Proceedings of the Third International

Scientific Conference “Intelligent

Information Technologies for Industry”
(IITI’18): Volume 1 Ajith Abraham
Visit to download the full and correct content document:
https://textbookfull.com/product/proceedings-of-the-third-international-scientific-confer
ence-intelligent-information-technologies-for-industry-iiti18-volume-1-ajith-abraham/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Proceedings of the Third International Scientific

Conference “Intelligent Information Technologies for
Industry” (IITI’18): Volume 2 Ajith Abraham

https://textbookfull.com/product/proceedings-of-the-third-
international-scientific-conference-intelligent-information-
technologies-for-industry-iiti18-volume-2-ajith-abraham/

Emerging Technologies in Data Mining and Information

Security Proceedings of IEMIS 2018 Volume 1 Ajith
Abraham

https://textbookfull.com/product/emerging-technologies-in-data-
mining-and-information-security-proceedings-of-
iemis-2018-volume-1-ajith-abraham/

Proceedings of the Fourth International Scientific

Conference Intelligent Information Technologies for
Industry IITI 19 Sergey Kovalev

https://textbookfull.com/product/proceedings-of-the-fourth-
international-scientific-conference-intelligent-information-
technologies-for-industry-iiti-19-sergey-kovalev/

Emerging Technologies in Data Mining and Information

Security Proceedings of IEMIS 2018 Volume 2 Ajith
Abraham

https://textbookfull.com/product/emerging-technologies-in-data-
mining-and-information-security-proceedings-of-
iemis-2018-volume-2-ajith-abraham/
The ShortTube 80 Telescope A User s Guide Neil T.
English

https://textbookfull.com/product/the-shorttube-80-telescope-a-
user-s-guide-neil-t-english/

Essentials of Hypertension: The 120/80 paradigm 1st

Edition Flávio Danni Fuchs (Auth.)

https://textbookfull.com/product/essentials-of-hypertension-
the-120-80-paradigm-1st-edition-flavio-danni-fuchs-auth/

Intermittent Fasting for Women Over 50 80 Recipes

Included 1st Edition Sharon Milson

https://textbookfull.com/product/intermittent-fasting-for-women-
over-50-80-recipes-included-1st-edition-sharon-milson/

Sew Your Own Wardrobe More Than 80 Techniques 1st

Edition Alison Smith

https://textbookfull.com/product/sew-your-own-wardrobe-more-
than-80-techniques-1st-edition-alison-smith/

Unpublished Fragments from the Period of Dawn (Winter

1879/80–Spring 1881) 1st Edition Friedrich Nietzsche

https://textbookfull.com/product/unpublished-fragments-from-the-
period-of-dawn-winter-1879-80-spring-1881-1st-edition-friedrich-
nietzsche/
Advances in Intelligent Systems and Computing 874
Ajith Abraham
Sergey Kovalev
Valery Tarassov
Vaclav Snasel
Andrey Sukhanov Editors

Proceedings of the
Third International
Scientific Conference
“Intelligent Information
Technologies for
Industry” (IITI’18)
Volume 1
Advances in Intelligent Systems and Computing

Volume 874

Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: kacprzyk@ibspan.waw.pl
The series “Advances in Intelligent Systems and Computing” contains publications on theory,
applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually all
disciplines such as engineering, natural sciences, computer and information science, ICT, economics,
business, e-commerce, environment, healthcare, life science are covered. The list of topics spans all the
areas of modern intelligent systems and computing such as: computational intelligence, soft computing
including neural networks, fuzzy systems, evolutionary computing and the fusion of these paradigms,
social intelligence, ambient intelligence, computational neuroscience, artificial life, virtual worlds and
society, cognitive science and systems, Perception and Vision, DNA and immune based systems,
self-organizing and adaptive systems, e-Learning and teaching, human-centered and human-centric
computing, recommender systems, intelligent control, robotics and mechatronics including
human-machine teaming, knowledge-based paradigms, learning paradigms, machine ethics, intelligent
data analysis, knowledge management, intelligent agents, intelligent decision making and support,
intelligent network security, trust management, interactive entertainment, Web intelligence and multimedia.
The publications within “Advances in Intelligent Systems and Computing” are primarily proceedings
of important conferences, symposia and congresses. They cover significant recent developments in the
field, both of a foundational and applicable character. An important characteristic feature of the series is
the short publication time and world-wide distribution. This permits a rapid and broad dissemination of
research results.

Advisory Board
Chairman
Nikhil R. Pal, Indian Statistical Institute, Kolkata, India
e-mail: nikhil@isical.ac.in
Members
Rafael Bello Perez, Universidad Central “Marta Abreu” de Las Villas, Santa Clara, Cuba
e-mail: rbellop@uclv.edu.cu
Emilio S. Corchado, University of Salamanca, Salamanca, Spain
e-mail: escorchado@usal.es
Hani Hagras, University of Essex, Colchester, UK
e-mail: hani@essex.ac.uk
László T. Kóczy, Széchenyi István University, Győr, Hungary
e-mail: koczy@sze.hu
Vladik Kreinovich, University of Texas at El Paso, El Paso, USA
e-mail: vladik@utep.edu
Chin-Teng Lin, National Chiao Tung University, Hsinchu, Taiwan
e-mail: ctlin@mail.nctu.edu.tw
Jie Lu, University of Technology, Sydney, Australia
e-mail: Jie.Lu@uts.edu.au
Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexico
e-mail: epmelin@hafsamx.org
Nadia Nedjah, State University of Rio de Janeiro, Rio de Janeiro, Brazil
e-mail: nadia@eng.uerj.br
Ngoc Thanh Nguyen, Wroclaw University of Technology, Wroclaw, Poland
e-mail: Ngoc-Thanh.Nguyen@pwr.edu.pl
Jun Wang, The Chinese University of Hong Kong, Shatin, Hong Kong
e-mail: jwang@mae.cuhk.edu.hk

More information about this series at http://www.springer.com/series/11156

Ajith Abraham Sergey Kovalev
•

Valery Tarassov Vaclav Snasel

•

Andrey Sukhanov
Editors

Proceedings of the Third

International Scientific
Conference “Intelligent
Information Technologies
for Industry” (IITI’18)
Volume 1

123
Editors
Ajith Abraham Vaclav Snasel
Scientific Network for Innovation VSB-Technical University of Ostrava
and Research Excellence Ostrava, Czech Republic
Machine Intelligence Research Labs
(MIR Labs) Andrey Sukhanov
Auburn, WA, USA Rostov State Transport University
Rostov-on-Don, Russia
Sergey Kovalev
Rostov State Transport University
Rostov-on-Don, Russia

Valery Tarassov
Bauman Moscow State Technical University
Moscow, Russia

ISSN 2194-5357 ISSN 2194-5365 (electronic)

Advances in Intelligent Systems and Computing
ISBN 978-3-030-01817-7 ISBN 978-3-030-01818-4 (eBook)
https://doi.org/10.1007/978-3-030-01818-4

Library of Congress Control Number: 2018958808

© Springer Nature Switzerland AG 2019

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, express 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 Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

This volume of Advances in Intelligent Systems and Computing contains papers

presented in the main track of IITI 2018, the Third International Scientific
Conference on Intelligent Information Technologies for Industry held in September
17–21 in Sochi, Russia. The conference was jointly co-organized by Rostov State
Transport University (Russia) and VŠB-Technical University of Ostrava (Czech
Republic) with the participation of Russian Association for Artificial Intelligence
(RAAI).
IITI 2018 is devoted to practical models and industrial applications related to
intelligent information systems. It is considered as a meeting point for researchers
and practitioners to enable the implementation of advanced information technolo-
gies into various industries. Nevertheless, some theoretical talks concerning the
state of the art in intelligent systems and soft computing were also included into
proceedings.
There were 160 paper submissions from 11 countries. Each submission was
reviewed by at least three chairs or PC members. We accepted 94 regular papers
(58%). Unfortunately, due to limitations of conference topics and edited volumes,
the program committee was forced to reject some interesting papers, which did not
satisfy these topics or publisher requirements. We would like to thank all authors
and reviewers for their work and valuable contributions. The friendly and wel-
coming attitude of conference supporters and contributors made this event a
success!
The conference was supported by Russian Fund for Basic Research (grant no.
18-07-20024 G).

September 2018 Ajith Abraham

Sergey M. Kovalev
Valery B. Tarassov
Václav Snášel
Andrey V. Sukhanov

v
Organization

Organizing Institutes

Rostov State Transport University, Russia

VŠB-Technical University of Ostrava, Czech Republic
Russian Association for Artificial Intelligence, Russia

Conference Chairs

Sergey M. Kovalev Rostov State Transport University, Russia

Alexander N. Guda Rostov State Transport University, Russia

Conference Vice-chair

Valery B. Tarassov Bauman Moscow State Technical University,

Russia

International Program Committee

Alexander I. Dolgiy JSC “NIIAS”, Rostov branch, Russia

Alexander L. Tulupyev
Alexander N. Shabelnikov
Alexander N. Tselykh
Alexander P. Eremeev
St. Petersburg Institute for Informatics
and Automation of the Russian Academy
of Sciences, Russia
JSC “NIIAS”, Russia
Southern Federal University, Russia
Moscow Power Engineering Institute, Russia

vii
viii Organization

Alexander V. Smirnov St. Petersburg Institute for Informatics

and Automation of the Russian Academy
of Sciences, Russia
Alexey B. Petrovsky Institute for Systems Analysis of Russian
Academy of Sciences, Russia
Alexey N. Averkin Dorodnicyn Computing Centre of Russian
Academy of Sciences
Alla V. Zaboleeva-Zotova Volgograd State Technical University, Russia
Anton Beláň Slovak University of Technology in Bratislava,
Slovakia
Dusan Husek Institute of Computer Science, Academy
of Sciences of the Czech Republic
Eid Emary Cairo University, Egypt
Eliska Ochodkova VSB-Technical University of Ostrava,
Czech Republic
František Janíček Slovak University of Technology in Bratislava,
Slovakia
Gennady S. Osipov Institute for Systems Analysis of Russian
Academy of Sciences, Russia
Georgy B. Burdo Tver State Technical University, Russia
Habib M. Kammoun University of Sfax, Tunisia
Hussein Soori VSB-Technical University of Ostrava,
Czech Republic
Igor B. Fominykh Moscow Power Engineering Institute, Russia
Igor D. Dolgiy Rostov State Transport University, Russia
Igor N. Rozenberg JSC “NIIAS”, Russia
Igor V. Kotenko St. Petersburg Institute for Informatics
and Automation of the Russian Academy
of Sciences, Russia
Ildar Batyrshin National Polytechnic Institute, Mexico
Ivan Zelinka VSB-Technical University of Ostrava,
Czech Republic
Jana Nowakova VSB-Technical University of Ostrava,
Czech Republic
Jaroslav Kultan University of Economics in Bratislava, Slovakia
Jiří Bouchala VŠB-Technical University of Ostrava,
Czech Republic
Jiří Hammerbauer University of West Bohemia, Czech Republic
Josef Paleček VŠB-Technical University of Ostrava,
Czech Republic
Juan Velasquez University of Chile, Chile
Konrad Jackowski Wrocław University of Technology, Poland
Leszek Pawlaczk Wrocław University of Technology, Poland
Marcin Paprzycki IBS PAN and WSM, Poland
Michal Wozniak Wroclaw University of Technology, Poland
Organization ix

