . . . 153
6.3 Porosity Control Based on Fuzzy-PID Adaptive Control . . . . . . . . 155
6.3.1 PID Control Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . 156
6.3.2 Random Test Design and Model Structure
Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
6.3.3 Control Simulation Single Input Single Output . . . . . . . . . 160
6.3.4 Fuzzy-PID Control Algorithm and Its Simulation . . . . . . . 160
6.3.5 Experimental Verification . . . . . . . . . . . . . . . . . . . . . . . . . 163
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Abbreviations
AC Alternating current
AVC Automatic voltage control
CCD Charge-coupled device
DCEP Direct current electrode positive
EM Expectation maximization
EMD Empirical mode decomposition
FA Factor analysis
GA Genetic algorithm
GTAW Gas tungsten arc welding
IMF Intrinsic mode functions
LDA Linear discriminant analysis
LLE Local linear embedding
LTE Local thermodynamic equilibrium
MIG Melt inert gas
MMC Maximum margin criterion
MST Minimal spanning tree
OLLDP Orthogonal local linear discriminant projection algorithm
PCA Principal component analysis
P-GTAW Pulsed gas tungsten arc welding
PMT Photomultiplier tube
SA Spectral angle
SCC Spectral correlation coefficient
SD Spectral distance
SID Spectral information divergence
SVM Support vector machine
TIG Tungsten inert gas
WPT Wavelet packet transformation
xiii
Chapter 1
Introduction
Abstract Aluminum alloys have promising application in automobile and aircraft

industries due to their high specific strength and good corrosion resistance. However,
porosity is easily produced during welding solidification due to the significant differ-
ence of hydrogen solubility in liquid and solid alloy, which is detrimental to welding
quality. Because of existence inside the weld seam, porosity has to be detected by
destructive test or nondestructive test, resulting in increasing the difficulty to meet
the need of efficient production with high quality for modern manufacturing. To
solve this problem, this chapter proposes a suitable method for on-line detection
of porosity defects during aluminum alloy arc welding, on the basis of introducing
the current intelligent welding technology. Furthermore, the chapter points out the
challenges the method faces.
Manufacturing is the mainstay of the national economy and a basic and strategic
industry for national construction. Since the 21st century, high-tech technologies
such as information technology, new materials technology, and aerospace technol-
ogy have been rapidly developed and widely used, which leads the vigorous devel-
opment of manufacturing industry worldwide [1, 2]. Welding is an indispensable
method of material processing in manufacturing [3]. With the synchronous advance-
ment of new industrialization and information technology, the deeply integration of
new generation of information technology and manufacturing technology has been
greatly promoted. The advanced intelligent welding technology combining informa-
tion technology, material science technology and welding manufacturing technology
has emerged [4–6].
1.1 Welding Technology
1.1.1 The Principle of the Welding Process
Welding is a process in which the weldment is atomically bonded by heating, or

pressurizing, or combination of both and with or without a filling material [7, 8].
© Springer Nature Singapore Pte Ltd. 2020 1

Y. Huang and S. Chen, Key Technologies of Intelligentized Welding
Manufacturing, https://doi.org/10.1007/978-981-13-7549-1_1
2 1 Introduction
Since the distance between internal atoms of solid materials such as metals (lattice)
is very small and a strong bond is formed between the atoms, they can maintain a
fixed shape as a whole. To join two separated metal components into a whole by the
action of atomic binding force, two difficulties need to be overcome:
(1) The connecting surface is not flat. Even with the most sophisticated machining,
the surface flatness can only reach the micron level, which is still far greater
than the order of magnitude 10−14 required for atomic union.
(2) The oxide film and other contaminants on the surface hinder the proximity of
the metal surface atoms to the lattice distance and form a bonding force. The
welding process is the process of overcoming these two difficulties.
1.1.2 Gas Tungsten Arc Welding and Improvement Process
According to the character of weld metal bonding, welding methods are generally
divided into three categories, namely, fusion welding, solid phase welding and braz-
ing. Each category can be divided into several types according to the welding heat
source and other obvious features. There are nearly 30 basic welding methods by
category.
These welding methods combine with metal cutting, hot cutting, pressure process-
ing, casting, heat treatment and other processing methods to form a comprehensive
material processing technology, which is the basic processing method of the modern
material processing industry. Among the various welding methods, arc welding is
the most widely used and has the largest proportion in welding production [9, 10].
According to statistics, the proportion of arc welding in the total amount of welding
production is more than 60% in some industrialized countries.
Arc welding is a kind of fusion welding. From the physical point of view, fusion
welding is to heat and melt the metal to be joined without pressure. During the
welding process, the atoms are become closed to each other by the flow of liquid
metal, which is transformed into solid state after solidification. According to the
electrodes used, arc welding is divided into two types: molten arc welding and non-
melting arc welding. The consumable electrode welding is a method of arc welding
using a metal wire as an electrode to simultaneously melt a filler weld [11], which
includes welding rod arc welding, submerged arc welding, melt inert gas (MIG)
welding, etc. Arc welding with non-consumable electrode is an arc welding method
using an infusible electrode for welding [12], which includes tungsten inert gas (TIG)
welding, plasma arc welding, etc.
Tungsten inert gas welding, which is also called gas tungsten arc welding (GTAW),
is an arc welding method using tungsten or tungsten alloy as an electrode and argon
gas as a shielding gas. Depending on the groove form of the weld, filler metal may
or may not be added. The filler metal is usually added from the front of the arc or it
can be preset between the groove or gap of the joint. Gas tungsten arc welding has
the following advantages [13–15]:
1.1 Welding Technology 3
(1) The protection is good. During the welding, the entire welding zone including
the tungsten electrode, the arc, the molten pool, the end of the filling wire
and the surface of the weldment near the molten pool are protected by argon
gas, which not only avoids oxidation and nitridation of the weld metal affected
by the surrounding air, but also eliminates the source of hydrogen. Therefore,
the weld metal is pure and the hydrogen content is small. In addition, since
argon gas neither chemically reacts with metal nor dissolves in liquid metal, the
metallurgical reaction of the molten pool during the welding process is simple
and easy to control, which provides good conditions for obtaining high quality
welds.
(2) The welding process is stable. The arc is very stable in argon after being ignited,
and the argon arc can be stably burned even at a lower arc voltage. Argon is a
monoatomic gas which does not decompose or absorb heat at high temperatures.
Therefore, the arc burned in argon has less heat loss, and the heat and force of
the arc acting on the electrode and the molten pool are substantially constant.
There is no droplet transition in the arc and the welding process is very stable.
(3) The weld is well formed. Neither oxygen intrusion nor chemically active reac-
tion occurs in the welding process, the surface tension is large. The heat input
is easy to adjust during the welding process, which is especially suitable for the
welding of thin plates and all-position welding. It is also an ideal method for
double-sided forming of single-sided welding. Since no spatter occurs during
welding, the weld bead is well formed. In addition, since TIG welding is an open
arc welding without droplet transfer, it is easy to be mechanized and automated.
1.2 Welding Defects of Aluminum Alloys
1.2.1 Application Range of Aluminum Alloys
As a low-density, high-specific strength structural material, aluminum alloy has good

corrosion resistance, good electrical and thermal conductivity [16, 17]. After more
than one hundred years of development, it has become the second largest metal in the
world after steel and it is the most widely used light alloys in the industry. Especially
in the middle of the 20th century, the application of aluminum alloys is shifted from
military to civilian. They were widely used in various fields of national production
and became an indispensable material in the construction of national basic economy.
In developed countries, the transportation industry is the industry that uses the
most aluminum. Aluminum alloys are the first choice for lightweight vehicles due to
their low density and high energy absorption per unit mass. At present, lightweight
is an important means of energy saving and emission reduction in the automotive
industry, which will inevitably lead to a large number of applications of aluminum
alloys in automobiles. In addition, aluminum alloys are also widely used in the
4 1 Introduction
manufacture of aircraft structures and launch vehicle arches [18, 19], such as skins,
siding, cabinets and so on.
China is the largest aluminum producer and consumer in the world. The alu-
minum processing industry is in a high-speed growth period. According to the “13th
Five-Year Plan” of the development of the non-ferrous metal industry of the Min-
istry of Industry and Information Technology, the compounded annual growth rate
of aluminum processed materials output from 2016 to 2020 will remain at 5.92%
according to the estimated production of 40 million tons of aluminum by 2020.
By 2020, China’s total aluminum consumption will reach 43 million tons, and the
compounded annual growth rate of 2016–2020 will reach 7.24%.
1.2.2 Study on Weldability of Aluminum Alloys
With the rapid development of the industrial economy, the demand for aluminum
alloy welded structural parts is increasing, leading to the intensive research on the
weldability of aluminum alloys. As is known, aluminum alloys have strong mechan-
ical properties and are easy to process [20]. However, due to the large specific heat
capacity and thermal conductivity, it is easy to oxidize and adsorb moisture in the air,
which causes stress deformation, oxide inclusions, pores and cracks in the weld [21,
22]. The porosity defect is one of the most difficult internal defects to be detected and
controlled in the welding of aluminum alloys. It not only affects the airtightness of the
aluminum alloy weld, but also reduces the effective working area and bearing area of
the weldment, thereby causing stress concentration and greatly reducing the fatigue
strength and toughness of the weld [23]. Crack defection can also be triggered when
the porosity defects are severe. Through the research and statistics on the defects on
the X-ray flaw detection film of the weld, it was found that most of them were pores
(about 79%), followed by slag inclusion, incomplete penetration and crack [24]. In
addition, the pores are common in various welding methods, including laser welding,
MIG welding and TIG welding.
All the internal defects of welding, especially the detection of air holes, have been
carried out after welding. Destructive testing or non-destructive testing is required.
However, the testing process is time-consuming and laborious, which greatly restricts
the welding production efficiency and is difficult to make up the porosity defects in
real time. With the rapid development of manufacturing industry, the “13th Five-Year
Plan” clearly proposes to promote the deep integration of industry and information,
and promote the development of “intelligent manufacturing”. Therefore, the online
quality inspection of the welding process is required to meet higher standards. Effi-
cient, reliable and stable testing methods are bound to be the future development
direction. Therefore, the development of real-time testing technology for welding
dynamic process and weld quality is of great significance for ensuring the con-
sistency and high reliability of welding products and realizing the automation and
intelligence of the welding process. As a result, intelligent welding manufacturing
technology came into being.
1.3 Intelligent Welding Manufacturing Technology 5
1.3 Intelligent Welding Manufacturing Technology
The intelligent welding manufacturing technology is proposed in the environment

