Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Geoscience Frontiers 11 (2020) 739–744

H O S T E D BY Contents lists available at ScienceDirect

Geoscience Frontiers
journal homepage: www.elsevier.com/locate/gsf

Focus Paper

Artificial intelligence in seismology: Advent, performance and future trends


Pengcheng Jiao a, Amir H. Alavi b, *
a
Ocean College, Zhejiang University, Zhoushan, Zhejiang 316021, China
b
Department of Civil and Environmental Engineering, University of Pittsburgh, Pittsburgh, PA, 15261, USA

A R T I C L E I N F O A B S T R A C T

Handling Editor: M. Santosh Realistically predicting earthquake is critical for seismic risk assessment, prevention and safe design of major
structures. Due to the complex nature of seismic events, it is challengeable to efficiently identify the earthquake
Keywords: response and extract indicative features from the continuously detected seismic data. These challenges severely
Seismology impact the performance of traditional seismic prediction models and obstacle the development of seismology in
Artificial intelligence
general. Taking their advantages in data analysis, artificial intelligence (AI) techniques have been utilized as
Machine learning
powerful statistical tools to tackle these issues. This typically involves processing massive detected data with severe
Deep learning
Internet-of-Things noise to enhance the seismic performance of structures. From extracting meaningful sensing data to unveiling seismic
events that are below the detection level, AI assists in identifying unknown features to more accurately predicting the
earthquake activities. In this focus paper, we provide an overview of the recent AI studies in seismology and evaluate
the performance of the major AI techniques including machine learning and deep learning in seismic data analysis.
Furthermore, we envision the future direction of the AI methods in earthquake engineering which will involve deep
learning-enhanced seismology in an internet-of-things (IoT) platform.

1. Introduction useful information to make reliable predictions and decisions in seis-


mology. Consequently, AI-enhanced seismology significantly relies on
Seismology studies earthquake at the multiscale using enormous how to use AI approaches (e.g., machine learning (ML) or deep learning
measuring data with particular focus on assessing the impact of the natural (DL)) to obtain effective detecting data among noise and detect earth-
disaster on civil infrastructure systems (Beroza, 2018). Fig. 1a demon- quake events that are at or below the noise level, as indicated in Fig. 1b.
strates the characteristics of earthquake and seismic events happened In recent years, seismology has dedicated the majority of its efforts to
during 1900–2013 with respect to the amplitude and depth (USGS, 2019). improving the phase detection sensitivity while addressing massive
Typically, studies in seismology are conducted with four objectives: (1) amount of real-time sensing data. Due to the inefficiency of visually
actions to reduce long-term the disaster of earthquake, (2) disaster prep- examining recording data from different phases by seismic experts, it is of
aration or adjustment, (3) disaster response strategies, and (4) desire to predict earthquake using real-time accumulate data in a rational
post-disaster recovery planning, which are known as the strategies of and reliable manner (Lary et al., 2015). The short (long)-term average
mitigation, preparedness, response, and recovery, respectively (Nabian (STA/LTA) prediction algorithms were developed and later replaced by
and Meidani, 2018). Dissatisfying the ideality of detecting every earth- phase association methods due to the lack of the former in distinguishing
quake event in seismogenic process, the capability and accuracy of impulsive transient seismic signals. The latter was evolved to group
continuously monitoring earthquakes are critically affected by the facts seismic phases together to decide whether earthquake happens (Shahin,
that: (1) many earthquake events are undetected and (2) huge amount of 2016). However, those traditional approaches become inadequate to
noisy data in seismic record (Perol et al., 2018). Recent progresses of the detect small seismic events that occur more frequently. Consequently, the
sensing, processing and analyzing techniques ensure the rapid develop- AI tools are applied to tackle the complicated scenarios in earthquake
ment of seismology, which is particularly resulted in the enhancement of prediction. AI plays an important role in modern seismic detection systems
computational power over massive seismic data. Whereas earlier research as it increases the detection efficiency and reduces the influence of noisy
mainly used traditional data mining methods, more recently, artificial data. Fig. 2 shows an illustration of the interrelation between AI and data
intelligence (AI) provides efficient tools to address those data and extract science techniques in seismology. Comparing different AI algorithms in

* Corresponding author.
E-mail address: alavi@pitt.edu (A.H. Alavi).
Peer-review under responsibility of China University of Geosciences (Beijing).