Milan Dado University of Žilina, Slovakia

Mohamed Mostafa Arab Academy for Science, Technology,
and Maritime Transport, Egypt
Nadezhda G. Yarushkina Ulyanovsk State Technical University, Russia
Nashwa El-Bendary SRGE (Scientific Research Group in Egypt),
Egypt
Nour Oweis VSB-Technical University of Ostrava,
Czech Republic
Oleg P. Kuznetsov Institute of Control Sciences of Russian Academy
of Sciences
Pavol Špánik University of Žilina, Slovakia
Petr I. Sosnin Ulyanovsk state technical university, Russia
Petr Saloun VSB-Technical University of Ostrava,
Czech Republic
Santosh Nanda Eastern Academy of Science and Technology,
Bhubaneswar, Odisha, India
Sergey D. Makhortov Voronezh state university, Russia
Stanislav Kocman VŠB-Technical University of Ostrava,
Czech Republic
Stanislav Rusek VŠB-Technical University of Ostrava,
Czech Republic
Svatopluk Stolfa VSB-Technical University of Ostrava,
Czech Republic
Tarek Gaber VSB-Technical University of Ostrava,
Czech Republic
Teresa Orłowska-Kowalska Wrocław University of Technology, Poland
Vadim L. Stefanuk Institute for Information Transmission Problems,
Russia
Vadim N. Vagin Moscow Power Engineering Institute, Russia
Vladimir V. Golenkov Belarus State University of Informatics and
Radioelectronics, Belarus
Vladimír Vašinek VŠB-Technical University of Ostrava,
Czech Republic
Yuri I. Rogozov Southern Federal University, Russia
Zdenĕk Peroutka University of West Bohemia, Czech Republic

Organizing Committee Chair

Alexander N. Guda Rostov State Transport University, Russia

x Organization

Organizing Vice-chair

Andrey V. Sukhanov Rostov State Transport University, Russia

Local Organizing Committee

Andrey V. Chernov Rostov State Transport University, Russia

Anna E. Kolodenkova Samara State Technical University, Russia
Ivan A. Yaitskov Rostov State Transport University, Russia
Jan Platoš VSB-Technical University of Ostrava,
Czech Republic
Maria A. Butakova Rostov State Transport University, Russia
Maya V. Sukhanova Azov-Black Sea State Engineering Institute,
Russia
Pavel Krömer VSB-Technical University of Ostrava,
Czech Republic
Vitezslav Styskala VSB-Technical University of Ostrava,
Czech Republic
Vladislav S. Kovalev JSC “NIIAS”, Russia
Contents

Invited Papers
Connected Vehicle Prognostics Framework for Dynamic Systems . . . . . 3
Omar Makke and Oleg Gusikhin
Human-Computer Cloud for Decision Support: Main Ontological
Models and Dynamic Resource Network Configuration . . . . . . . . . . . . . 16
Alexander Smirnov, Tatiana Levashova, Nikolay Shilov,
and Andrew Ponomarev
Enterprise Total Agentification as a Way to Industry 4.0:
Forming Artificial Societies via Goal-Resource Networks . . . . . . . . . . . . 26
Valery B. Tarassov

Data Mining and Knowledge Discovery in Intelligent Information

and Control Systems
Context-Dependent Guided Tours: Approach
and Technological Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Alexander Smirnov, Nikolai Shilov, and Oleg Gusikhin
Retention to Describe Knowledge of Complex Character
and Its Formalization in Category Theory . . . . . . . . . . . . . . . . . . . . . . . 51
A. V. Zhozhikashvily and V. L. Stefanuk
Proximity of Multi-attribute Objects in Multiset Metric Spaces . . . . . . . 59
Alexey B. Petrovsky
Evidence Theory for Complex Engineering System Analyses . . . . . . . . . 70
Boris Palyukh, Vladimir Ivanov, and Alexander Sotnikov
Analysis of Software Development Process in Respect
to Anomaly Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Denis Zavarzin and Tatyana Afanaseva

xi
xii Contents

A Model of Multiagent Information and Control System

Distributed Data Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Eduard Melnik and Anna Klimenko
Representation and Use of Knowledge for the Reconfiguration
of the Mechanical Transport System . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Stanislav Belyakov, Marina Savelyeva, and Igor Rozenberg
Synthesis of Adaptive Algorithms for Estimating the Parameters
of Angular Position Based on the Combined Maximum Principle . . . . . 107
Andrey Kostoglotov, Sergey Lazarenko, Anton Penkov, Igor Kirillov,
and Olga Manaenkova
Synthesis of Intelligent Discrete Algorithms for Estimation with Model
Adaptation Based on the Combined Maximum Principle . . . . . . . . . . . . 116
Andrey Kostoglotov, Sergey Lazarenko, Igor Pugachev,
and Alexey Yachmenov

Ontological Modeling, Semantic Technologies

and Knowledge Engineering
The Approach to Extracting Semantic Trees from Texts to Build
an Ontology from Wiki-Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Nadezhda Yarushkina, Aleksey Filippov, Vadim Moshkin,
and Ivan Dyakov
Ontology-Based Semantic Models for Industrial IoT
Components Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Nikolay Teslya and Igor Ryabchikov
An Approach to Optimization of Ray-Tracing in Volume
Visualization Based on Properties of Volume Elements . . . . . . . . . . . . . 148
Nikolai Vitiska, Vladimir Selyankin, and Nikita Gulyaev
Discovering of Part-Whole Relations Used in Architectural
Prototyping of Project Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Petr Sosnin and Anna Kulikova
Multi-level Ontological Model of Big Data Processing . . . . . . . . . . . . . . 171
Victoria V. Bova, Vladimir V. Kureichik, Sergey N. Scheglov,
and Liliya V. Kureichik
Ontological Modeling for Industrial Enterprise Engineering . . . . . . . . . 182
Alena V. Fedotova, Vadim V. Tabakov, Michael V. Ovsyannikov,
and Jens Bruening
Contents xiii

Designing the Knowledge Base for the Intelligent Inertial Regulator

Based on Quasi-optimal Synthesis of Controls Using the Combined
Maximum Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Andrey Kostoglotov, Sergey Lazarenko, Alexander Agapov,
Zoya Lyaschenko, and Irina Pavlova

Fuzzy Graphs, Fuzzy Networks and Fuzzy Inference for Planning

and Cognitive Modelling
Assessing the Software Developer’s Quality Using Fuzzy Estimates . . . . 203
Tatiana Afanasieva and Vlad Moiseev
A Fuzzy Control Method for Priority Driven Embedded Device . . . . . . 213
Karolina Janosova, Michal Prauzek, Jaromir Konecny, Monika Borova,
and Martin Stankus
Interpretability of Fuzzy Temporal Models . . . . . . . . . . . . . . . . . . . . . . 223
Alexander N. Shabelnikov, Sergey M. Kovalev, and Andrey V. Sukhanov
Control of the Cognitive Process in Hard Real-Time Environment
in the Context of the Extended Stepping Theories of Active Logic . . . . . 235
Michael Vinkov, Igor Fominykh, and Sergey Romanchuk
Method of the Maximum Dynamic Flow Finding in the Fuzzy
Graph with Gains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Victor Kureichik and Evgeniya Gerasimenko
Hybrid Bioinspired Algorithm of 1.5 Dimensional Bin-Packing . . . . . . . 254
Boris K. Lebedev, Oleg B. Lebedev, and Ekaterina O. Lebedeva
Community Detection in Online Social Network Using Graph
Embedding and Hierarchical Clustering . . . . . . . . . . . . . . . . . . . . . . . . 263
Vang Le and Vaclav Snasel
Cognitive Visualization of Carbon Nanotubes Structures . . . . . . . . . . . . 273
Vadim A. Shakhnov, Lyudmila A. Zinchenko, Vadim V. Kazakov,
Andrei A. Glushko, Vladimir V. Makarchuk, and Elena V. Rezchikova
Decision Making Model for Outsourcing by Analysis of Hierarchies
of T. Saaty Under Fuzzy Environment . . . . . . . . . . . . . . . . . . . . . . . . . . 280
Natalia Egorova and Yana Sorokina

Evolutionary Modeling, Bionic Algorithms

and Computational Intelligence
Forecasting of Results of Dynamic Interaction Between Space Debris
and Spacecrafts on the Basis of Soft Computing Methods . . . . . . . . . . . 293
Boris V. Paliukh, Valeriy K. Kemaykin, Yuliya G. Kozlova,
and I. V. Kozhukhin
xiv Contents