facing the challenge of strengthening the resource environment and the cost rising in
the manufacturing industry. The meaning is realizing advanced welding technology
based on the sensing technology and information processing technology, by combin-
ing artificial intelligence and traditional welding methods [25–27]. The information
is the most critical part. Modern cybernetics defines information as “the exchange
of content in the process of interaction and interrelationship between the cognitive
subject and the external environment”. The information of manufacturing, such as
welding, is a specialized area of information. Its nature and attributes have a com-
monality with information. At the same time, it has its own characteristics in the
field. The welding manufacturing process has not only a multi-field of information,
but also a strong coupling of parameters and random interference.
The rapid development and application of new technologies, such as computers,
electronics, and information, have promoted the transformation of traditional man-
ufacturing into a process with information-based digital and intelligent characteris-
tics, meaning an era of information-driven modern manufacturing. From the view
of information, numbers are symbols used to represent the quantitative relationship
between things and things. They are the carrier of information and the expression of
its physical meaning. The effective transmission of digital information can be realized
through the network, which is a technical approach to drive manufacturing activi-
ties. Therefore, information technology-based digital equipment and manufacturing
environments with related digital resources support are the basis of digital manufac-
turing technology, which gradually transforms traditional manufacturing experience
into an intelligent manufacturing technology which is recordable, quantifiable and
comparable evaluation.
With the improvement of the requirements for high quality, high efficiency of
welding products, the acquisition and processing of quality information in the product
manufacturing process have become an important part of the welding manufacturing.
After several decades of research, sensor monitoring technology has made great
progress in welding electrical signals, acoustic signals, and visual signals [28–32].
1.3.1 Sensing Technology Based on Electrical Signals
The welding current is the most intuitive electrical signal data in the welding process.
Hall sensors are usually used to achieve the welding current collection. The Hall sen-
sor is a detection element based on Hall effect, with advantages of high precision,
good linearity, high bandwidth, and large measurement range. The basic working
principle is that when the primary current passes through the conductor, the primary
magnetic field is generated. At this time, the secondary coil generates the correspond-
ing secondary magnetic field for compensation of magnetic field. The Hall element
outputs a compensation current, and the compensation current can accurately reflect
the primary current and voltage.
6 1 Introduction
In the arc welding process, various types of defects will occur if the welding
process is unstable. The welding current is sensitive to the fluctuation [33, 34]. In
addition, when the relative position of the welding torch and the workpiece changes
during the welding process, it usually causes static and dynamic changes of the arc
signal. This change can also be extracted as a characteristic signal to achieve tracking
control of both vertical and horizontal directions. This method has been applied in the
industry and is called Automatic Voltage Control (AVC) method, which is to send the
extracted arc voltage signal to the AVC controller for data processing and compare
it with the given arc voltage value. The deviation value is calculated and converted
to the adjustment value in vertical movement of the welding torch, determining the
direction and strength of the driving signal output to the vertical motion mechanism
of the torch. As early as 1983, some scholars studied the relationship between arc
voltage and arc length. After more than 30 years, the research on arc electric signal
was continuously improved, and the automatic detection and real-time control of
weld position was realized [35–38]. The arc sensor has the characteristics of simple
structure, low cost, good real-time performance and fast response. It has a good
application in weld tracking, but it cannot reflect the penetration and formation of
the weld.
It is worth mentioning that the plasma electrical signal is also one of the com-
monly used detection methods in the laser welding process. Zhao et al. [39] detected
plasma electrical signals and analyzed in the frequency domain by power spectral
density analysis. They concluded that there were significant differences between the
fluctuating characteristics of the plasma under different welding modes. Qiu et al.
[40] found the corresponding relation between plasma plume shape and electrical
signal by analyzing the fluctuation of the plasma plume shape and the change of the
detected electrical signal.
1.3.2 Sensing Technology Based on Sound Signals
Sound sensing information is a very important source of information in the weld-

ing process, which contains a wealth of welding status information and is one of
the important signals for monitoring welding quality. In the arc welding process,
along with the generation of arc sound, the arc behavior, process stability, welding
parameters, and droplet transfer mode are all closely related to the welding sound
information [41, 42]. Tam et al. [43] put a headset on the welding operator during
the test and sent the welding sound to the welder with a delay. As time went on, the
welding quality of the welder operation became worse and worse. The experiment
showed that the sound signal of the welding process has a certain influence on the
welder judging welding quality. Lv et al. [44–46] proposed a set of algorithms for
extracting the features on the weld penetration status of the TIG welding based on
the arc acoustic signal, and realized the extraction of a 36-dimensional eigenvector
group in different penetration states. They established a BP Adaboost neural network
for weld penetration prediction model with a recognition rate of 94%, based on the
36-dimensional eigenvector group.
In general, the current research on welding sound is mainly concentrated in plasma

perforation welding [47], as well as the detection of droplet transfer in the GMAW
process. There is little research on the GTAW process and its application is extremely
limited.
1.3.3 Sensing Technology Based on Visual Images
Vision plays an important role in the process of human perception of the external envi-
ronment. It is estimated that external information obtained through visual accounts
for more than 80% of the total amount of external information obtained by humans.
In the field of welding, visual information is also the key to the welder’s work in
welding. Welding visual information in the welding dynamic process imitates the
welder’s human eye function, with the advantages of intuitive display and mature
technology. Image processing based on the acquired visual information can filter out
arc, spatter, smoke and other interference to a certain extent, so as to obtain some
key information in the welding process.
The visual signal is mainly collected by means of arc light and auxiliary light
source to observe the visible area of the molten pool. The amount of information
collected is rich, including the melting width and melting length of the weld pool,
which can directly reflect the welding quality [48–50]. Depending on the light source,
vision sensing systems are generally divided into two types: active vision and passive
vision [51, 52].
The active vision sensing method illuminates the welding zone by means of an
auxiliary light source, which can eliminate the interference of the arc light on the
welding image during the welding process. Among active vision sensors, the laser
structured light sensor is the most widely used. Feng et al. [53] collected and analyzed
the welding current and voltage by using current and voltage sensors, while collecting
images of GMA welding pool by high-speed camera. The study mainly analyzes the
influence of welding speed on the droplet transfer behavior of the welding process.
The dynamic behavior of the droplet transfer can be clearly observed by the molten
pool images captured by high-speed camera with a sampling frequency of 5000
frame per second. Although the active image sensor has good image quality, the cost
is generally expensive, which limits the wide application of this method in practice.
Unlike active vision sensing technology, passive vision directly uses arc light to
illuminate the weld and pool area. Before entering the CCD camera, the arc must be
filtered to avoid interference from the strong arc to the weld image. Xu et al. [54]
designed a special filter system to block the glare to obtain a clear outline of the weld
front end image. Furthermore, they applied this sensor to the weld seam tracking
and guidance of gas tungsten arc welding and realized the tracking of straight and
curved welds in the welding process. Scholars from Shanghai Jiao Tong University
[55–57] designed a binocular vision sensing system according to the characteristics
of aluminum alloy molten pool, which can realize the weld-seam correction before
welding and the observation of the behavior of the weld pool during the welding
process. Moreover, a predictive model of back weld width on the basis of the front
information is established. In addition, considering the importance of backside melt-
8 1 Introduction
ing information for real-time penetration monitoring, a three-light visual sensing