https://doi.org/10.1016/j.gsf.2019.10.004
Received 19 August 2019; Received in revised form 27 September 2019; Accepted 17 October 2019
Available online 6 November 2019
1674-9871/© 2019 China University of Geosciences (Beijing) and Peking University. Production and hosting by Elsevier B.V. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
P. Jiao, A.H. Alavi Geoscience Frontiers 11 (2020) 739–744

Fig. 1. (a) Characteristics of earthquake and seismic events occurred during 1900–2013 (USGS, 2019), and (b) AI-enhanced seismic analysis in detecting “small”
seismic events and addressing noisy data.

seismology, supervised and unsupervised ML approaches have attracted capturing subtle functional relationships between variables without a
considerable research attention (Sick et al., 2015; Wang et al., 2019; Woo, need to assume prior form of the relationship, complicated information
2019). The ML algorithms (e.g., artificial neural networks (ANN), genetic between factors in earthquake (e.g., time, location, magnitude, and
programming (GP), self-organizing map (SOM), support vector machines possible damages) can be extracted. On the contrary, almost all of the
(SVM), and decision tree (DT)) are used to train to find implicit de- traditional statistical methods can only provide the existing relationships
terminations for seismic events. As one of the cutting-edge algorithms in between seismic data when prior knowledge of the nonlinear nature of the
the ML arena, deep learning uses the ANN concept to learn generalized variables is known. Another advantage of AI over the traditional methods
representations of data sets in different domains to define complex is that it fosters increasing computational efficiency in making decisions
nonlinear relationships between variables. The outstanding capabilities of over earthquake while decreasing error rate (Azamathulla Md, 2013;
ML in sequential data sets lead it to be an optimal choice for the phase Karbasi and Md Azamathulla, 2017). On the other hand, a limitation of the
association problem in seismology, which processes one element of a AI methods is that they are extremely parameter sensitive, especially when
sequence at a time and thus, suitable for phase association in a real-time dealing with complicated experimental datasets. A viable solution to
seismic network. Fig. 2b summarizes the published AI-based studies in tackle this issue is to use robust optimization algorithms (e.g. genetic al-
seismology during 1999–2019. As seen, ANNs are the most widely-used gorithm, particle swarm optimization, Tabu search) for an optimized
ML methods in this domain. Moreover, there has been recently a control of the parameters of the AI method.
growing interest in deploying deep learning techniques in seismology.
2.1. Machine learning
2. The AI approaches in seismic analysis
As a branch of AI, ML involves systems capable of automatically
Attempting to mimic human cognition capability, AI is a computa- learning from data, identifying patterns and making decisions. The
tional solution to address engineering problems that are difficult to solve salient beauty of ML is that it enables computers to learn without being
using conventional approaches (Alavi and Gandomi, 2012). In principal, explicitly programmed. Most of the ML-based methods are essentially
AI uses the given seismic training data to identify the relationship between inspired by biological learning. In seismology, ML uses series of tech-
the inputs and corresponding outputs, as shown in Fig. 3a (Shahin, 2016). niques to find the inherent rules and dependences between data and then
Such essence of AI leads to its advances in seismology, especially consid- classify or regress them. Also, ML is commonly used to categorize and
ering the massive detected seismic data with heavy noise. Due to the in- analyze unseen patterns or features in detected data since it, unlike
adequacy of physics-based models developed using the first principles to seismologists that analyze data using intuition and logics, discovers un-
define the underlying relationships between seismic data, AI has attracted considered features beyond human capability (Kong et al., 2019). Fig. 3b
critical attention in recent years (Li et al., 2018). Taking an AI approach displays the main components of ML, which can be grouped into super-
toward data mining, processing and analyzing substantially increases the vised and unsupervised (Salehi and Burgueno, 2018). The former typi-
accuracy and efficiency of earthquake detection, which offers exciting cally consists of regression and classification methods, and the latter
opportunities to develop seismic networks for multipurpose. Some ex- includes reduction and clustering techniques. There is also another
amples are detection and phase picking, early warning, ground-motion category called semi-supervised learning algorithms that can organize
prediction, tomography, geodesy, etc. Thanks to the capability of AI in the data as well as make predictions. However, characterizing into

Fig. 2. Existing AI and data science studies in seismology. (a) Illustration of the interrelation between AI and data science techniques in seismology. (b) Summary of
the published AI-based studies in seismology during 1999–2019.