A Fuzzy Evaluation of Quality for Color Vision Disorders Diagnostic . . . 303

Monika Borova, Jaromir Konecny, Michal Prauzek,
and Karolina Janosova
Improving the Efficiency of Solution Search Systems Based
on Precedents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Alexander Eremeev, Pavel Varshavskiy, and Roman Alekhin
Efficient Feature Selection Algorithm Based on Population
Random Search with Adaptive Memory Strategies . . . . . . . . . . . . . . . . 321
Ilya Hodashinsky, Konstantin Sarin, and Artyom Slezkin
Applying Fuzzy Computing Methods for On-line Monitoring
of New Generation Network Elements . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Igor Kotenko, Igor Saenko, and Sergey Ageev
An Approach to Similar Software Projects Searching and
Architecture Analysis Based on Artificial Intelligence Methods . . . . . . . 341
Yarushkina Nadezhda, Guskov Gleb, Dudarin Pavel,
and Stuchebnikov Vladimir
Analysis of the Dynamics of the Echo State Network Model Using
Recurrence Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Emmanuel Sam, Sebastian Basterrech, and Pavel Kromer
Query Answering over Some Extensions of Allen’s Interval Logic . . . . . 362
Gerald S. Plesniewicz
Fuzzy Topological Approach to a Solid Control Task . . . . . . . . . . . . . . 373
Andrei Kostoglotov, Vladimir Taran, and Vladimir Trofimenko
Time Series Grouping Based on Fuzzy Sets and Fuzzy Sets Type 2 . . . . 382
Anton Romanov and Irina Perfilieva
Using the Concept of Soft Computing to Solve the Problem
of Electromagnetic Compatibility Control . . . . . . . . . . . . . . . . . . . . . . . 390
Vladimir Taran, Aleksey Shandybin, and Elena Boyko
Generation of Efficient Cargo Operation Schedule at Seaport
with the Use of Multiagent Technologies and Genetic Algorithms . . . . . 401
Olga Vasileva and Vladimir Kiyaev
Resource Managing Method for Parallel Computing Systems
Using Fuzzy Data Preprocessing for Input Tasks Parameters . . . . . . . . 410
Anastasia Voitsitskaya, Alexader Fedulov, and Yaroslav Fedulov

Cognitive Technologies on the Basis of Sensor and Neural Networks

Post-processing of Numerical Forecasts Using Polynomial Networks
with the Operational Calculus PDE Substitution . . . . . . . . . . . . . . . . . . 423
Ladislav Zjavka and Stanislav Mišák
Contents xv

Visibility Loss Detection for Video Camera Using Deep

Convolutional Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
Alexey Ivanov and Dmitry Yudin
Neural Network Control Interface of the Speaker Dependent
Computer System «Deep Interactive Voice Assistant DIVA»
to Help People with Speech Impairments . . . . . . . . . . . . . . . . . . . . . . . . 444
Tatiana Khorosheva, Marina Novoseltseva, Nazim Geidarov,
Nikolay Krivosheev, and Sergey Chernenko
Hierarchical Reinforcement Learning with Options and United
Neural Network Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Vadim Kuzmin and Aleksandr I. Panov
New Approaches to Discrete Modeling of Natural Neural Networks . . . 463
Oleg Kuznetsov, Ludmila Zhilyakova, Nikolay Bazenkov,
Boris Boldyshev, and Sergey Kulivets
Meta-Optimization of Mind Evolutionary Computation Algorithm
Using Design of Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Maxim Sakharov and Anatoly Karpenko
An Interpreter of a Human Emotional State Based on a Neural-Like
Hierarchical Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Konstantin V. Sidorov, Natalya N. Filatova, and Pavel D. Shemaev
A Vessel’s Dead Reckoning Position Estimation by Using
of Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
Victor V. Deryabin and Anatoly E. Sazonov
Evaluation of Neural Network Output Results Reliability
in Pattern Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
Daniil V. Marshakov, Vasily V. Galushka, Vladimir A. Fathi,
and Denis V. Fathi
A Neural-Like Hierarchical Structure in the Problem of Automatic
Generation of Hypotheses of Rules for Classifying the Objects
Specified by Sets of Fuzzy Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
Konstantin V. Sidorov, Natalya N. Filatova, and Pavel D. Shemaev
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
Invited Papers
 Connected Vehicle Prognostics Framework
for Dynamic Systems

Omar Makke and Oleg Gusikhin(&)

Research and Advanced Engineering, Ford Motor Company,

20300 Rotunda Drive, Dearborn, MI 48124, USA
{omakke,ogusikhi}@ford.com

Abstract. Connected vehicle analytics has a promise to substantially advance

vehicle prognostics and health management. However, the practical imple-
mentation of connected vehicle prognostics faces a number of challenges, such
as the limitation of communication bandwidth resulting in potential loss of data
that is critical for adequate prognostics models. The paper discusses a modelling
framework for connected vehicle prognostics for dynamic systems that allows
addressing connectivity limitations and memory constraints. The framework is
based on a hybrid prognostics approach combining in-vehicle physics-based
data aggregation model and cloud-based data-driven prognostics leveraging
cross-vehicle and external data sources. The application of the framework is
illustrated by models for brake pads wear and cabin air filter prognostics.

Keywords: Prognostics
Connected vehicles
Machine learning

Big data analytics

1 Introduction

Vehicle prognostics and health management has always been a topic of great interest
for the automotive industry. It is one of the early areas of intelligent systems appli-
cations targeting on-board and service bay diagnostics and prognostics [1]. In recent
years, due to increased complexity of the vehicle systems and the emergence of con-
nected and autonomous vehicles, the interest in prognostics and health management has
soared.
Connected vehicle technology brings game changing opportunities to advance
vehicle prognostics and health management [2]. Connected vehicles can take advantage
of cloud resources to build complex models, and they can leverage data from multiple
vehicles and other external sources to infer complex relationships and offer enhanced
user experience. Even existing on-board monitoring and diagnostics systems can be
enhanced using connected vehicle technology. For example, modern vehicles are
equipped with the oil-life monitoring system that tracks oil wear by analyzing duty-
cycle and typically provide advanced notice to the driver when the oil life is at 5%.
However, for scheduling purposes, it is more convenient to estimate the “time to oil
change” rather than the oil wear percentage. By using connected vehicle data, it is
possible to project the time to oil wear, assuming consistent duty cycle. Then, a
notification containing projected time to service can be provided to the customer.

© Springer Nature Switzerland AG 2019

A. Abraham et al. (Eds.): IITI 2018, AISC 874, pp. 3–15, 2019.
https://doi.org/10.1007/978-3-030-01818-4_1
4 O. Makke and O. Gusikhin

Extending the service need prediction to other components, such as brake pads,
tires, batteries, air filter, etc. will provide a comprehensive service-scheduling package,
which offers a new level of user experience and service efficiency. This capability can
insure that the dealer has the right replacement parts available. It can also improve the
efficiency of service parts inventory. For fleet owners, it facilitates better scheduling to
minimize the impact of vehicle downtime on the fleet operations.
Connected vehicle analytics has a promise to develop such models utilizing exiting
vehicle data to project the time to service without adding costly sensors or other
hardware. However, practical implementation of connected vehicle prognostics faces a
number of challenges such as on-board sensor data, memory availability, and limita-
tions of the wireless communication bandwidth [3]. These limitations result in potential
data loss of data which may be crucial for adequate prognostics of the given compo-
nent. To overcome these challenges, the design in [4] proposed the model for con-
nected vehicle brake prognostics, where an on-board system implements a duty-cycle
aggregation model to alleviate limitations of communication channel while a cloud
based model uses cross-vehicle data in lieu of the lack of direct measurement of brake
wear. In this paper we generalize and formalize the approach presented in [4] by
providing a theoretical framework to build such prognostics models for the components
which can be modelled as dynamic systems.
This paper is organized as follows. Section 2 provides an overview of the proposed
framework. Section 3 illustrates the framework using brake pad prognostics as an
example. Section 4 presents the detailed mathematical foundation behind the frame-
work. Section 5 demonstrates how the same framework can be applied to a different
domain area, such as the cabin air filter prognostics. Section 6 concludes the paper and
discusses future work.

2 Connected Vehicle Prognostics Framework

This section describes a prognostics framework for a class of problems commonly

found in automotive vehicles. For the discussion of this paper, note that typically
prognostics are classified into 3 categories [5] summarized in Fig. 1.

Fig. 1. Three different prognostics models

The first is physics based prognostics. In this approach, elaborate physics models
are needed with details about many of the components material properties, such as
coefficient of friction, mass, heat capacity, and other material properties. The second
approach is a data-driven approach which is based solely on data without any
 Connected Vehicle Prognostics Framework for Dynamic Systems 5

consideration to the underlying physics. The goal here is either to use machine learning
or other statistical tools in order to identify patterns in the data and make projections on
when to service a component. In our framework, we will utilize a hybrid approach
where an in-vehicle physics-based model generates data in a meaningful way, and this
data is sent periodically to the cloud to make predictions using data-driven approaches,
which feeds back a wear result to the vehicle. These in-vehicle models should be
designed to be embeddable (in particular, through over the air updates) into vehicle
control modules, such as modem, engine controller, brake controller, etc.
The problem can be formulated as follows. Suppose that the dynamic behavior of a
component can be described with a physics-based dynamic model, requiring a vector
u of inputs which affects a state vector x. The wear of the component is assumed to be a
function of the inputs and the states. The goal is to find a prognostics algorithm which
outputs “expected time to service” by calculating the “wear” of the component, and
based on the past history, predicts a time to “service component by”.
To minimize the reliance on multiple parameters related to material properties and
environment, machine learning is used. The algorithm must run in the vehicle within
deterministic time period, must have resistance to connectivity issues, must require
fixed size of storage space in the vehicle, and must consume a deterministic amount of
data when transferring the data from the vehicle to the cloud. The framework which
implements such algorithms with these requirements is shown in Fig. 2.

Fig. 2. Connected vehicle prognostics framework

Using this approach, a fleet manager with a known number of vehicles can pur-
chase a data plan adequate enough to support the data transfer, or can ensure that the
fleet vehicles are connected to the cloud at least once every specified period of time,
such as day, a week, or similar. The vehicle generates data based on duty cycle, and the
cloud utilizes this data in machine learning. As will be shown in the following sections,
the method of generating data is an important factor in combining the physics approach
with machine learning.
6 O. Makke and O. Gusikhin

3 Brake Pad Prognostics

The main objective of the brake pad prognostics system is to give customers a close
approximation of the date when they need to change their brake pads. The algorithm
described here will serve as an example to help in understanding the general framework
for prognostics explained in the next section. The classical approach to this problem is
to create an elaborate physics model which relies heavily on material properties of the
pads and execute it in the vehicle [6]. One simple form of prognostics is a distance-
based prediction. A driver is asked to inspect the brake pads every 10k to 15k miles or
similar, found in the vehicle’s manual. Therefore, it is expected that using a physics
based model, yet a simplified model compared to an elaborate 3D finite element model
as shown in [7], a more accurate prediction can be made. The goal is to predict the wear
so that at the next oil change pads can be replaced.