system is designed to obtain not only the geometry of the front molten pool, but also
the penetration form of the back side [58, 59].
In general, visual sensing technology has the advantages of being intuitive and
informative, and is the most widely used in industrial production. However, due to the
large equipment and limited acquisition principle, it is difficult to collect clear image
under special working conditions. The subsequent processing is also very difficult.
1.3.4 Sensing Technology Based on Arc Spectrum
The wavelength and intensity information obtained by the spectrometer can be used
to qualitatively and even quantitatively analyze the elements. In view of the unique
advantages of spectroscopic analysis, it has been widely used in geology, metallurgy,
machinery, environment, materials, life and medicine [60, 61]. In recent years, atomic
emission spectrometry has attracted the increasing attention of welding workers.
Since the arc spectrum is informative and has no excellent contact with the welding
system, it will be the most promising new method for real-time detection of weld
defects [62, 63].
Under normal conditions, the metal atoms in the base metal and the wire and the
atoms in the shielding gas are in the lowest energy ground state. When excited by
arc energy, the atom jumps from the ground state to the excited state. The atoms in
the excited state are very unstable, and they will emit energy in the form of radiation
while transitioning to the lower energy level according to the selection rule, thus
forming an emission spectrum [64]. Arc emission spectroscopy is the most abundant
signal source in the welding process. It not only contains the metal elements in
the welding consumables [65, 66], but also contains nitrogen, hydrogen, oxygen
and other gas components. The above information is beneficial to understand the
physical characteristics of the arc, such as arc temperature and particles density in
the arc column. Since the arc plasma has the function of transmitting energy, when the
arc state changes, its heat output also changes, which may induce welding defects.
Therefore, the monitoring and quality of the welding process can be realized by
collecting and analyzing the arc spectrum [67–69].
At present, the research on arc spectral information mainly focuses on two aspects:
(a) conducting arc plasma diagnosis, and indirectly measuring the parameters of
arc physical properties including arc temperature and electron density based on the
assumption that the arc-based plasma has a local thermodynamic equilibrium state
[70, 71]; (b) On-line detection of welding process based on extracted features, includ-
ing arc length, welding parameters, penetration status, etc.
1.3.4.1 Plasma Diagnosis
As a low-temperature (thermal) plasma, the temperature of the welding arc is dis-

tributed between 103 and 104 K. Its high-temperature characteristic limits the use
of detection methods such as thermocouples and infrared imagers. In order to gain
Fig. 1.1 a Schematic of experimental setup, b Spectral intensity under different gas flow rates [74]
a deep understanding of the physic properties of the arc, non-contact spectroscopy

has attracted the attention of researchers. The collection and analysis of its plasma
emission spectrum is the most basic research.
Wang et al. [72] explored the plasma characteristics of laser-MIG hybrid welding.
By collecting spectral signals, it was observed that the intensity of metal elements
is much stronger than that of single heat source arc welding, while the intensity of
protective gas is weak. Oezmert [73] et al. observed that the light intensity in the
830–1000 nm band is enhanced with the increase in penetration in the laser welding
of copper. Wang et al. [74] constructed a system to study the effect of auxiliary
shielding gas on laser welding plasma, as shown in Fig. 1.1a. The results show
that as the gas flow increases, the penetration increases, laser plasma light intensity
is reduced, as shown in Fig. 1.1b. The experiment shows that the addition of the
auxiliary gas compresses the plasma volume while increasing the energy conversion
efficiency.
Literature [75] shows the difference in the arc intensity of the normal weld and
the weld defect, as shown in Fig. 1.2. In literature [76], the intensity distributions
of Fe I (538 nm) and Ar I (696 nm) lines during GMA welding were obtained by
monochromator and high-speed camera, and the temperature distribution of the arc
was obtained by Fowler-Milne method. A lot of researches have been done in this
area in China. For example, Wang et al. [77] studied on pulsed TIG welding (the
protective gas is Ar) and DC TIG welding (protective gas is 80% Ar + 20% N2 ) by
Fowler-Milne method, using optical filters and CMOS sensors to build an arc spectral
information acquisition system based on high-speed photography. Figure 1.3a, b
show the calculated spectral intensity and arc temperature, respectively.
As the main characteristic parameter of plasma, electron temperature is an impor-
tant parameter for describing the state of the arc and plays an important role in the
application of plasma. There are many measuring methods, including laser scatter-
ing, speckle photography, Talbo-Mor method and spectroscopy, among which spec-
troscopy is the most widely used in the field of welding. It is known that a welding
arc can be regarded as a discharge phenomenon of plasma, and the temperature of
electrons, cations, and neutral particles is raised by collision with each other. During
the welding process, the electron temperature at each moment of the arc is measured
to form an electronic temperature curve Te . Each point on the curve Te corresponds to
the actual weld. The description of the plasma by the SaHa. M. N formula shows that
10 1 Introduction
Fig. 1.2 a Defective seam b plasma spectrum associated with correct seam, and c plasma spectrum
associated with the defect [75]
Fig. 1.3 Time variation of a intensity and b temperature in one cycle in the 125 A-200 A arc
welding [77]
Te is closely related to the ionization potential and electron density in the plasma, so
the arc information related to these parameters is related to the curve Te . Specifically,
the arc temperature during the welding process is mainly affected by parameters such
as arc current, arc voltage, shielding gas composition, and flow rate.
Using Boltzmann curve method and Stark broadening method, Liu and Hao [78,
79] measured arc spectrum in magnesium alloy arc welding and laser-arc hybrid
welding, respectively. Moreover, the influence of welding parameters on electron
temperature Te and electron density is explored, shown in Fig. 1.4 [80, 81]. Yu et al.
[82] investigated the relationship between electron temperature and arc length and
bath size during the P-GTAW process of aluminum alloys. It was found that electron
temperature and electron density would first increase and then decrease, when the
arc length increased from 2 to 4 mm. Thereby determining the arc length at which
Fig. 1.4 a Variation in the intensity of Mg atom lines with the arc current 60 A b 140 A c Charac-
teristic variation of Te with different arc currents d variation of electron density with different arc
currents [80, 81]
the heat input efficiency is maximum. In order to solve the problem that the electron
temperature cannot be solved due to insufficient number of line spectra, Zou et al.
[83] proposed a method based on continuous spectrum to solve the Te in laser welding
process. After obtaining the electron temperature, the researchers tried to use it for
monitoring the welding process.
Liu et al. [84] observed that the electron temperature is positively correlated with
the heat input. When the weld transferred from partially penetrated to complete
penetration, the molten pool moved down and the electron temperature increased
sharply. The characteristics of the weld seam state can be the feature to be detected.
In addition, they also use Te for the inspection of the welding process. Since when
the plasma and the molten pool are unstable, the weld will be discontinuous and the
electron temperature fluctuates, corresponding to the defects. Sibillano et al. [85] used
a covariance mapping technique to correlate electron temperature with weld defects
in real time, and to determine the presence of defects when the electron temperature
exceeds a set threshold. Harooni et al. [86] performed a zero-gap lap laser welding
test on magnesium alloys, using plasma sensors to collect plasma information, and
calculating the electron temperature, as shown in Fig. 1.5, to obtain the relationship
between the pores and the spectral signals. It is indicated that the oxide film causes
the generation of pores.
12 1 Introduction
Fig. 1.5 a The calculated electron temperature for laser induced plasma with laser power of 1200 W
b cross-sectional views of the welds without and c with oxide layer [86]
1.3.4.2 On-Line Detection Based on Spectral Signals
As mentioned above, the features extracted from the arc plasma can characterize the
weld state and are therefore increasingly applied to on-line monitoring of welding
quality. As a new means, arc spectroscopy has received a lot of attention in recent
years, and it has more vitality than traditional methods based on sensing technologies
of welding electrical signals and visual images [87–89]. Zhanget al. [90] extracted
8 spectral wavelength bands of interest and used wavelet packet decomposition to
achieve the purpose of de-pulse interference, analyzing the reflection of characteristic
parameters such as SNR, RMS, variance and kurtosis on oil contamination defects.
Figure 1.6 shows the correspondence between the characteristic signals extracted
from one of the spectra and the defects. Zhang and Chen [91] incorporate spectral
signals on the basis of voltage and sound signals, which greatly increases the recogni-
tion rate of welds. Sibillanoet al. [92] built the system, shown in Fig. 1.7, to calculate
the electron temperature under different laser source conditions, and obtained the
corresponding relationship with the penetration depth, as shown in Fig. 1.8. As the
depth of fusion increases, Te decreases. The author analyzed that it is related to the
placement of the fiber optic probe when acquiring spectral information.
In literature [93], the spectrometer (SM240-USB) was used to collect the spectral
data of the interaction surface of the molten pool in GMA welding, the arc Te was
calculated by the line intensity of Fe and Mn elements, and the welding defects
were monitored by Te . The sampling frequency of the spectrometer is 45 Hz, and
the welding speed is 10 mm/s. Therefore, each sampling period of the spectral data
corresponds to a length of 0.22 mm on the weld. It can be seen from Fig. 1.9 that
there is a good correspondence between the several typical welding defects and
Fig. 1.6 a Defect weld of seam oxidation 2 b SNR(SOI-7-D) = 14.26425 c SNR(SOI-7-K) =