740
P. Jiao, A.H. Alavi Geoscience Frontiers 11 (2020) 739–744

supervised learning and unsupervised learning, ML in seismology is


developed using probability theory in five steps, including (1) collecting
and partitioning seismic data for training and testing, (2) preprocessing
to clean, format and remove/recover seismic data, (3) training model
uses numerical optimization algorithms to tune the seismic variables, (4)
evaluating model with respect to the prediction accuracy using the test
data, and (5) generating new data for prediction using an ML algorithm.
In general, ML can be categorized into three applications in seismology
including code accelerator tool to reduce the computational cost of
deterministic models, developing an empirical model if deterministic
model is not possible, and addressing classification problems (Lary et al.,
2015). Within the ML arena, ANNs have found many applications in
seismology (Azamathulla Md et al., 2005; Alavi and Gandomi, 2011a, b).
ANNs together with many other ML methods are considered as black-box
model generators since they are not able to develop practical prediction
equations. This limitation has been tackled by other ML methods such as
GP and Decision Tree (DT). In general, a neural network consists of
several processing components that are typically grouped in layers, e.g.,
an input layer, multi hidden layers, and an output layer (Salehi and
Burgueno, 2018). In seismology, an ANN model starts with the propa-
gation of information at the input layer. The network evaluates and ad-
justs the weights on the presentation of the training dataset and finds a
set of the weights using the learning techniques to produce the re-
lationships between the input and output data with smallest errors
(Giacinto et al., 1997). Through the process of training, the performance
of the ANN model can be validated using independent datasets. The main
steps of the ANN analysis in seismology, as well as the main methods used
in each step, are summarized in Fig. 3c.

3. Perspectives and future trends

3.1. Perspectives – from machine learning to deep learning

Traditional ML models are limited due to the lack of abilities to


address natural data in raw form. Developing an ML model typically
requires designing feature extractors for raw data transformations form.
For example, pixel values of images can be transformed to suitable
feature vectors or internal representations and thus, learning subsystems
can classify and detect patterns in the inputs (Schmidhuber, 2015;
DeVries et al., 2018). DL is a subset of ML that uses multiple layers to
extract higher level features from the raw data progressively. The most
well-known DL architectures are deep neural networks (DNN), con-
volutional neural networks (CNN), deep belief networks (DBN), and
recurrent neural networks (RNN). Essentially based on ANNs, DL consists
of critically deeper layers of neurons, which has been applied to solve
classification and segmentation problems in seismology. DL is a type of
representation learning method that allows to learn from raw data to
automatically discover the representations needed for classification or
detection (LeCun et al., 2015). Representations are transformed into
more abstract levels using simple modules, which are assembled to
structure multiple levels of representation. Taking the advantages of
increasing the amount of available computational data while requiring
little engineering by hand, DL has outperformed over other ML methods
(Nabian and Meidani, 2018). New DL learning architectures and algo-
rithms are likely to assist this progress (Huang et al., 2017). Earthquake
detection and characterization have recently been significantly improved
by the methodological advances in AI. DL has recently been used to
monitor and locate earthquakes, determining the arrival time of P-wave,
first motion polarities and other outputs (Ross et al., 2018, 2019).

3.2. Future trends

Fig. 3. AI in seismology. (a) Principle and objectives of AI systems in seis- Although previous ML-based seismic research has been mainly
mology. (b) Characteristics and (c) main calculation steps of machine learning- focused on deploying supervised learning methods, unsupervised
enabled seismic model. learning methods are likely to play a crucial role in seismology in near
future. The fact is that major learning mechanisms in human and animals

741
P. Jiao, A.H. Alavi Geoscience Frontiers 11 (2020) 739–744

Eventually, it is of our belief that the major progress of the AI-enhanced


seismology is likely to be achieved by combining DL with complex
reasoning. The AI techniques have shown promising performance when
analyzing huge and complicated seismic data and extracting meaningful
findings in seismology. Future developments of the AI-enhanced seismic
analysis could be deployed in the following three aspects:

3.2.1. Data-driven seismic prediction systems


Performance of the AI-based seismic prediction models crucially de-
pends on the detected data. As a consequence, the durability and accuracy
of earthquake monitoring are of significance to the prediction systems.
Other than simply analyzing and monitoring seismic data, data-driven
prediction models aim to cover the techniques of earthquake monitoring
and data transferring as well. The AI-enabled seismic prediction models
Fig. 4. Perspective and future trends of the AI-enhanced seismology from Ma- are required to take into account not only how to extract meaningful results
chine Learning (ML) to Deep Learning (DL). from massive amount of detecting data (i.e., seismic noise), but also how to
ensure sustainable detection over a long time period. The AI techniques
can be deployed to eliminate the influence of seismic noise by setting self-
are unsupervised – discovering the world by observing it, rather than
adjustable limit states. Limit states can be defined to avoid the impact of
being told on all the objects (Gentine et al., 2018). Fig. 4 illustrates the
small-amplitude seismic noise. Smart, data-driven seismic prediction
perspective of the AI-based seismic analysis and envisioned DL as one of
system can adjust the limit states with respect to time period, frequency of
the future trends for AI seismic prediction models.
happened events, detection precision, analyzing cost, etc. On the other
Furthermore, the majority of the AI-related studies in seismology in the
hand, it is important to improve the performance of seismic monitoring
last two decades have been dedicated to using conventional ML ap-
sensors such as using self-powered seismic sensors that harvest energy
proaches such as ANNs, decision trees and SVMs (Bergen et al., 2019). On
from the environment. The AI-enabled seismic prediction models are ex-
the contrary, seismic systems developed using DL methods are still in their
pected to reduce the energy dissipation of seismic sensors and improve the
infancy. Interestingly, the limited reported DL-based studies have already
demonstrated the sustainability and accuracy of addressing detecting data data transformation for detection (Larose et al., 2015). Moreover, it is
important to take into account the influence of location and change
comparing with others. Although DL has been used for earthquake
detection and analysis, new paradigms are still required to replace predication model accordingly. Obviously, an AI model trained by the
seismic data from North America will be different from the model devel-
rule-based manipulation of symbolic expressions by operations on large
oped based on the data from Middle East.
vectors. For example, earthquake events might be linked to glaciers form
and melt in a mountainous region (Masih, 2018). However, lack of studies
3.2.2. Deep learning seismic prediction systems
has been reported on connecting AI-based climate changing models with
DL is recently a cutting-edge AI method in seismology capable of
seismic activities. In the future, AI can be expanded to investigate the
interpreting enormous seismic data and identifying implicit features.
relationship between global climate changes and seismic events.
Different from conventional ML algorithms based on hand-crafted features

Fig. 5. AI in Gartner’s Hype Cycle and vision for the future development of the DL-enhanced seismology in IoT platform.