Fig. 3. Brake pad prognostics system architecture

The architecture of the brake pad prognostics system is shown in Fig. 3. In general,
the characteristics of the wear on the pad depend on many variables. For the aim of
prognostics, a simplified model will be used, keeping in mind that machine learning
will be applied later to improve the model’s accuracy. In this simplified lumped model,
the wear on the brake pad is a function of the instantaneous temperature of the pad and
the instantaneous energy extracted (the power) from the vehicle’s kinetic energy as heat
[6]. In a lab environment, one may create a setup with the average temperature of the
pad being controlled so that it only fluctuates in a range DT (such as 50 °C fluctua-
tions), and the pad pressure is applied to a rotating disk for measurable durations, and a
table similar to Table 1 is generated. The pad mass is modelled by

dT P
¼  r ð T  T1 Þ ð1Þ
dt mC
dm
¼ wðT; PÞ ð2Þ
dt
 Connected Vehicle Prognostics Framework for Dynamic Systems 7

Here, w(T, P) is the wear from Table 1 and T∞ = Tamb. Initial conditions are
T = Tamb, m = m0, and mC is the mass multiplied by specific heat capacity of the brake
pad, and is assumed to be constant for a short duration of time, measured in days. It is
assumed that the processor solving Eq. (1) does not have access to w(T, P), and
therefore it cannot calculate the change in the mass instantaneously. The temperature
T in Eq. (1) is initialized to Tamb at ignition. This model’s accuracy will deteriorate
when almost all the mass of the pads is worn out, but at that time, the algorithm will
already have predicted that it is time to change the brake pads. Although this
approximation has errors, due to the fact that big data and machine learning will be
used as will be shown later, the error is expected to be compensated for, since it is not a
random error.

Table 1. Showing w(T, P). Here, ai,j (g/s) is the average grams lost given a power at a average
temperature for 1 s.

w Temperature (Celsius)
[-40 0] [0 40] [40 80] …
Power (Watts)

[0 100] a0,0 a0,1 a0,2 …

[100 200] a1,0 a1,1 a1,2 …

[200 300] a2,0 a2,1 a2,2 …

… … … … …

The power P is derived from vehicle speed and the friction brake torque for each
wheel after considering the loading effect ratios and the heat partitioning factor. Based
on the calculated temperature and power at a given time t, for given (T, P), Table 1,
provides the wear factor w(T,P), so that the instantaneous wear is:

dW ¼ wðT; PÞdt ð3Þ

t
Zf
W ¼ wðT; PÞdt ð4Þ
0

Here, W is the total wear, in grams, of the brake pad until a final time tf.. Both
Eqs. (3) and (4) which is dependent on material properties will be solved in the cloud.
The wear in this model is a function of two variables, the state T and the input P. To
solve the model, create a 2D mesh in the vehicle’s memory with T and P as axes,
similar to Table 1. Set a minimum and maximum for each axis:

40  C  T  400  C and 0W  P  10000Ws ð5Þ

Partition each axis into non-intersecting intervals. Smaller partitions give higher
accuracy but require more storage. Each axis may have different number of partitions.
8 O. Makke and O. Gusikhin

This will create a 2D mesh with cells in it. Let set EM be the set of all points M which
lie in the same cell. Suppose that the algorithm runs at a rate s ms. (e.g. 5 ms). Then, at
every s ms, calculate using Eq. (1) both P(t), and T(t) where t is the current time.
Locate the set EM in the 2D mesh such that EM contains point M with coordinates (P(t),
T(t)). Then increment:

ValueðEM Þ ¼ ValueðEM Þ þ PðtÞs ð6Þ

Then, it is evident that the set EM is a set of points approximating infinitesimal

energy values which can occur at separate times. The sum of the values of all the sets
EM for all points M in the 2D mesh is equivalent to simply integrating P(t) over time.
Figure 4 shows a hypothetical power curve as a function of time, and the values of the
pad’s temperature which fall within the same range are collected together. Since the set
EM at (T, P) is the accumulation of P(t)s in a temperature and power range, EM is the
measure of how much energy the brake pad has dissipated at (P, T). This is also shown
in Fig. 4. Since the integration formula of wear is approximated by summation, which
is commutative, the final wear of Eq. (4) can be calculated by:
X
W¼ Cells
wðT; PÞECell ð7Þ

In order to utilize big data and machine learning, the data in the 2D mesh is sent to
the cloud. The data has a known size in terms of kilobytes, and hence the requirements
can be known before implementing it in the vehicle. The final architecture of the
dynamic system is shown in Fig. 5.

Fig. 4. Clustering a hypothetical curve based on both T and P. The intensity of each cell in the
2D mesh represents the sum of PðtÞs at different values of t, at coordinate (P,T).

The cloud can use big data and machine learning to derive the actual coefficients w
(T,P) if field measurements can be provided. To achieve this, experimental vehicles can
be equipped with sensors which generate data for the cloud to learn the wear factors for
each cell in Fig. 4. Alternatively, measurements can be taken during a service visit at
the dealer. Then Table 1 is constructed using machine learning and is made available in
the cloud.
 Connected Vehicle Prognostics Framework for Dynamic Systems 9

Fig. 5. Brake prognostics algorithm is separated between vehicle and cloud. Here, m is assumed
constant for short durations in the vehicle.

An alternative to the aggregation method used in Eq. (6) is to only increment an

integer counter in every cell, instead of aggregating a quantity. Section 4 will elaborate
more about the mathematics behind this. The process is as follows. Pick any cell EM. If
the data in EM is the accumulated energy P(t)s at given (P, T), it is possible to divide
that total energy by P to obtain the cumulative time the brake pad was used at the given
temperature and power. By multiplying the cumulative time in each cell by its tem-
perature or power coordinate, one can obtain a diagram similar to that in Fig. 4.
Engineers can study temperature distributions of the usage of the pad, power distri-
butions, or simply the time pads are used at a given temperature and power. Engineers
can then make better assumptions about how the pads are used in order to improve their
designs. In this approach, instead of the parameter “time” being the axis and power and
temperature are the calculated values, temperature and power become the axes and time
becomes the aggregated value. This can be viewed as a method to compress data
generated over time, and E(T, P) becomes a simple counter as will be seen in the next
section.

4 Non-linear Dynamic Systems Prognostics

This section addresses prognostics of components which can be represented using non-
linear models. It is desired to have a universal approach for these components. In this
approach, we would like to reduce the number of experiments conducted in the lab to
characterize the wear of the components, and replace them by using connected vehicles
and machine learning. Without loss of generality, the initial state of the system is
assumed to be known for a new part, and is 0. Suppose that the state space model for a
system is

x_  ¼ f ðx ðtÞ; uðtÞÞ ð8Þ

_ ¼ yðtÞ ¼ gðx ðtÞ; uðtÞÞ

W ð9Þ
10 O. Makke and O. Gusikhin

Under the assumption that: if the degradation of the states which are used in
prognostics is slow compared to the dynamics of the system, then the states repre-
senting the degradation are assumed to be constant for that duration (a day or a week),
then the input vector x* can be separated into:
2 3
x
x ¼ ½x0 ; x1 ; . . .; xn jbx 0 ; bx 1 ; . . .; bx m T ¼ 4  5 ð10Þ
b
x
 
@ xbk 
where    1 for 0  k  m ð11Þ
@t

And then all xbk can be treated as constants for short periods of time.
W_ in Eq. (9) represents the instantaneous wear, and hence W is the total wear. The
goal is to calculate W on a different processor (in the cloud) than that which calculates
x*.
W is a vector of total wear of several subcomponents of the component under
monitor, and intuitively, we can select the worst value to be an answer. The correct
answer will depend on the nature of the problem and its impact on customers.
The following qualitative conclusion can then be made: The state Eq. (8) captures
how the component is being used, and the output Eq. (9) calculates a measure of
interest, based on the usage. The vehicle calculates and aggregates the behavioral
model, while the cloud calculates any y of interest by using a corresponding g(x*,u),
which itself can be learnt using big data and a true measurement of y using either a
sensor or manual measurement of selected few components. Algorithm 1 is used to
aggregate the data in the vehicle.

Algorithm 1 In-Vehicle Data Aggregation

1: Create hyperrectangle of dimension n + r where n and r are the
dimensions of the state and input vectors respectively
2: Identify the minimum and maximum of each axis in the hyperrectangle
4: Partition the hyperrectangle into independent partitions
5 At Key OFF -> ON: Set initial conditions and load values from memory
8: while KEY ON
9: Solve equation (8) numerically
10: Locate partition E in hyperrectangle containing point M identified
by the values . If any of the
coordinates exceed the boundary, assume their value is the
boundary
11: Increment the value in partition E identified in step 10 by 1.
= +1
12: end while

The equations are discretized and the errors are assumed to be negligible since a
customer has few weeks to change the brake pads, oil, air filter, and not few minutes. In
 Connected Vehicle Prognostics Framework for Dynamic Systems 11

order to calculate W from Eq. 9, instead of integrating g(x*,u) over time, by summing
the values in time, sample by sample, we partitioned the range of g(x*,u) into cells
where each cell represents accumulation of values of W _ at same pair (x*, u), and
summed the values in the partition (the is commutative). This is described in details as
follows.
By integrating Eq. (9), W can be obtained by summing over all samples N as

X
N
W¼ gðxðti Þ; uðti ÞÞs ð12Þ
i

Without loss of generality, assume W and g to be one dimensional. Since

gðxðtÞ; uðtÞÞ is memoryless, then for the same inputs, the same output is obtained.
Define ci to be the value of g at time ti. The value ci represents the wear value at the
conditions of (x, u). It is possible that at different values of t, (x, u) is the same, and
hence g(x, u) is the same.