9.58691 [90]
Fig. 1.7 Experimental set-up of the optical sensor used during Nd: YAG laser experiments [92]
electron temperature curves. Mirapeix et al. [94] collected arc information based on
photodiode DET36A and compared it with the performance of spectrometer BWTEK
BCR112E. The results showed that with the appropriate filter, the photodiode can
be made the same judging as the spectrometer for a specific defect, as shown in
Fig. 1.10. Mirapeix et al. [95] performed on-line weld quality assessment through Te
curve to detect common defects such as oxidation, infusibility, etc., and the change
of defects is different by electronic temperatures of different element, which can be
seen from Fig. 1.11. Based on this research, Cobo et al. [96] used Te to conduct on-
line inspection of welding quality in tube sheet welding of ENSA steam generators.
14 1 Introduction
Fig. 1.8 Fe(I) electron temperature as a function of the penetration depth (a) for CO2 laser welding
processes and (b) for LASAG Nd:YAG laser welding processes at a welding speed of 50 mm/s [92]
Fig. 1.9 Arc electronic temperature curve corresponding to different welding defections [93]
The defect in Fig. 1.12 is caused by discontinuities of the protection of the welding
area.
The research on arc spectroscopy for on-line welding defect detection also
includes: the literature [97] extracted the wavelength information corresponding to
the maximum intensity in the background spectrum, and used it as the characteristic
signal to effectively detect the weld defect; the literature [98] used the spectrometer
to obtain the spectral data of the arc and its spatial information. The defects were
detected without determining the wavelength of the emission spectrum, which pro-
vided a new idea for the detection of the spectrum; the literature [99] used an optical
smart sensor to measure the amount of plasma radiation around the welding arc and
performed real-time analysis, using Kalman filter to reduce measurement noise and
an intelligent fuzzy system to evaluate the welding quality.
Jia et al. [100] comparatively analyzed the spectral data of underwater welding
and air arc welding. The two sides are basically the same in light intensity except for
the 656.28 nm wavelength, which indicates that the underwater arc plasma contains
a large amount of H elements. Furthermore, in [101], the influence of the variation
of welding parameters on the hydrogen spectral information in the TIG welding
(a) (b)
photodiodes
spectrometer
(c)
Fig. 1.10 a experimental setup b plasma RMS profile (spectrometer) c photodiode voltage profile
(766 nm filter) [94]
Fig. 1.11 Identification of a defect with different species [95]

16 1 Introduction
Fig. 1.12 Detection of

defective weld a Welded
tube. b Associated T profile
[96]
of Al alloys was investigated. It was proposed that the H I 656.28 nm and Ar I