ci ¼ gðxðti Þ; uðti ÞÞ ð13Þ

Since the summation is commutative, collect the values of g(x,u) at different times
together for which the pair (x,u) are similar and call this set Si. Therefore, Si is a
collection of values ci. Define gSi to be the sum of the values in set Si.

gSi ¼ ni ci ð14Þ

Then, W can be written as

X X
W¼ gSi s ¼ c i ni s ð15Þ
i i

But ni has already been calculated in the vehicle, in step 11 of the algorithm, as EM
and therefore, the wear W is found to be
X
W¼ ci Ei s ð16Þ
i

Note that the slow changing states, which are used for prognostics, are assumed to
be constants over short durations. But, whenever the cloud updates the wear prediction,
it calculates new values for xbk and sends them to the vehicle. Hence the model
approximation remains dynamic, running at two different rates, yielding a better
approximation than assuming that these states are always constant. The wear calcu-
lation then feeds into a time prediction algorithm which completes the prognostics for
the component. The time prediction algorithm can vary in complexity as needed, and
this not an issue because it runs in the cloud where resources are not as constraint as the
vehicle. Figure 6 shows the machine learning system in the cloud.
The last remaining part is to derive the values of ci in the cloud. By either providing
manual wear measurements W, or inserting sensors in a subset of the vehicles to
12 O. Makke and O. Gusikhin

measure this wear, we can use machine learning and optimization to derive the outputs
ci. Moreover, Eq. (16) has the form of a linear regression which also resembles a
simple neural network where the activation function is just a pass-through node and in
this case, the weights of the network are the coefficient ci. and the inputs are Eis.

Fig. 6. Machine learning in cloud.

However, if W is a non-autonomous dynamic system and not a function of only

x and u, as previously assumed, recurrent neural networks can be used to approximate
the wear [8]. Care should be taken in this case to take enough data points throughout
the wear of the component to have enough data to model the dynamics of the system.
Figure 7 shows the final architecture of the system.

Fig. 7. The proposed prognostics system. The algorithms are executed on two compute entities,
in the vehicle and in the cloud.

If needed, extra inputs can be added to the system in Fig. 6, such as geographical
location, season, vehicle type, etc. when a neural network is used to add more infor-
mation to the system to reduce errors unaccounted for in the models, such as effects of
salt, dust, etc.

5 Cabin Air Filter Prognostics

In this section, we will illustrate the application of the framework to cabin air filter
prognostics. Given air velocity Uin of the air feeding into the air filter, and concen-
tration of particulates before the air filter Nin, it can be shown [9] that the rate of change
of accumulation of particles, A, in the air filter to be:
 Connected Vehicle Prognostics Framework for Dynamic Systems 13

dA
¼ Uin Nin ð1  Pð AÞÞ ð17Þ
dt
In Eq. (17), P(A) is the filter penetration defined as the ratio of the particle con-
centration after the air filter over the particle concentration before the inlet.

Nout
Pð AÞ ¼ ð18Þ
Nin

The wear here is A, the accumulation of particles in the air filter, and the larger A,
the more wear. By integrating Eq. (17) and assuming initial wear is 0, the total wear
can be found to be:

Zt
AðtÞ ¼ Uin ðsÞNin ðsÞ½1  PðAðsÞÞds ð19Þ
0

To transform this model to fit in the connected prognostics framework, start by

noting that the air velocity Uin is related to blower speed B in percentage, recirculation
air door opening R in percentage, and vehicle speed V. Assume that there exists a
sensor in the vehicle which can be used to approximate particle concentration Nout.
Then

Nout
Nin ¼ ð20Þ
Pð AÞ

The system’s inputs are blower speed B, recirculation door position R, the con-
centration Nin, and the vehicle speed V, and the states are air velocity into the air filter
Uin, the concentration after the filter Nout, and the wear A. A is assumed to be slow
changing for the duration of few hours or days, depending on the region. Moreover,
here, Uin is assumed to be approximated as a function with arguments A, V, B and R,
and that it is not a dynamic system. This assumption is equivalent to saying that the
steady state response of Uin after changing any of the inputs is reached quickly. Under
this assumption, only A, V, B and R are needed to be known. Therefore,

Uin ¼ f ðA; B; R; V Þ ð21Þ

x ¼ ½xj^xT ¼ ½Uin ; Nout jAT and u ¼ ½B; R; Nin ; V T ð22Þ

Apply the framework as follows: Create a 4D hyperrectangle with axes B, R, V, and

Nout. Uin is ignored because it can be calculated directly from V, B, R, and A. The value
of A is known in the vehicle and the cloud. Knowing P(A) requires either knowing Nin
or Nout, but Nout is chosen since it is directly measured. Partition each axis into sets
where each set represents a cell. Assume air blower has 7 levels, and air recirculation
door has 5 levels. Nout axis is divided into 20 levels, from 0 to 1000 µg/m3. Vehicle
speed is partitioned into 4 levels from 0 to 80 mph. The hyperrectangle E is sent to the
cloud on timely basis based on the geographical location. The initial value of P(A) is
14 O. Makke and O. Gusikhin

known by the OEM. The OEM also has derived the mapping between A and P
(A) based on experimental data. The cloud calculates the wear for each cell Ei of value
ni in the hyperrectangle

_ cell ¼ A_ cell ¼ f ðA; B; R; V Þ Nout ð1  Pð AÞÞni

W ð23Þ
Pð AÞ

And then, the total wear A is found by summing the individual wears in each cell,
which happens in the cloud as shown in Eq. (24). The parameters except ni are known
from the axes coordinates, and ni is accumulated in the vehicle.

X
Cells
Nout
A¼ f ðA; B; R; V Þ ð1  Pð AÞÞni ð24Þ
i
Pð AÞ

After executing Eq. (24), a new value of P(A) is obtained from the already existing
lookup table. The problem reduces to knowing the function f. This function can be
obtained by using neural networks as global approximators, and data can be generated
before production by measuring in a vehicle the air velocity under different conditions.
The system is shown in Fig. 8. The function f is chosen to be in the cloud since it can
be improved over time. The values of A and P(A) are sent to the vehicle for use in other
features.

Fig. 8. In cabin air filter prognostics

6 Conclusion

In this paper, a connected vehicle prognostics framework for dynamic systems is

proposed and discussed. The approach presented here generalizes and formalizes the
model which is presented in [4]. In this framework, the in-vehicle physics based model
generates fixed-sized data in a form suitable for machine learning and big data analysis
where neural networks can be applied as universal approximators for wear functions.
Since the states under monitor are assumed to change slowly over time compared to the
rest of the states of the dynamic system, it is shown that it is possible to combine big
 Connected Vehicle Prognostics Framework for Dynamic Systems 15

data analysis and dynamic systems theory to create a hybrid solution. Furthermore, the
theoretical insight is provided to show what needs to be aggregated in the vehicle, what
assumptions are made, and how to map this data to machine learning inputs. The data
generation is simplified to aggregating the time in a hyperrectangle which has the states
and inputs of the dynamic system as axes. The application of the framework to two
dynamic models: brake pad wear prediction and in cabin air filter life, is discussed. If
more accuracy is desired, more elaborate models can be applied under same
framework.
The cloud system can use machine learning to predict the time to service com-
ponent, which can combine both machine learning and input from customers’ calendar
information to optimize the service scheduling for the vehicle.
This approach provides resistance to connectivity issues since the aggregated data
in the vehicle has a long time window, in terms of days, to connect once to the cloud to
transfer the aggregated data and obtain a new wear value.
The verification of the accuracy of the models still requires large set of field data.
Upon receiving sufficient amount of data, we will analyze the prediction and if nec-
essary, we will apply higher fidelity models. The future work includes transforming
other models of components and subsystems such as fuel pumps, tires, and batteries to
this framework.

References
1. Gusikhin, O., Rychtyckyj, N., Filev, D.: Intelligent systems in the automotive industry:
applications and trends. Knowl. Inf. Syst. 12, 147–168 (2007)
2. Siegel, J., Erb, D., Sarma, S.: A survey of the connected vehicle landscape–architectures,
enabling technologies, applications, and development areas. IEEE Trans. Intell. Transp. Syst.
19, 2391–2406 (2017)
3. Zhang, Y., Gantt Jr., G., Rychlinski, M., Edwards, R., Correia, J., Wolf, C.: Connected
vehicle diagnostics and prognostics, concept and initial practice. IEEE: Trans. Reliab. 58,
286–294 (2009)
4. Zagajac, J., Chopra, A., Krivtsov, V., Gusikhin, O.: Method and apparatus for connected
vehicle system wear estimation and maintenance scheduling. US Patent Application
US2016/0163130 A1 (2016)
5. Pecht, M.: Prognostics and Health Management of Electronics. Wiley-Interscience, New York
(2008)
6. Zhang, S., Chen, W., Li, Y.: Wear of friction material during vehicle braking. In: SAE
Technical Paper 2009-01-1032 (2009)
7. Bakar, A.R.A., Ouyang, H., Khai, L.C., Abdullah, M.S.: Thermal analysis of a disc brake
model considering a real brake pad surface and wear. Int. J. Veh. Struct. Syst. 2(1), 20–27
(2010)
8. Kambhampati, C., Graces, F., Warwick, K.: Approximation of non-autonomous dynamic
systems by continuous time recurrent neural networks. In: Proceedings of the IEEE-INNS-
ENNS International Joint Conference on Neural Networks (2000)
9. Billings, E.C.: Effects of particle accumulation in aerosol filtration. Ph.D. dissertation,
California Institute of Technology. http://resolver.caltech.edu/CaltechETD:etd-09172002-
113217. Accessed 14 May 2018
Another random document with
no related content on Scribd:
manufacture of textile machinery he added that of general machinery
and large tools for cutting, boring, rifling, planing and slotting. He had
a great reputation in his day, but his work seems to have been more
that of a builder of standard tools than an originator of new tools and
methods.
Charles Holtzapffel, another well-known engineer of that
generation, was the son of a German mechanic who came to
London in 1787. He received a good education, theoretical as well
as practical, and became a skilled mechanician and a tool builder of
wide influence. His principal book, “Turning and Mechanical
Manipulation,” published in 1843 in three volumes, is an admirable
piece of work. Covering a field much wider than its title indicates, it is
the fullest and best statement of the art at that time; and scattered
through it there is a large amount of very reliable mechanical history.
By 1840 the number of men engaged in tool building was
increasing rapidly, and it is impossible to consider many English tool
builders who were well known and who did valuable work, such as
Lewis of Manchester, B. Hick & Son of Bolton, and others. One
noteworthy man, however, ought to be mentioned—John George
Bodmer, who was neither an Englishman, nor, primarily, a tool
builder.[71] He was a Swiss who worked in Baden and Austria, as
well as in England, and his fertile ingenuity covered so many fields
that a list of the subjects covered by his patents occupy six pages in
the “Transactions of the Institution of Civil Engineers.”
[71] For a “Memoir” of Bodmer see “Transactions of the Institution of Civil
Engineers,” Vol. XXVIII, p. 573. London, 1868.