641.63 nm line intensity ratio I H /I Ar signals could be used to monitor the abnormal
hydrogen content in the arc. The method effectively eliminates the interference of
non-hydrogen factors on the hydrogen spectral line. The peak component of the I H /I Ar
signal can be used to predict the occurrence of porosity in the weld joint. Moreover,
the experiments were performed by man-made defects. The corresponding I H /I Ar
curve is measured, shown as Fig. 1.13. It can be seen that the I H /I Ar curve has peaks
that obviously exceed the threshold Q at the artificial pore position. The results of
the literature based on the I H /I Ar curve for porosity prediction are in good agreement
with the actual weld, which provides a basis for us to further explore the formation
mechanism, the monitoring and the online control of porosity defects.
Theoretically, the real-time detection of the electron temperature Te can reflect
the change of the arc state, thereby predicting specific welding defects. However, in
fact, the welding process is influenced by the uncertain factors, such as non-stationary
randomness, nonlinearity and environmental interferences. Thus, the Te curve cannot
always reflect the change of the welding process stably and accurately, and the
accuracy rate for defect prediction is low. It is necessary to use some mathematical
tools and intelligent classification methods to process the original data. Combing
the extracted useful features and Te , the accuracy of welding defect detection can be
improved to meet the requirements of online control applications.
1.4 Challenges of Spectral Sensing Technology 17
Fig. 1.13 Curve IH/IAr corresponding to surface defection on welding bead [101]
1.4 Challenges of Spectral Sensing Technology
From the above, the rich information of the arc spectral data helps us to explore the
nature of the welding process. However, the difficulty of data mining for spectral
data is increased due to its large scale and complex structure, leading to reduced
computational rates, “dimension disasters” caused by high-dimensional features,
and so on. How to efficiently and accurately mine the effective information from
rich spectral data is one of the most important steps in data analysis. Secondly,
how to effectively control the welding process according to the acquired spectral
characteristics is still a big problem. At present, most control methods in the field of
welding focus on arc sensing and visual sensing, and remarkable research has been
made. Nevertheless, the research on arc spectrum is still very little. Next, the current
research situation will be introduced around these two aspects.
1.4.1 Data Dimension Reduction
There are many kinds of methods of data dimensionality reduction, which are widely
used in the fields of pattern recognition and data analysis. According to the distribu-
tion structure of the data, these methods can be divided into two categories, namely, a
linear method and a nonlinear method [102]. Essentially, both of them map data from
a high-dimensional measurement space map to a low-dimensional feature subspace.
The linear dimension reduction method is performed by linear combination of fea-
tures. The widely used methods include principal component analysis (PCA), factor
18 1 Introduction
analysis (FA) and linear discriminant analysis (LDA) [103], all of which assume that
the data obeys typical Gaussian distribution. The main idea of PCA is to synthesize
a set of new and unrelated variables by linear transformation [104]. The characteris-
tics of these variables are arranged in descending order of variance. Only variables
with the sum of variance contributions greater than 0.85 are selected to represent the
original data, achieving the purpose of dimension reduction. PCA strives to maintain
the essence of the original information after dimension reduction, and compares the
variance of the projection direction to determine the importance of the data in the
direction. This operation may make the data sample grouping together, making it
impossible to distinguish between categories. As a result, the features obtained by
PCA do not perform well in the established classification model. Different from the
PCA algorithm, the core of LDA is to make the new features of the mapping to
be distinguished as easy as possible. The projection direction it finds can make the
same type of data more compact, and the heterogeneous data more and more discrete
[105]. Factor analysis assumes that low-dimensional data points satisfying Gaussian
analysis generate high-dimensional data after double action of linear transformation
and error perturbation. It explores the basic structure of observed sample data by
studying the internal relations between the original explicit variables, and uses a
small number of factors to represent their data structures, through which they can
represent the core information of the original data.
Yu et al. [106] removed the redundant information in the arc data based on the PCA
method and discussed the relationship between the feature information and defects
caused by the wire feeding problem, as shown in Fig. 1.14a. Using BP-AA, the three
situations were automatically classified and identified, as shown in Fig. 1.14b. Zhang
et al. [107] extracted six characteristics from H I 656.28 nm and Ar I 641.69 nm lines,
which have different sensitivity to defects. The feature extraction was performed
based on Fisher discriminant. The results showed that the peak area ratio was the
best signal for monitoring the welding process. Colombo et al. [108] extracted the
principal component with a sum of the three variances as high as 75% of the original
spectral data, and discussed the relationship between the principal component and the
gap and weld position. The literature [109, 110] used the SFFS algorithm to classify
and extract the arc spectrum in TIG welding, including the linear and continuum
of the characteristic elements. These spectral data extracted by the SFFS algorithm
can better distinguish between good welding and defective welding. The ratio of the
line spectrum to its associated continuum can avoid the uncertainty caused by the
randomness of the element line selection.
Linear dimension reduction methods have the advantages of fast operation speed
and good generalization ability. However, linearity is often only an ideal situation.
Data sample sets tend to exhibit nonlinear structures, even highly nonlinear. At this
time, linear dimensionality reduction methods cannot reveal the nature implied in
the data samples.
Another random document with
no related content on Scribd:
cheerfully took his turn at the pump, and so saved the captain and
me any anxiety on that account.
It had been no pleasure cruise after we passed Brisbane, and
became worse every day. There was not a dry place on board,
unless it was our throats. Everybody was constantly drenched with
the sea, and no one had a good square meal during the last four
days; but there was no discontent, everything was taken in good
part, and many a tough yarn was told while they were lashed to the
rail to keep themselves from being washed overboard.
After two days sheer battling for our lives, the wind died down, and
a steady southerly wind sprang up. This soon brightened our
prospects, and added considerably to our comfort. How thankful we
were for the peace and quiet after the rough and tumble experience
we had just passed through! The sea became as smooth as a mill
pond with just a steady south wind blowing, that drove us about five
knots an hour through the water. All our effects were brought on deck
and dried, and our sails, which had been considerably damaged,
were repaired, and on the fourteenth day we arrived at Cookstown.
Our passengers were soon landed, and Captain Brown took the little
vessel well into the river and moored her there until he decided what
he was going to do himself. I landed the following day, and soon
found that the Palmer was as far off as ever. The rainy season had
set in, and the roads were impassable. Whole districts between
Cookstown and the Palmer were under water, the rivers were
swollen and in flood, and no stores of any sort could be bought on
the road.
To describe Cookstown as I first saw it would be impossible. It
resembled nothing so much as an old English country fair, leaving
out the monkeys and merry-go-rounds. Tents were stuck up at all
points. Miserable huts, zinc sheds, and any blessed thing that would
shelter from the sun’s fierce heat and rain, were used as habitations.
There were thousands of people living in the tents and sheds, and
the place literally swarmed with men of all nationalities. Large plots
had been pegged out in the main street, on these were erected
either corrugated iron sheds, or large tents, and here all sorts of
merchandise was sold, cheap enough to suit all purses, but the wet
season was on, and there was no way of getting to Palmer. Parties
of men left every day in the rain and slush to try and reach what
seemed such a land of promise, but many returned saying that it was
no use trying, as the rivers could not be crossed. Hundreds of these
men lived out in the scrub with just a couple of blankets thrown over
some twigs for shelter, no fire being needed except for cooking. All
the scum of Melbourne, Sydney and Brisbane were gathered
together here, thieves, pickpockets, cardsharpers and loafers of
every description. This class had not come to dig for gold with pick
and shovel from mother earth’s bosom, but to dig it out of honest
men’s pockets by robbery and murder, and the robbing of tents in
their owners’ absence was becoming a daily occurrence, for
gathered there were the good, bad, very bad and indifferent.
One day a party of three men returned after having got as far as
the Normanby River. They had been caught between two streams,
and could neither get backward or forward. The patch on which they
were imprisoned was only a few feet above water, and for some time
they were not sure if they would not be swept off and drowned, as
the island was only about one mile long and a quarter of a mile wide.
Whilst they were searching for means to get over to Normanby
they made a gruesome discovery, one by no means uncommon.
There at their feet lying together were five dead bodies. They had
been starved to death, and under the head of each man was a small
leather bag of gold, averaging in weight about six pounds each.
What a terrible irony of fate—shut in between the waters and starved
to death, with over five thousand pounds between them! The bodies
were all shrunken and black, so burying them where they lay, the
party took the gold and divided it. A couple of days afterwards they
were able to swim their horses over the stream and return to
Cookstown.
There were several instances told about this time of miners who
had reached the diggings before the wet season had set in, gathered
a stock of gold, then finding their stores giving out, were forced to
pack up and retrace their steps for a fresh supply. Many, on that
terrible return journey, were struck down by the sun’s intense heat,
and after using their last small stock of food, died a miserable lonely
death by starvation, their treasures of gold powerless to buy them an
ounce of food.
It was quite a common occurrence for miners travelling up from
Cooktown with plenty of stores and provisions, but no cash, to arrive
on the banks of a swollen river, over which there was no means of
crossing, and to see on the other side of the river a party of men on
their way down to the coast with bags of gold, but with hungry, empty
stomachs. There they were, looking across at each other, one
holding up a bag of flour, and the other shaking his gold purse, each
powerless to help the other. Such was the lot of many of the diggers
at that time, but all the horrors, the suffering and death that took
place in that mad rush for gold, will never be known. ’Tis better so, I
saw men return from the gold fields, with thousands of pounds worth
of gold in their possession, but with frames so emaciated and ruined
with what they had gone through on their return journey, that their
very existence was a burden to them, their horses, dogs, and even
their boots had been eaten to keep them alive. It is a fact that they
have boiled their blucher boots for a whole day, and then added any
weeds they could find to make a broth of, so tenacious of life were
they.
There were hundreds of men idle in Cooktown. They had no
means of buying an outfit, even if the road to Palmer had been
passable, and many of them had no desire to go any further. These
could easily be distinguished from those who really wanted work
during the waiting time, so many there were that anyone who wanted
a man might easily get him for a whole day’s work for a good square
meal. Men would walk about among the tents and whenever they
saw food there they would beg. Many were getting a living by their
wits and knavery, and it was not safe to be about alone after dark,
unless you were well armed and prepared for these light-fingered
gentry. And yet the leading articles in the newspapers at that time
were painting in glowing terms the bustle and activity going on in the
rising city of Cooktown, declaring that any man who could use a
hammer or tools of any description could earn a pound a day.
Feeling a bit disheartened at the grim realities that I had
witnessed, and after knocking about Cooktown for a week, I called
on Captain Brown, and asked him if he was going to take the
“Woolara” back to Newcastle.
“No,” he replied, “I have sold her, and made a jolly good thing out
of her, too, and I’m going to have a try to get to the Palmer. What are
you going to do?”
“Well, I am undecided at present, there are so many returning
disheartened, and broken down in health, and they give such bad
accounts of the road to be travelled over before you reach the
Palmer, that I don’t care about tackling it alone.”
“Well, look here,” said the captain, “I have done very well by this
venture so far, and I don’t care about returning without having a try
for the diggings, even if I have to return. What do you say to us
joining forces, and trying our luck together. I will buy three horses
from the next squad that returns, and use one for a pack horse.”
I agreed to his plan, and the following day about a dozen
horsemen rode into Cooktown. They had been a month on the road,
several times they had narrowly escaped drowning, while trying to
cross the Normanby river. They had lost nearly the whole of their
provisions, and one of their mates had been seized by an alligator
before their eyes, while they were powerless to help him. Then they
had been obliged to kill two of their horses for food. They willingly
sold us three horses at fifteen pounds each, but strongly advised us
not to try the road for at least two months, or to wait for the end of
the rainy season. But the thought of the gold beyond made us eager
to take our chance. Had we gone back to Newcastle without trying,
our friends would have chaffed us unmercifully.
The next day we began our preparations. We bought a tent, two
small picks, two small spades and one gun. Captain Brown had a
gun and revolver. I had a revolver, and the gun that was bought was
for me, and a good supply of ammunition. As we were going where
money was of no value and food invaluable, and everything
depended on our being able to carry sufficient provisions, we got a
good supply of the best. We had cocoa, extract of beef, preserved
meat, tea and sugar, two hundred pounds of flour—this was divided,
one hundred pounds to the pack horse, and fifty pounds to each of
our horses—two large billy cans, a couple of drinking pots, two
knives, two basins, a tinder box and burning glass. When we were all
packed and ready to start, we looked like a couple of mountebanks
off to a village fair.
It was a fine morning when we started, but before we had got ten
miles from Cooktown our horses were sinking in the mire. Road
there was none, it was just a track or belt of morass, into which one
sank at times knee deep, and as night came on it rained in torrents,
so we picked out a dry piece of ground, and pitched our tent for the
night. We then hobbled the horses with about ten fathoms of line to
keep them from straying.
We slept well that night, for we were dead tired, and had we been
lying on a feather bed in a good hotel instead of on a piece of ground
that might soon be under water, we should have slept no better. As it
had ceased raining when we awoke we started on our way again
after we had breakfasted, and got along very well until noon. Coming
to a place where there was good grass for the horses we decided not
to go any farther that day, but to let the horses have the benefit of a
good feed.
The following morning we started early, and at noon met a party of
diggers returning from the Palmer. They had been fortunate enough
to get a fair amount of gold they said, but what a terrible condition
they were in, thin and emaciated as skeletons, with barely a rag to
cover them. Three of their party had been lost crossing the Laura
river, and one had died of sunstroke on the road.
“What is it like further ahead, mates?” asked Brown.
“Well, it is only just passable to the Normanby river from here. I
don’t think you will be able to cross it with your packs. We had to
swim it, holding on to the horse’s tails, and then we lost some of our
little stock of food, it was a narrow squeak for us all, horses and men,
but we are here, thank God, safe so far.”
Brown gave them a small tin of beef essence, and a few ship’s
biscuits that he had brought with him. The gratitude of the poor
hungry fellows was pitiful to see, then they offered us some of their
hardly won gold for it, which we promptly refused.
“No, no, mates,” said Brown. “You chaps have earned and suffered
enough for that. Keep it, and take care of it, and may you live to
enjoy it.”
We camped all together that night, after sitting yarning for some
hours, and when we had all eaten a very hearty breakfast we
separated, each party going on its way, like ships that pass in the
night, never to meet again.
Our track that day was very bad, just slush and mire, the horses at
every step sinking up to their knees. We were ready and expected to