Bodmer was born at Zurich in 1786. After serving his

apprenticeship he opened a small shop for millwright work near that
city. A year or so later he formed a partnership with Baron d’Eichthal
and with workmen brought from St. Etienne, France, he started a
factory in an old convent at St. Blaise, in the Black Forest, first for
the manufacture of textile machinery and later, in 1806, of small
arms.
“Instead of confining himself to the ordinary process of gun-making
by manual labour, Mr. Bodmer invented and successfully applied a
series of special machines by which the various parts—more
especially those of the lock—were shaped and prepared for
immediate use, so as to insure perfect uniformity and to economise
labour. Amongst these machines there was also a planing machine
on a small scale; and Mr. Bodmer has been heard to observe how
strange it was that it should not have occurred to him to produce a
larger machine of the same kind, with a view to its use for general
purposes.”[72] He does not seem to have used the process of milling
until much later. Bodmer was thus among the first to discern and to
realize many of the possibilities of interchangeable manufacture, Eli
Whitney having begun the manufacture of firearms on the
interchangeable basis at New Haven, Conn., about 1800, only a few
years before. Why Bodmer’s attempt should have failed of the
influence which Whitney’s had is not quite clear. A possible
explanation may lie in the fact that the use of limit gauges does not
seem to have been a part of Bodmer’s plan. This use was
recognized by the American gun makers as an essential element in
the interchangeable system almost from the start.
[72] Ibid., p. 576. (The italics are ours.)

Bodmer was appointed, by the Grand Duke of Baden, director of

the iron works and military inspector with the rank of captain and for
a number of years much of his energy was given to the development
of small arms and field artillery. He invented and built a 12-pound
breech-loading cannon in 1814, which he had tested by the French
artillery officers. It failed to satisfy them, and was sent a few years
later to England, where it was decently buried by the Board of
Ordnance.
The following year he built a flour-mill at Zurich for his brother.
Instead of each set of stones being driven by a small waterwheel, all
the machinery connected with the mill was driven by a single large
wheel through mill gearing. The millstones were arranged in groups
of four. “Each set could be started and stopped separately, and was
besides furnished with a contrivance for accurately adjusting the
distance between the top and bottom stones. By means of a hoist of
simple construction, consisting in fact only of a large and broad-
flanged strap-pulley and a rope-drum, both mounted on the same
spindle (the latter being hinged at one end, so that it could be raised
and lowered by means of a rope), the sacks of grain or flour could be
made to ascend and to descend at pleasure, and the operatives
themselves could pass from one floor to any other by simply
tightening and releasing the rope.[73] The shafting of this mill was
made of wrought iron, and the wheels, pulleys, hangers, pedestals,
frames, &c., of cast iron, much in accordance with modern
practice.”[74] This was several years before Fairbairn and Lillie began
their improvements at Manchester.
[73] Apparently the modern belt conveyor.
[74] “Memoir,” p. 579.

Bodmer went to England for the first time in 1816 and visited all
the principal machine shops, textile mills and iron works. He returned
in 1824 and again in 1833, this time remaining many years. On his
second trip he established a small factory for the manufacture of
textile machinery at Bolton, in which was one of the first, if not the
first, traveling crane.[75] At the beginning of his last and long
residence in England, Bodmer appointed Sharp, Roberts &
Company makers of his improved cotton machinery, which they also
undertook to recommend and introduce. This arrangement was not
successful, and a few years later, in partnership with Mr. H. H. Birley,
Bodmer started a machine shop and foundry in Manchester for
building machinery.
[75] Ibid., p. 581.

Nearly all of the machinery for the Manchester plant was designed
and built by Bodmer himself and it forms the subject of two
remarkable patents, granted, one in 1839 and the other in 1841.[76]
The two patents cover in reality nearly forty distinct inventions in
machinery and tools “for cutting, planing, turning, drilling, and rolling
metal,” and “screwing stocks, taps and dies, and certain other tools.”
“Gradually, nearly the whole of these tools were actually constructed
and set to work. The small lathes, the large lathes, and the planing,
drilling, and slotting machines were systematically arranged in rows,
according to a carefully-prepared plan; the large lathes being
provided, overhead, with small traveling cranes, fitted with pulley-
blocks, for the purpose of enabling the workmen more economically
and conveniently to set the articles to be operated upon in the lathes,
and to remove them after being finished. Small cranes were also
erected in sufficient numbers within easy reach of the planing
machines, &c., besides which several lines of rails traversed the
shop from end to end for the easy conveyance on trucks of the parts
of machinery to be operated upon.”[77] There were, in addition to
these, however, “a large radial boring machine and a wheel-cutting
machine capable of taking in wheels of 15 feet in diameter, and of
splendid workmanship, especially in regard to the dividing wheel,
and a number of useful break or gap-lathes, were also constructed
and used with advantage. It is especially necessary to mention a
number of small, 6-inch, screwing lathes, which, by means of a
treadle acting upon the driving gear overhead, and a double slide-
rest—one of the tools moving into cut as the other was withdrawn,—
screw cutting could uninterruptedly proceed both in the forward and
in the backward motion of the toolslide, and therefore a given
amount of work accomplished in half the time which it would occupy
by the use of the ordinary means. Some of the slide-lathes were also
arranged for taking simultaneously a roughing and finishing cut.”[78]
[76] The first of these is described in the American Machinist of March
13, 1902, p. 369.
[77] “Memoir,” p. 588.
[78] Ibid., p. 597-598.

The latter part of Bodmer’s life was spent in and near Vienna,
working on engines and boilers, beet sugar machinery and
ordnance; and at Zurich, where he died in 1864, in his seventy-ninth
year.
Bodmer does not seem to have originated any new types of
machine tools, with the exception of the vertical boring-mill, which he
clearly describes, terming it a “circular planer.” It was little used in
England, and has been considered an American development.
It is hard now to determine how far Bodmer has influenced tool
design. It was much, anyway. Speaking of the patent just referred to,
John Richards, who has himself done so much for tool design, says,
“Here was the beginning of the practice that endured.” He has
described some of Bodmer’s tools in a series of articles which show
a standard of design greatly in advance of the practice of his time.[79]
Another writer says of Bodmer, “He seems always to have
thoroughly understood the problems he undertook to solve.” “One is
lost in admiration at the versatility of the inventive genius which could
at any one time—and that so early in the history of machine design
—evolve such excellent conceptions of what was needed in so many
branches of the mechanics’ art.”[80]
[79] American Machinist, Vol. XXII, pp. 352, 379, 402, 430, 457, 478,
507, 531, 559, 586, 607, 637.
[80] Ibid., Vol. XXV, p. 369.

Bodmer was elected a member of the Institution of Civil Engineers

in 1835, and his standing among his contemporaries is shown by the
fact that thirty-five pages in the “Transactions” of the Institution for
1868 are given to his memoir. For a foreigner to have won respect
and distinction in the fields of textile machinery, machine tools and
steam engines in England, where all three originated, was surely
“carrying coals to Newcastle.” Not only did he succeed in these
fields, but he invented the traveling crane, the chain grate for boilers,
the Meyer type of cut-off valve gear, the rolling of locomotive tires,
and introduced the system of diametral pitch, which was long known
as the “Manchester pitch,” from its having originated in his plant at
Manchester.
Though Bodmer was never regularly engaged in the building of
machine tools, his contribution to that field is far too great to be
forgotten.
 CHAPTER VIII
JAMES NASMYTH
We know more of the life of Nasmyth than of any of the other tool
builders. Not only did Smiles give an account of him in “Industrial
Biography,”[81] but fortunately Nasmyth was induced in later life to
write his recollections, which were published in the form of an
autobiography, edited by Smiles.[82] With the exception of Sir William
Fairbairn, he is the only great engineer who has done this. His
intimate knowledge of the rise of tool building, the distinguished part
he himself had in it, and his keen and generous appreciation of
others, make his record valuable. We have already quoted him in
connection with Maudslay, and wherever possible will let him tell his
own story.
[81] “Industrial Biography,” Chap. XV. Boston, 1864.
[82] “James Nasmyth, Engineer, An Autobiography,” edited by Samuel
Smiles. London, 1883.

Unlike most of the early mechanics, James Nasmyth came from a

family of distinction dating from the thirteenth century. They lost their
property in the wars of the Covenanters and his direct ancestors took
refuge in Edinburgh, leaving their impress on the city as the
architects and builders of many of its most famous and beautiful
buildings. Alexander Nasmyth, the father of James, was a well-
known artist, the founder of the Scotch School of Landscape
Painting, and a friend of Burns, Raeburn and Sir Walter Scott. He
was a landscape architect and enough of an engineer to be included
in Walker’s engraving of “The Eminent Men of Science Living in
1807-1808,” reproduced in Fig. 8. He invented the “bow-string” truss
in 1794, the first one of which was erected over a deep ravine in the
island of St. Helena, and also the setting of rivets by pressure
instead of hammering. This last, by the way, was the result of trying
to do a surreptitious job on Sunday without outraging the fearsome
Scotch “Sawbath.” Alexander Nasmyth was one of the six men on
the first trip made on Dalswinton Loch, October 14, 1788, by the
steamboat built by Symington for Patrick Miller. This was the second
trip of a steam-propelled vessel, the first one being that of John Fitch
on the Delaware, August 22, 1787. It was an iron boat with double
hulls and made about five miles an hour. It barely escaped being the
first iron vessel, as Wilkinson’s iron boat on the Severn was
launched less than a year before. The picture of this trial trip which
has come down to us was made by Alexander Nasmyth at the
time.[83]
[83] Ibid., pp. 28-31.