meet with hardship on the road, but to realize the suffering to man
and horse dragging themselves along that quagmire is better felt
than described. Every moment we were afraid of them breaking
down, and when about two p.m. we got on a stretch of solid ground,
we pitched our tent, and gave them a good rest. So far we had not
seen a living bird or animal since leaving Cooktown. Had we been
depending on our guns supplying our larder with food we should
have had to go short, fortunately for us we were not.
The next day it was terribly hot, and, to add to our discomfort, we
had several heavy showers, which soon wet us through and through.
When these stopped and the sun came out again our clothes
steamed on us, just as though we were near a fire; this and the
steam arising from the ground made us feel faint and feverish. We
were also pestered with a common little house fly that swarmed
around us and was a perfect nuisance. At sunset we felt we could go
no farther, so pitched our tent on a patch of stony ground close to a
creek, where there was good grass, so we hobbled the horses and
let them graze.
We turned in early, for we were dead tired, and the mosquitoes
were buzzing round in myriads, with their incessant cry of “cousin,
cousin,” when about midnight we were roused by a tremendous row
near us, a peculiar indescribable noise was coming down from the
creek, which we could not account for. We both sprang up and
seized our guns, but the night was pitch dark. What it might be we
did not know, we did not go out, but remained in our tent on the
defensive. Never had either of us heard anything like it; it was as one
often hears, “sufficient to raise the dead.” We began to wonder if we
were surrounded by a mob of the blacks, who were lurking around
us, or was it the spirits of those who had perished on this lone track,
and who were trying to make us return to civilization, but whatever it
was, it was awful and above all the noise could be heard quite
distinctly—a piercing yell of pain, such as no human being or animal
we knew could utter. Thinking to frighten the blacks, if it were indeed
they, we shouted out to each other in different tones and names, to
give them the impression that we were neither alone or unarmed.
When the welcome daylight came we Went in search of the
horses. We could only find two, but on the bank of the creek, not far
from the tent, was the forepart of our third horse. It was bitten off
right under the forelegs, all the rest was gone. There on the ground
and in the soft mud were the signs of a struggle, and the marks of
some big body having been dragged towards the water. Close to the
water were the tracks of a huge alligator, and where it had come out
of and entered the creek, a deep furrow had been turned up by its
tail. This explained the noise in the night, it was the struggle and
death agony of the poor beast, it must have been drinking at the
creek and been seized by the alligator. This was a very serious loss
to us, and made us feel quite disheartened.
We remained where we were until noon. Then crossed the creek
and went on our way—our horses more heavily weighted than before
owing to the loss of the packhorse—and at sundown we pitched our
tent. Our fire was barely lighted to boil the billy for tea, when three
men crawled up to the tent. We were so surprised, that for the
moment we stood still looking at them, for they looked like
scarecrows with their clothes hanging in rags upon them.
“For God’s sake mates, give us something to eat, we are starving,
we have lost everything crossing the Normanby.”
“Aye, aye, lads,” said Brown. “Come up to the fire, and you shall
share our meal. Have you come from the Palmer?”
“No, we could not get there. It is six weeks since we left Cooktown,
and we are trying to get back. Our provisions gave out, and we could
neither go forward or get back, owing to the district being flooded
and impassable. Three days ago the strength of the river eased
down a bit, and we managed to cross by strapping our bits of
clothes, and the little food we had on the horses’ backs, then we got
on their backs and forced them into the water, but the current was so
great that they were borne down the stream, so we slipped off, and
getting hold of the horses’ tails with one hand, we swam with the
other. We managed to cross, but it was a desperate undertaking, and
we were so done up that we were too weak to tie up the horses. We
just lay where we landed and went to sleep. We never saw the
horses again, and have not the slightest idea what has become of
them. And now mates, we are stranded here, without a bite of food,
and unless you can help us here we must die; we can go no farther.
What is it to be?”
“Well, strangers,” said Brown, “my mate here and I were bound for
the Palmer. We have had a tough job of it so far, and we have had
quite enough of it. Hal a good meal, and rest yourselves well, and
we’ll all go back together.”
The poor fellows could hardly find words to thank him. They ate a
hearty meal, and washed it down with a good pot of tea, and very
soon after were in a sound sleep.
Brown and I sat talking far into the night. To tell the truth I was not
sorry he had decided to return, for with one thing and another, I had
begun to ask myself whether the game was worth the candle, and
seemed all at once to have sickened of the roaming about, and felt
that the ups and downs of sea life were luxury in comparison to
hunting for goldfields.
The following day we divided the stores between the two horses,
and prepared to tramp back to Cooktown.
CHAPTER XXV
We Return to Cooktown
The first day of our return journey we travelled as far as the creek
where we had lost our horse the day before. The poor fellows we
had rescued were completely done up, so Captain Brown
determined to go along slowly, and so give them a chance to pick up
their strength. Their names, they told us, were James Whitefield,
Henry Bagly and Thomas Pain. Whitefield, it seems, had been on
almost every goldfield in the colonies, and had three times been in
possession of twenty thousand pounds worth of gold. According to
his own account, which I afterwards verified, the man had not a
friend in the world, or a relative living. He was utterly indifferent to
worldly possessions, and after returning from the Victorian goldfields
had spent, or squandered, twelve thousand pounds in Melbourne in
three weeks. A woman in Burk Street took his fancy, and he bought
and furnished a house for her that cost him five thousand pounds,
then, after living with her there for ten days, he grew restless and
cleared out to the Charter Tower goldfields. He could neither read
nor write distinctly, because, as he said, he had no use for either.
The other two men were runaway sailors, who had been working
ashore for twelve months at Brisbane before starting for the Palmer.
The following morning we swam the creek after firing our guns and
shouting to scare any alligators that might be about. The creek was
about two hundred feet across, and for about sixty feet from the
south shore the depth was only about four feet, then the bed
suddenly dropped and the current rushed very strongly until the north
shore was reached, and there the landing was very bad as the scrub
came right down to the water. The way we crossed was as follows: A
small line was made fast to the after part of the saddles and
stretched along each horse’s back and a half hitch round its tail. The
horses were then driven into the water, and at once began to swim
across. Captain Brown and Whitefield hung on to the rope of one
horse, and the other two men and I took the other. Before we started
Brown told me to keep next the horse and watch it closely, and to
keep my sheath-knife handy for fear the current might sweep it away.
Brown’s horse led, and we stood to watch it land. When about half
way across Whitefield let go the rope, and with a swift stroke brought
himself alongside the horse on the lower side, then he kept one hand
on the saddle and used the other to propel himself. This eased the
horse somewhat, and he got over fairly easily.
After they had safely landed, Brown called out to me to ease all
weight off the horse. We started, and I swam alongside the horse like
Whitefield had done. The other men held on to the rope with one
hand and swam with the other, and we got along first class until
about fifty feet from the other side, when I felt my feet touch
something, and my heart came into my mouth. The next minute the
horse seemed to be jerked backward, and terrified he began to
plunge, snorting and neighing. Then I heard Whitefield sing out:
“Cut the rope! Cut the rope!”
I drew my knife, and while holding on to the saddle with my left
hand, reached over and cut the rope near the saddle, in my haste
cutting a gash in the horse’s back. At the touch of the knife, and with
the strain from behind relieved, the horse plunged ahead, and in a
minute we landed. I looked round for the other men, but they had
gone under.
“Whatever was the matter, Brown?” I asked.
“Well I don’t know,” he replied. “We saw the fellows go under, and
saw the horse floundering, and Whitefield called out cut the rope,
and if it had not been cut at that moment, the horse would have gone
under, and you, too, I expect.”
“But what do you think took them under?” I persisted. “We were
going along all right at first. Do you think it was an alligator,
Whitefield?”
“Oh, no,” he replied, “if it had been he would have gone for the
horse first. I think there must be a dead tree, or a snag down there,
and they must have struck it and been drawn down in the eddy. They
are dead enough by this time, anyhow.”
“But good heavens, mates, it’s awful,” said Brown, “to think we all
had breakfast together, and now two of us are dead. Were they
friends of yours, Whitefield, you seem to take it pretty coolly if they
were?”
“No,” he replied, “I didn’t know them. We met on the road over the
Normanby river, and beyond their names, I know nothing about
them, except that they had been sailors. They were jolly good mates
—I know that much, anyhow. As to my taking it coolly, well, mates,
my fussing about it would not bring them back, it may be our turn
next, we are not in Cooktown yet. I expect they suffered less in that
last lap of their race in life, than in any other part, and by this time
they’ll have learnt the grand secret.”
“Well, look here,” said Brown, “spread the tent and make some
tea, and I’ll go along the bank and see if there is any sign of their
bodies washing up.”
Whitefield and I soon had the tent spread, and the tea made. The
horses were hobbled, their loads taken off, and they were turned out
to graze. There was not much grass in the place, but a small shrub
that grew in abundance they ate freely of and seemed to enjoy.
Strange to say, although all our stores had been in the water there
was not much damaged. The two small bags of flour I thought would
have been ruined, but they were not. The water had only formed the
flour into a cake on the outside, but the inside was all right.
When the billy was set on to boil I strolled along the bank to meet
Brown, whom I saw was coming back. When I was close to him I
suddenly espied, about twenty yards from the edge of the river, a
bundle tied up with a stick through it, as though it had been carried
over a man’s shoulder. I walked towards it, and Brown, seeing it too,
walked over towards it. He gave it a kick with his foot, and the next
minute was on his knees untying it.
“Some Johnny’s swag,” he said, as he opened the bundle.
The covering was a piece of tent duck, inside it were a pair of
socks, and a wool shirt, both filthy dirty, rolled up inside the shirt was
a piece of canvas, which had apparently been the sleeve of a canvas
jacket. Both ends were tied with a strong grass like flax, and inside
was about eleven pounds of fine gold, that looked just like birdseed.
“Halves, Brown,” I said.
“Oh, no, not halves, mate,” he replied.
I drew my revolver and covered him.
“Why not?” I asked, my temper rising to a white heat at the sight of
the gold.
Brown smiled:
“Put back that revolver,” he said, “you mad-brained young beggar.
What about the other chap shan’t we give him a bit, he needs it just
as much as we do.”
“Oh, yes,” I replied, feeling a bit ashamed, “I agree to that.”
So we shared it out, five pounds each for Brown and me and one
pound for Whitefield. He thanked us, and said he had no claim to any
share, as he was only a stranger, and we were old mates. Who he
was, or what had become of the owner of the swag will never be
known. It was evident he had come from the diggings and had safely
crossed the river. Perhaps he was another of those without food,
who became exhausted, went mad, under the broiling sun, and had
wandered off, or he may only have lain down to sleep and during the
night had been seized by one of the alligators, which were very
numerous in the Normanby at the early stages of the gold rush. The
truth will never be known.
After we had eaten a good feed of damper and tea, we caught the
horses, loaded them up and continued our journey. It was terribly
rough the first few miles. The track was just a spongy quagmire, into
which we and the horses sank knee deep and could hardly pull our
feet out again so great was the suction. And every now and then the
poor beasts would look pitifully at us, as they bravely tried to get
along. However, just at sunset, we found a pitch of dry ground and
rested there for the night.
The following day we got along a little better, but our stores were
getting very low, and the sky began to look very threatening, and the
next morning we were up and off at daylight, but we had only gone a
few miles when the storm burst over us, and the rain came down in
sheets. We spread the tent, but it leaked like a sieve, while the
thunder and lightning was awful. We were soon wet through to the
skin, and everything else we had was in the same condition. We
were afraid to let the horses stray for fear of losing them altogether.
All night the rain came down in torrents, and when daylight came the
whole face of the country was a sheet of water;
“Pack up, lads,” said Whitefield, “we must get away from here
before the floods come down, and then we shall get bogged and that
will be the end of us. I’ve been through that once, and had to shoot
as good a horse as a man need wish for, he was slowly sinking in the
bog. I could not get him out, and the pitiful look in his eyes as he
sank deeper and deeper was more than I could stand, so I just
ended his misery by putting a bullet in his brain, so let’s get on while
we can.”
We managed to make a pot of tea, for we had very little else by
now, and started off again, but what a journey! Every hole and hollow
was full of water, and first one animal and then the other would
stumble into them, both man and beast, I think, had the roughest
time of their life that day, for at the best of it we were nearly up to our
knees, and sometimes a good bit above them. At sundown Brown
wanted to camp, but Whitefield urged us to push ahead until we
reached more solid ground. After a few miles of this quagmire, which
seemed to get worse, and when it was near midnight, we came up to
some bushes or scrub; we found the ground was a little higher and,
though still wet and sloppy, we felt we could go no further, so here
we camped for a few hours’ rest.
At daylight we found, to our surprise, that we were near a camp of
men making for the Palmer. There were quite twenty of them, and
they seemed to be well supplied with stores and horses, in fact, they
looked the most likely and best equipped party that I ever saw on the
way to the goldfields. They had two light-built carts, made specially
for that purpose. These carts were four-wheeled, of light, tough
material, the seams were well puttied and painted and over all the
outside was a cover of strong painted canvas, with two cane wood
runners underneath. When crossing the rivers, the horses were
taken out of the shafts, and the harness was put into the cart with the
stores, the horses would then swim over to the other side, taking the
end of a long line with them. On landing, the other end of the line
was made fast to the cart, and the horses who were on the river
bank easily pulled it across, thus keeping the stores dry. It was a
capital idea and had been well thought out, and would answer its
purpose well. They also had with them a powerful dog of the
Newfoundland breed that had been trained to swim across the
creeks and rivers with a light rope. The party were prepared for any
emergency that might offer itself, and their outfit must have cost a
good sum of money. When Whitefield saw them he offered to go and
assist them for his food, until they arrived at the diggings. Such was
the fascination that the goldfields held for this man. The party readily
accepted the offer of his services, and he joined them at once.
After watching the party start off, we also continued our journey,
and arrived in Cooktown twenty-four hours later. Many were the
enquiries made of us as to the state of the roads and prospects of
reaching the Palmer. There were still hundreds of men waiting in
idleness at Cooktown for the rainy season to pass. The place
seemed worse than when we left it, for wherever you turned there
were the loafers hanging round in scores. Brown was able to dispose
of his horses and tent for forty pounds, clearing ten pounds by the
deal, for horses were scarce and dear, and he might have got more if
he had stood out for it. We sold our gold to the bank and received
from them cash and notes to the value of two hundred and ten
pounds each. Then we put up at a second class restaurant and that
day I posted a money order, value one hundred and fifty pounds, to a
friend in Sydney, to bank for me until I came back, and in the event
of my death it was to be sent to my mother in Liverpool, and Captain
Brown posted a draft to his wife at Newcastle, New South Wales. It
was not safe by any means to have it known about the town that you
had any money on you, especially after dark, as there were plenty of
men in Cooktown at that time who would have cut your throat for
half-a-crown, and think themselves well off to get that much.
CHAPTER XXVI
A Trip to the Cannibal Islands and Captain
Brown’s Story
We stayed together in Cooktown for a couple of weeks, and then