James Nasmyth was born in 1808, the tenth in a family of eleven

children. Like all of his brothers and sisters, he inherited his father’s
artistic tastes. If he had not been an engineer he would probably
have become distinguished as an artist. He was ambidextrous, and
to the end of his life his skill with his pencil was a constant source of
pleasure and convenience. The notebook in which the later record of
his mathematical ideas is contained, is crowded with funny little
sketches, landscapes, little devils and whimsical figures running in
and out among the calculations. The leaf in this book on which he
made his first memorandum of the steam hammer is shown in Fig.
23. In 1817, Watt, then in his eighty-first year, visited Edinburgh and
was entertained at the Earl of Buchan’s, where Alexander Nasmyth
met him at dinner. Watt delighted all with his kindly talk, and
astonished them with the extent and profundity of his information.
The following day Watt visited Nasmyth to examine his artistic and
other works. James Nasmyth, a nine-year-old boy, returning from
school, met him at the doorstep as he was leaving, and never forgot
the tall, bent figure of “the Great Engineer.”
 Figure 22. James Nasmyth
from an etching by paul rajon
 Nasmyth’s father had a private workshop which was well equipped
for those days. Nasmyth played there from childhood and had
mastered the use of all the tools while still a schoolboy. “By means of
my father’s excellent foot lathe,” he says, “I turned out spinning tops
in capital style, so much so that I became quite noted amongst my
school companions. They would give any price for them. The peeries
were turned with perfect accuracy, and the steel shod, or spinning
pivot, was centered so as to correspond with the heaviest diameter
at the top. They could spin twice as long as the bought peeries.
When at full speed they would ‘sleep,’ that is, turn round without a
particle of waving. This was considered high art as regarded top-
spinning.”[84] He established a brisk business in these, in small brass
cannon, and especially in large cellar keys, which he converted into
a sort of hand cannon, with a small touch-hole bored into the barrel
and a sliding brass collar which allowed them to be loaded, primed,
and then carried around in the pocket.
[84] Ibid., p. 89.

He haunted all the shops and foundries in the neighborhood,

making friends with the skilled workmen and absorbing the mysteries
of foundry work, forging, hardening and tempering, and those arts
which were handed down from man to man. Speaking of Patterson’s
old shop, Nasmyth says: “To me it was the most instructive school of
practical mechanics. Although I was only about thirteen at the time, I
used to lend a hand, in which hearty zeal made up for want of
strength. I look back on these days, especially to the Saturday
afternoons spent in the workshops of this admirably conducted iron
foundry, as a most important part of my education as a mechanical
engineer. I did not read about such things; for words were of little
use. But I saw and handled, and thus all the ideas in connection with
them became permanently rooted in my mind....
“One of these excellent men, with whom I was frequently brought
into contact, was William Watson. He took special charge of all that
related to the construction and repairs of steam engines,
waterwheels, and millwork generally. He was a skillful designer and
draughtsman and an excellent pattern maker. His designs were
drawn in a bold and distinct style, on large deal boards, and were
passed into the hands of the mechanics to be translated by them into
actual work.”[85]
[85] Ibid., p. 92.

After telling of various workmen, Nasmyth says: “One of the most

original characters about the foundry, however, was Johnie Syme.
He took charge of the old Boulton & Watt steam engine, which gave
motion to the machinery of the works.... Johnie was a complete
incarnation of technical knowledge. He was the Jack-of-all-trades of
the establishment; and the standing counsel in every out-of-the-way
case of managing and overcoming mechanical difficulties. He was
the superintendent of the boring machines. In those days the boring
of a steam engine cylinder was considered high art in excelsis!
Patterson’s firm was celebrated for the accuracy of its boring.
“I owe Johnie Syme a special debt of gratitude, as it was he who
first initiated me into that most important of all technical processes in
practical mechanism—the art of hardening and tempering steel.”[86]
From another of his friends, Tom Smith, Nasmyth picked up the
rudiments of practical chemistry, as it was then understood.
[86] Ibid., p. 93.

Traveling with his father from time to time, he had good

opportunities for meeting many distinguished engineers and of
visiting the great iron works, the most famous of which was the
Carron Iron Works. “The Carron Iron Works,” he writes, “are classic
ground to engineers. They are associated with the memory of
Roebuck, Watt, and Miller of Dalswinton. For there, Roebuck and
Watt began the first working steam engine; Miller applied the steam
engine to the purposes of navigation, and invented the Carronade
gun. The works existed at an early period in the history of British iron
manufacture. Much of the machinery continued to be of wood.
Although effective in a general way it was monstrously cumbrous. It
gave the idea of vast power and capability of resistance, while it was
far from being so in reality. It was, however, truly imposing and
impressive in its effect upon strangers. When seen partially lit up by
the glowing masses of white-hot iron, with only the rays of bright
sunshine gleaming through the holes in the roof, and the dark, black,
smoky vaults in which the cumbrous machinery was heard rumbling
away in the distance—while the moving parts were dimly seen
through the murky atmosphere, mixed with the sounds of escaping
steam and rushes of water; with the half-naked men darting about
with masses of red-hot iron and ladles full of molten cast-iron—it
made a powerful impression upon the mind.”[87]
[87] Ibid., p. 109.

By the time he was seventeen Nasmyth had become a skilled

model maker. While he was still attending lectures in the Edinburgh
School of Arts and in the University, he had built up quite a brisk
business in engine models, for which he charged £10 each. He
made his brass castings in his own bedroom at night, arranging a
furnace in his grate. He had a secret box of moulding sand and
rammed his patterns gently so as not to awaken his father who slept
below. In the morning the room would be all clean and gave no
indication that it was serving for a foundry as well as a bedroom, and
by some miracle he managed to complete his practical education
without burning down the house. In 1827, when he was nineteen, he
built a steam road carriage which ran about the streets of Edinburgh
for many months, but the condition of the Scotch roads was such as
to make a machine of this kind almost useless. When he went to
London he broke it up, and sold the engine and boiler for £67.
From inspecting the engines constructed by different makers,
Nasmyth became impressed with the superiority of those turned out
by the Carmichaels of Dundee. “I afterwards found,” he writes, “that
the Carmichaels were among the first of the Scottish engine makers
who gave due attention to the employment of improved mechanical
tools, with the object of producing accurate work with greater ease,
rapidity, and economy, than could possibly be effected by the hand
labor of even the most skillful workmen. I was told that the cause of
the excellence of the Carmichaels’ work was not only in the ability of
the heads of the firm, but in their employment of the best engineers’
tools. Some of their leading men had worked at Maudslay’s machine
shop in London, the fame of which had already reached Dundee,
and Maudslay’s system of employing machine tools had been
imported into the northern steam factory.”[88] These reports built up
an ambition, which developed into a passion, to go to London and
work in Maudslay’s shop under “this greatest of mechanics.”
[88] Ibid., p. 123.

Consequently, in the spring of 1829, he went with his father to

London and made application to Maudslay to work with him as an
apprentice. Maudslay told them in the friendliest way, but
unmistakeably, that he had had no satisfaction from gentleman
apprentices and that he had definitely settled that he would never
employ one again. He showed them about his shop, however, and
began to melt when he saw the boy’s keen interest and intelligent
appreciation of everything about him. Nasmyth had brought with him
some of his drawings and one of his engine models. At the end of
the visit he mustered courage to ask Maudslay if he would look at
them. The next day Maudslay and his partner looked them over. “I
waited anxiously. Twenty long minutes passed. At last he entered the
room, and from a lively expression in his countenance I observed in
a moment that the great object of my long cherished ambition had
been attained! He expressed, in good round terms, his satisfaction at
my practical ability as a workman engineer and mechanical
draughtsman. Then, opening the door which led from his library into
his beautiful private workshop, he said, ‘This is where I wish you to
work, beside me, as my assistant workman. From what I have seen,
there is no need of an apprenticeship in your case.’[89]
[89] Ibid., p. 129.

“Mr. Maudslay seemed at once to take me into his confidence. He

treated me in the most kindly manner—not as a workman or an
apprentice, but as a friend. I was an anxious listener to everything
that he said; and it gave him pleasure to observe that I understood
and valued his conversation. The greatest treat of all was in store for
me. He showed me his exquisite collection of taps and dies and
screw-tackle, which he had made with the utmost care for his own
service. They rested in a succession of drawers near to the bench
where he worked....
“He proceeded to dilate upon the importance of the uniformity of
screws. Some may call it an improvement, but it might almost be
called a revolution in mechanical engineering which Mr. Maudslay
introduced. Before his time no system had been followed in
proportioning the number of threads of screws to their diameter.
Every bolt and nut was thus a specialty in itself, and neither
possessed nor admitted of any community with its neighbors. To
such an extent had this practice been carried that all bolts and their
corresponding nuts had to be specially marked as belonging to each
other....
“None but those who lived in the comparatively early days of
machine manufacture can form an adequate idea of the annoyance,
delay, and cost of this utter want of system, or can appreciate the
vast services rendered to mechanical engineering by Mr. Maudslay,
who was the first to introduce the practical measures necessary for
its remedy.”[90]
[90] Ibid., pp. 131-132.

There was no place in all England where Nasmyth could have

learned more. He was in close personal contact with one of the best
mechanics in the world. He had Maudslay’s warmest personal
interest and heard all the discussions of the engineers and famous
men who used to come to the workshop. “Among Mr. Maudslay’s
most frequent visitors was Gen. Sir Samuel Bentham, Mr. Barton,
director of the Royal Mint, Mr. Bryan Donkin, Mr. Faraday, and Mr.
Chantrey, the sculptor. As Mr. Maudslay wished me to be at hand to
give him any necessary assistance, I had the opportunity of listening
to the conversation between him and these distinguished visitors. Sir
Samuel Bentham called very often. He had been associated with
Maudslay during the contrivance and construction of the block
machinery. He was brother of the celebrated Jeremy Bentham, and
he applied the same clear common sense to mechanical subjects
which the other had done to legal, social and political questions.
“It was in the highest degree interesting and instructive to hear
these two great pioneers in the history and application of mechanics
discussing the events connected with the block-making machinery.
In fact, Maudslay’s connection with the subject had led to the
development of most of our modern engineering tools. They may
since have been somewhat altered in arrangement, but not in
principle. Scarcely a week passed without a visit from the General.
He sat in the beautiful workshop, where he always seemed so
happy. It was a great treat to hear him and Maudslay fight their
battles over again, in recounting the difficulties, both official and
mechanical, over which they had so gloriously triumphed.”[91]
[91] Ibid., pp. 151-152.

While with Maudslay, Nasmyth designed and built an index milling

machine for finishing the sides of hexagon nuts. After Maudslay’s
death in 1831, he remained a few months with Mr. Field to finish
some work in hand, and then left to start in business for himself.
Nasmyth speaks in the kindliest terms of Mr. Field, and doubtless
would have had more to say about him if his relationship with
Maudslay had not been so close.
Joshua Field was a man to be appreciated. He was a draftsman at
the Portsmouth dockyard when the block machinery was being built,
and showed so clear a grasp of the work in hand that Bentham had
him transferred to the Admiralty at Whitehall. In 1804 he left the
service and went to Maudslay’s, when he was at Margaret Street
and employed about eighty men. He rose steadily, was taken into
partnership in 1822, at the same time as Maudslay’s eldest son, and
was the senior partner after Maudslay’s death when the firm was at
the height of its long prosperity. He was one of those consulted in the
laying of the Atlantic cable and in the designing of machinery for
doing it.
“Mr. Field was one of the founders of the Institution of Civil
Engineers, the origin of which was very humble. About the year
1816, Mr. Henry Robinson Palmer, who was then a pupil of the late
Mr. Bryan Donkin, suggested to Mr. Field the idea of forming a
society of young engineers, for their mutual improvement in
mechanical and engineering science; and the earliest members were
Mr. Henry Robinson Palmer, Mr. William Nicholson Maudslay, and
Mr. Joshua Field. To these three were shortly added Mr. James
Jones, Mr. Charles Collinge, and Mr. James Ashwell. They met
occasionally in a room hired for the purpose, and to them were soon
attracted others having the same objects in view. Mr. Field was the
first chairman of the Institution, being elected to that post on the sixth
of January, 1818. Subsequently he became, in 1837, a vice-
president, an office he filled until he was elected president in 1848,
and in 1849, and he continued to the last to be an active member
and warm supporter of the Institution.”[92] Mr. Field did everything in
his power to give Nasmyth a start, allowing him to make the castings
for some machine tools which he proposed to finish later for use in
his own plant.
[92] Memoir, in “Transactions of the Institution of Civil Engineers,” Vol.
XXIII, p. 491. 1863.

Nasmyth returned to Edinburgh and took temporary quarters in a

little outbuilding 16 feet by 24 feet, within a few minutes’ walk of his
father’s home. He hired one mechanic, Archie Torry, who remained
with him the rest of his life and became one of his principal foremen.
His power plant consisted of one husky laborer who turned a crank.
Together they finished up the castings brought from Maudslay &
Field’s, making first a lathe, then a planer 20 inches by 36 inches,
and with these a few boring and drilling machines. He carried the
expense of this by doing some work for an enthusiastic inventor of a
wonderful rotary steam engine. Nasmyth honorably informed the
inventor that his machine would not work, but as the inventor was
bent on spending his money, Nasmyth executed the work for him,
and the proceeds enabled him to build his machinery.
In a few months he was ready to begin. He went to Liverpool and
Manchester looking for a location, and soon made many powerful
friends in both cities. In 1831 he rented a single floor in Manchester,
27 feet by 130 feet, with power, and ten days later Archie followed
with the tools. It was a particularly fortunate time and place for
starting such an enterprise. The success of the Liverpool &
Manchester Railway, just opened, created a great demand for
locomotives and for machine tools. Orders came in fast, and the
planer especially was busy all the time. If its profits were anything
like those of Clement’s planer, it must have been a very heavy
earner. As the business grew, Nasmyth added more tools, always
making them himself and steadily improving their design and
construction.
 He soon outgrew his quarters; and in 1836 he secured land at
Patricroft, a mile or so outside of the city, admirably located between
the new railway and the Bridgewater Canal, and built a new plant
which he called the Bridgewater Foundry. In the new foundry he
used the first worm-geared tilting pouring-ladle. As it eliminated a
common and very dangerous source of accidents, he refrained from
patenting it and in a short time its use was universal. He formed a
partnership with Holbrook Gaskell, who took the business end of the
enterprise, and the firm of Nasmyth & Gaskell had a very prosperous
career until, sixteen years later, Mr. Gaskell was forced to retire on
account of ill health.
Nasmyth built machine tools of all kinds. In 1836 he invented the
shaper which was long known as “Nasmyth’s Steel Arm.”
Descriptions and illustrations of some of Nasmyth’s tools may be
found at the end of his autobiography,[93] in Buchanan’s “Mill
Work,”[94] and in the American Machinist.[95] He patented but few of
his inventions, relying for protection mainly upon the reputation
which he soon established. “In mechanical structures and
contrivances,” he says, “I have always endeavored to attain the
desired purpose by the employment of the fewest parts, casting
aside every detail not absolutely necessary, and guarding carefully
against the intrusion of mere traditional forms and arrangements.
The latter are apt to insinuate themselves, and to interfere with that
simplicity and directness of action which is in all cases so desirable a
quality in mechanical structures. Plain common sense should be
apparent in the general design, as in the form and arrangement of
the details; and a character of severe utility pervade the whole,
accompanied with as much attention to gracefulness of form as is
consistent with the nature and purpose of the structure.”[96] This was
written in later life. While his later work was in thorough conformity
with these principles, it was some time before he freed himself from
the tradition of Greek style in machine frames. He was one of those,
however, who led the way into the more correct practice indicated
above, though he was probably not so influential in this direction as
Whitworth.
[93] p. 400 et seq.
[94] Volume of Plates.
 [95] Oct. 14, 1909, p. 654.
[96] Autobiography, p. 439.

His greatest invention unquestionably was that of the steam

hammer, which came about in an interesting way. He had built a
number of locomotives for the Great Western Railway. This railway
operated a line of steamers from Bristol to New York and was
planning a ship larger and faster than any then built, to be called
“The Great Britain.” It was to be a side-wheeler and the plans called
for a large and heavy paddle shaft, 30 inches in diameter. Mr.
Humphries, its designer, wrote to Nasmyth asking for help, saying so
large a shaft could not be forged with any of the hammers then in
use. Nasmyth saw at once the limitations of the prevailing tilt
hammer—which was simply a smith’s hand hammer, enlarged, with
a range so small that it “gagged” on large work,—and that the design
of large hammers must be approached in an entirely new way. “The
obvious remedy was to contrive some method by which a ponderous
block of iron should be lifted to a sufficient height above the object
on which it was desired to strike a blow, and then to let the block fall
down upon the forging, guiding it in its descent by such simple
means as should give the required precision in the percussive action
of the falling mass. Following up this idea,” he writes, “I got out my
‘Scheme Book,’ on the pages of which I generally thought out, with
the aid of pen and pencil, such mechanical adaptations as I had
conceived in my mind, and was thereby enabled to render them
visible. I then rapidly sketched out my steam hammer, having it all
clearly before me in mind’s eye. In little more than half an hour after
receiving Mr. Humphries’s letter narrating his unlooked-for difficulty, I
had the whole contrivance, in all its executant details, before me in a
page of my Scheme Book, a reduced photograph copy of which I
append to this description. (See Fig. 23.) The date of this first
drawing was the twenty-fourth of November, 1839....”[97]
[97] Ibid., p. 240.
Figure 23. First Sketch of the Steam Hammer Nov. 24, 1839
Figure 24. Model of the First Steam Hammer
 In the South Kensington Museum, London

“Rude and rapidly sketched out as it was, this, my first delineation

of the steam hammer, will be found to comprise all the essential
elements of the invention.[98] Every detail of the drawing retains to
this day the form and arrangement which I gave to it forty-three
years ago. I believed that the steam hammer would prove practically
successful; and I looked forward to its general employment in the
forging of heavy masses of iron. It is no small gratification to me now,
when I look over my rude and hasty first sketch, to find that I hit the
mark so exactly, not only in the general structure but in the details;
and that the invention as I then conceived it and put it into shape, still
retains its form and arrangements intact in the thousands of steam
hammers that are now doing good service in the mechanical arts
throughout the civilized world.”[99]
[98] Compare Nasmyth’s sketch, Fig. 23, with Fig. 24, which was taken
from the model of his first hammer now in the South Kensington Museum
(Exhibit No. 1571). The description of it in the catalog is as follows:
“It consists of a base plate with a large central opening through which
projects the top of the anvil, so that a blow on the anvil is not transmitted to
the base plate. On the plate are secured two standards which form guides
for the hammer-head or tup, and also support an overhead cylinder, the
piston of which is connected with the tup by a piston rod passing through
the bottom of the cylinder. Steam is admitted to this cylinder by a stop valve
in the form of a slide, and then by a slide valve on the front of the cylinder,
which by a hand lever can be moved so as to let steam in below the piston
and so raise the heavy tup. When it is lifted to a height proportionate to the
energy of the blow required, the steam is by the slide valve permitted to
escape and the hammer falls upon the forging placed on the anvil. The
cylinder is therefore only single-acting, but the top is closed, and a ring of
holes communicating with the exhaust pipe is provided at a little distance
down inside. In this way an air cushion is formed which helps to start the
piston downwards when a long stroke is being taken, and also the steam
below the piston is permitted to escape when the tup has been lifted as high
as it can safely go. Soon after its invention the steam hammer was greatly
increased in power by accelerating the fall of the tup by admitting steam
above the piston in the downstroke and so changing it into the usual
double-acting steam hammer.” Cat. Machinery Collection, Part II, p. 255.
[99] Autobiography, p. 242.