Captain Brown was offered the command of a small vessel trading
between Cooktown, Townsville, and the Solomon Islands, sometimes
calling at Port Moresby, New Guinea. He at once offered me the
berth of mate in her, and I gladly accepted, as it was quite a new part
of the world to me, and just what I wanted. The “Pelew” was a smart
little schooner of a hundred and fifty tons, could sail like a water
witch, and was a right staunch little craft. We shipped three deck
hands, one a young Danish seaman, who had cleared out from an
English ship at Brisbane, and two Kanakas. The Dane was a smart,
active young fellow, his only drawback being that he could not speak
a word of English, but it was evident he would soon learn. The
Kanakas were two splendid types of the Solomon Islanders, they
were sharp, intelligent men and could speak “pigeon” English. In
their younger days they had been slaves on a Queensland sugar
plantation, but for the last two years they had been on one of the
missionary schooners cruising among the Pacific Islands. They took
life very merrily, and were always laughing, no matter what had to be
done—they got some fun out of it. Work was no trouble to them, and
when there was no work going on they would wrestle with each
other, tumbling each other about until the perspiration rolled off them,
but they never lost their tempers over it, but would finish up with a
hearty laugh. Sometimes they would get the young Danish sailor to
wrestle with them, but they could do just what they liked with him, he
was muscular and strong, but they were slippery as eels, and twisted
and twirled as though there was not a bone in their bodies, and
always slipped out of his fingers before he could get a grip on them.
It was great fun to Captain Brown and me to see the Kanakas,
Tombaa and Panape, trying to teach Neilson, the Dane, to speak
English, and Neilson trying to teach them Danish. That seemed the
only thing they could not get any fun out of. At last Panape gave it
up, and would not have it at any price.
“That no tam good,” said he, shaking his head. “Good fellow white
man—speak Englis’—no that allee samee you. You no takee allee
same good fellow captain—good fellow, mate?”
“No,” said Neilson in English.
“You no tam good, then,” said the Kanaka. “All good fellow speak
Englis’. Me good fellow—me speak Englis’. Tombaa, he good fellow
man, too—he speak allee samee missiony man, he teach us to say
prayer to ‘Big Fellow Master’ (God), prayer belong sleep, prayer
belong get up. Tombaa you speak white fellow commandments.”
I drew nearer to them, anxious to hear a Kanaka’s version of the
ten commandments. Tombaa stood up, and throwing his chest out
like a proud turkey cock, he delivered the following version:—
The Ten Commandments in Kanaka.
I. Man take one fellow God, no more.
II. Man like him God first time, everything else behind.
III. Man no swear.
IV. Man keep Sunday good fellow day, belong big fellow
Master.
V. Man be good fellow longa father, mother belonga him.
VI. Man no kill.
VII. Man no take him mary belonga ’nother fellow man.
VIII. Man no steal.
IX. Man no tell him lie ’bout ’nother fellow man.
X. ’Supose man see good fellow something belonga
’nother fellow man, he no want him all the time.
I was much amused at their interpretation, what it lacked in length
was made up by the clear definition of the meaning of the ten
commandments, and these two lived up to it.
We left Cooktown with a general assortment of cargo for
Townsville, and a few deck passengers. The wind being fair and the
weather fine, we made the passage in fifty-four hours, anchoring
inside Magnetic Island. Our cargo and passengers were soon
landed, and the schooner loaded for Port Moresby, New Guinea. The
cargo consisted of cloth, prints, calicoes, ribbons of all sorts and
colours, tobacco (horrid stuff), spirits, axes and various joinery tools,
etc., and some agricultural implements. We also had four
passengers—German officials—going to the German settlement,
north-east New Guinea.
We left the port at sunrise. The weather was fine, one of those
lovely tropical days when the sky blends its prismatic hues and the
easterly breeze, as it whistles through the shrouds, brings new life
and energy into one’s veins. The sea all around was covered with
silver-crested waves and as the little “Pelew” cut her way through the
sparkling waters she sent them like showers of jewels along her
painted sides. What a joy it was to me to be once more on the
ocean, to feel once more the motion of the vessel beneath my feet,
and to quaff the salt breeze that was like the wine of life. We had a
delightful passage, but owing to the numerous reefs and shoals we
were kept constantly on the lookout. These seas require the most
careful navigation, and I was surprised to find that Captain Brown
seemed quite at his ease among the reefs, although, when I
mentioned this and asked him about his life in these regions, I could
never get any very definite answers from him. However by putting
two and two together, from his chance remarks, I came to the
conclusion that he had been what is known as a “blackbird catcher,”
an “island scourger,” a “dealer in living ebony,” or a “sandlewood
thief.”
We made the passage to Port Moresby in five days. As soon as
we anchored in the bay three native crafts came off for our cargo, the
agent who was in the first boat seemed half a savage himself, and
had a most repulsive face. Captain Brown gave orders that no one
was to leave the ship on any pretext whatever, except the German
passengers, and they did not seem to like the job either, but that was
what they had come out for. No natives were allowed to come on
board. Their appearance was not very inviting, they were quite
naked, with the exception of a strip of pounded bark or cocoanut
fibre round their waist, their woolly heads were decked with shells
and tufts of grass, while round their necks each had a necklace of
shark’s teeth. Though fine, well-built, powerful looking fellows, their
features were not what we should call handsome, as their foreheads
are low and retreating, the face broad, the cheek-bones prominent,
the nose flat and the lips thick. We heard that there was an English
missionary living amongst them and doing a good work.
After delivering what goods we had for the store-keeper, we
received orders to proceed to Gaurdalcana in the Solomon Islands,
and deliver the balance to the store-keeper there.
Captain Brown then told me that the natives of the Solomon
Islands were cannibals, “so you had better be careful while we are
amongst these islands, and,” he continued, with a sly twinkle in his
eyes “you have to be very cautious in dealing with them, for they are
very partial to roast sailor. I had a terrible experience on one of the
islands some years ago. I was in a smart little brig, cruising among
the Islands. We were out on a blackbird (native) catching expedition.
We sailed into the bay at the south-east point of San-Christobal. The
brig ‘Carl’ of blackbird notoriety, had been there a few times, and
after getting a number of the natives on board to trade as they
thought, they had been invited into the saloon, and their eyes were
dazzled by the beads and toys and other things spread on the table.
Unsuspecting of any treachery they stayed until the gentle rolling of
the vessel caused them to ask with some surprise what it meant, by
this time the ship was well under way, and fast leaving San-
Christobal behind them. They tried to rush on deck, but found
themselves covered with the rifles of some of the ship’s crew, they
were soon overpowered and made prisoners and put into the hold
with others who had been lured to the vessel by the same device—
all to be sold as slaves to the North Queensland planters—but we
were not aware of this at the time.
“Well, as we drew up towards the head of the bay we suddenly
grounded on a reef, and while we were rushing about backing and
filling the sails, the natives swam off in hundreds and boarded the
vessel on all sides. We let go the ropes and seized whatever we
could lay our hands on to defend ourselves, but in a minute three of
our men were beaten to death with clubs. The captain was aft by the
wheel, and as soon as I saw the natives climbing over the rail I drew
my knife and sprang aft near him, and together we fought like
demons. But the copper-coloured fiends thronged round us, and one
big fellow at last got a blow in with a club that laid the captain
senseless on the deck. But his triumph was short, and mine too, for I
ripped him open with my knife, and the next minute was knocked
senseless on the deck myself. When I came to, I was on the floor of
a hut on shore, trussed like a fowl, with my arms and legs bent
behind me and lashed together. I struggled and twisted to get my
hands free, but it was no use, I could not do it. I raved and shouted
for some one to come and put me out of my misery. At last, as if in
answer to my cry, one of the women came and looked in, and seeing
me struggling, she picked up a club, and smashed me on the head
with it, and again I became senseless. The next thing I remembered
was being rolled over and over and my flesh being pinched by two or
three natives. After jabbering among themselves for a few minutes
they left me, and directly afterwards I heard the captain’s voice
shouting not far off, and a lot of jabber among the natives. I could not
see what was going on, but I knew that they were taking the poor
fellow to kill and roast him. I tore at my bonds, until the lashings cut
into the flesh. Suddenly a horrible yell burst on my ears, and I knew it
was the captain’s death cry. I shook like a leaf, and the perspiration
rolled off me like raindrops. I was on the rack with torture, knowing
full well what was before me, and that at any minute my turn might
come. I swooned away with horror at the thought, to be brought to
later by a burning stick being thrust into my face. I saw four of the
devils were in the hut, and a whole crowd outside. They put a small
spar through my arms, and two of them lifted me up between them,
like a Chinaman carrying a load. As they carried me along towards a
large fire in the middle of a clearing, near a large hut, like a meeting
house, my stomach and face were scraping the ground, and, oh!
God, what a terrible sight met my eyes. There just in front of me was
the roasting body of the poor skipper. He had been a bad devil in his
time and many an islander had suffered at his hands, but they had
got their revenge on him for it.
“The head man or chief now spoke to a big powerful savage, and
the latter approaching me with a large knife, was about to plunge it
into me to rip me open, when the head man, who was jumping about
before me, suddenly fell forward on his face and lay still. The others
looked on and shouted. Then some of the elder ones, seeing there
was something wrong, walked up to the prostrate chief, and touched
him. Finding he did not move, they turned him over, but he was
dead. I thought they would fall on me at once when they realized
this, but they only set up a great wail and beat their breasts with their
hands. Then two of the old men spoke up, and all was quiet. After
they had done speaking several of the men came to me, and I
thought my last moment had come, but, to my surprise, they gently
untied my hands and feet. For a few minutes I was unable to stand,
but as soon as I could, one of the old men picked up the spear and
club of the dead chief, placed them in my hands, and pointed to the
hills. I was not long in taking advantage of my freedom, and made
tracks at once. I could hardly believe that I was free, and expected
every minute to hear them coming after me.
“Why I had been spared was a mystery to me then, but I
afterwards learned that they released me through some superstitious
fear, and a belief that the spirit of their dead chief had entered into
me, had I been so minded they would have made me chief of the
tribe; this they tried to make me understand when the old man
placed in my hands the spear and club belonging to the dead chief. It
would have made no difference to me had I known, all I wanted was
to put as many miles as possible between the cursed place and
myself.
“I remained in hiding for a couple of days up among the hills, and,
strange to say, I never saw a single native come near to the place
where I was. Another thing I noticed in my wanderings was the
absence of children. I don’t remember seeing a single youngster. As
a rule there are plenty of them knocking about on most of these
islands, so I came to the conclusion that this was an island where it
is the custom for nearly all the children of both sexes to be killed by
their parents, perhaps eaten, too. I lived on bananas, cocoanuts, and

PDF Key Technologies of Intelligentized Welding Manufacturing The Spectral Diagnosis Technology For Pulsed Gas Tungsten Arc Welding of Aluminum Alloys Yiming Huang Ebook Full Chapter

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDF Key Technologies of Intelligentized Welding Manufacturing The Spectral Diagnosis Technology For Pulsed Gas Tungsten Arc Welding of Aluminum Alloys Yiming Huang Ebook Full Chapter

Uploaded by

Copyright:

Available Formats

Key Technologies of Intelligentized

Welding Manufacturing The Spectral

Key Technologies of Intelligentized Welding

Friction Stir Welding : Dissimilar Aluminium Alloys

Advances in Welding Technologies for Process

Welding : principles and practices Bohnart

Key Technologies of Magnetically Coupled Resonant

Machining of Light Alloys: Aluminum, Titanium, and

Automation in the Welding Industry: Incorporating

Encyclopedia of Aluminum and Its Alloys, Two-Volume Set

ISBN 978-981-13-7548-4 ISBN 978-981-13-7549-1 (eBook)

Library of Congress Control Number: 2019935997

© Springer Nature Singapore Pte Ltd. 2020

As a cross-disciplinary study of arc physics, heat transfer, metallurgy, and

effectively express the information of welding parameters, joint quality, product

Tianjin, China Yiming Huang

3 Basic Characteristics of Arc Spectrum in P-GTAW Process . . . . . . . 43

5 Dimension Reduction of Arc Spectrum for Porosity Detection

Abstract Aluminum alloys have promising application in automobile and aircraft

1.1 Welding Technology

1.1.1 The Principle of the Welding Process

Welding is a process in which the weldment is atomically bonded by heating, or

© Springer Nature Singapore Pte Ltd. 2020 1

1.1.2 Gas Tungsten Arc Welding and Improvement Process

1.2 Welding Defects of Aluminum Alloys

1.2.1 Application Range of Aluminum Alloys

As a low-density, high-specific strength structural material, aluminum alloy has good

1.2.2 Study on Weldability of Aluminum Alloys

1.3 Intelligent Welding Manufacturing Technology

The intelligent welding manufacturing technology is proposed in the environment

1.3.1 Sensing Technology Based on Electrical Signals

1.3.2 Sensing Technology Based on Sound Signals

Sound sensing information is a very important source of information in the weld-

In general, the current research on welding sound is mainly concentrated in plasma

1.3.3 Sensing Technology Based on Visual Images

ing information for real-time penetration monitoring, a three-light visual sensing

1.3.4 Sensing Technology Based on Arc Spectrum

1.3.4.1 Plasma Diagnosis

As a low-temperature (thermal) plasma, the temperature of the welding arc is dis-

a deep understanding of the physic properties of the arc, non-contact spectroscopy

1.3.4.2 On-Line Detection Based on Spectral Signals

Fig. 1.6 a Defect weld of seam oxidation 2 b SNR(SOI-7-D) = 14.26425 c SNR(SOI-7-K) =

Fig. 1.11 Identification of a defect with different species [95]

Fig. 1.12 Detection of

of Al alloys was investigated. It was proposed that the H I 656.28 nm and Ar I

1.4 Challenges of Spectral Sensing Technology

1.4.1 Data Dimension Reduction

A Trip to the Cannibal Islands and Captain

We stayed together in Cooktown for a couple of weeks, and then

You might also like