742
P. Jiao, A.H. Alavi Geoscience Frontiers 11 (2020) 739–744

with relatively high computational cost, DL uses optimal features learned Engineering at the University of Pittsburgh.
by the network to improve the classification accuracy and computation-
ally efficiency. It is expected that DL will play a significant role in the next References
generation of seismic prediction systems. Fig. 5 remarks AI techniques in
Gartner’s Hype Cycle 2018 (Garter’s hype cycle, 2019). It can be seen that Alavi, A.H., Gandomi, A.H., 2011a. Prediction of principal ground-motion parameters
using a hybrid method coupling artificial neural networks and simulated annealing.
DL is at the peak of inflated expectations. Note that DL has been marked at Comput. Struct. 89, 2176–2194.
the peak since the year of 2015, suggesting as 2–5 years prior to the plateau Alavi, A.H., Gandomi, A.H., 2012. Energy-based numerical models for assessment of soil
of productivity (Kotecki, 2018). Given the nature of 5–10 years delay of liquefaction. Geosci. Front. 3 (4), 541–555.
Alavi, A.H., Gandomi, A.H., Modaresnezhad, M., Mousavi, M., 2011b. New ground-
deploying new techniques in engineering, we envision an extensive
motion prediction equations using multi expression programing. J. Earthq. Eng. 15,
application of DL in seismology in the next decade. 511–536.
Alavi, A.H., Jiao, P., Buttlar, W.G., Lajnef, N., 2018. Internet of Things-enabled smart
cities: state-of-the-art and future trends. Measurement 129, 589–606.
3.2.3. Deep learning-enhanced seismology in internet of things platform
Azamathulla Md, H., 2013. A review on application of soft computing methods in water
IoT is defined as an internet-based platform that enables advanced resources engineering. In: Yang, X.S., Gandomi, A.H., Talatahari, S., Alavi, A.H.
services by interconnecting things through interoperable information (Eds.), Metaheuristics in Water, Geotechnical and Transport Engineering. Elsevier
and communication technologies (ICTs). This way, the IoT technology eBook ISBN: 9780123983176.
Azamathulla Md, H., Deo, M.C., Deolalikar, P.B., 2005. Neural networks for estimation of
allows devices to share information using different approaches (e.g., as scour downstream of a ski-jump bucket. ASCE J. Hydraul. Eng. 131 (10), 898–908.
pervasive and ubiquitous computing, embedded devices, or sensor net- Bergen, K.J., Chen, T., Li, Z., 2019. Preface to the focus section on machine learning in
works). It is expected that 50–100 billion devices will be connected to the seismology. Seismol. Res. Lett. 90 (2A), 477–480.
Beroza, G., 2018. Aftershock forecasts turn to AI. Nature 560, 556–557.
Internet via IoT by the end of 2020 (Alavi et al., 2018). In seismology, IoT DeVries, P.M.R., Viegas, F., Wattenberg, M., Meade, B.J., 2018. Deep learning of
refers to the platforms where monitoring devices are equipped with local aftershock patterns following large earthquake. Nature 560, 632–634.
computers and intelligent software to increase the machine-to-machine Garter’s hype cycle, 2019. Hype Cycle for Emerging Technologies, 2018. Gartner.
Available at: https://blogs.gartner.com/smarterwithgartner/files/2018/08/PR_
communications in local monitoring networks. This will eventually 490866_5_Trends_in_the_Emerging_Tech_Hype_Cycle_2018_Hype_Cycle.png.
lead to developing a smart and connected global seismic monitoring Gentine, P., Pritchard, M., Rasp, S., Reinaudi, G., Yacalis, G., 2018. Could machine
network. In this new paradigm, DL-enhanced monitoring devices are be learning break the convection parameterization deadlock? Geophys. Res. Lett. 45,
5742–5751.
used to collect data locally, and cloud computing techniques are applied
Giacinto, G., Paolucci, R., Roli, F., 1997. Application of neural networks and statistical
to transmit and process seismic data in real-time (DeVries et al., 2018). pattern recognition algorithms to earthquake risk evaluation. Pattern Recognit. Lett.
Fig. 5 envisions the future development of the DL-enhanced seismology 18, 1353–1362.
Huang, L., Dong, X., Clee, T.E., 2017. A scalable deep learning platform for identifying
in an IoT platform. Taking the advantages of the rapid development of
geologic features from seismic attributes. Lead. Edge 36 (3), 194–280.
IoT, DL seismic models can readily be functionalized in the IoT platform Karbasi, M., Md Azamathulla, H., 2017. Prediction of scour caused by 2D horizontal jets
to enable an advance real-time seismic monitoring system. using soft computing techniques. Ain Shams Eng. J. 8, 559–570.
Kong, Q., Trugman, D.T., Ross, Z.E., Bianco, M.J., Meade, B.J., Gerstoft, P., 2019. Machine
learning in seismology: turning data into insights. Seismol. Res. Lett. 90 (1), 3–14.
4. Conclusions Kotecki, J., 2018. Deep learning’s ‘permanent peak’ on Gartner’s hype cycle. Machine
Learning in Practice. Available at: https://medium.com/machine-learning-in-practi
Earthquake prediction is of significance to risk assessment, prevention ce/deep-learnings-permanent-peak-on-gartner-s-hype-cycle-96157a1736e.
Larose, E., Carriere, S., Voisin, C., Bottelin, P., Baillet, L., Gueguen, P., Walter, F.,
and safe design of major structures. However, it is typically challengeable Jongmans, D., Guillier, B., Garambois, S., Gimbert, F., Massey, C., 2015.
to characterize earthquake response and unveil features from continu- Environmental seismology: what can we learn on earth surface processes with
ously detected, massive noisy data. To address those severe challenges in ambient noise? J. Appl. Geophys. 166, 62–74.
Lary, D.J., Alavi, A.H., Gandomi, A.H., Walker, A.L., 2015. Machine learning in
seismology, the AI techniques have been used as powerful statistical tools geosciences and remote sensing. Geosci. Front. 7 (1), 3–10.
to address the data-related issues. Along the way, AI has demonstrated LeCun, Y., Bengio, Y., Hinton, G., 2015. Deep learning. Nature 521, 436–444.
advantages for mass adoption, which is emerged into seismology to open a Li, Z., Meier, M., Hauksson, E., Zhan, Z., Andrews, J., 2018. Machine learning seismic
wave discrimination: application to earthquake early warning. Geophys. Res. Lett.
promising direction for AI-enhanced seismic analysis. This is in contrast to 45, 4773–4779.
the traditional approaches dominated in the field of earthquake until now. Masih, A., 2018. An enhanced seismic activity observed due to climate change:
The exciting debut of ML and its robust branches such as DL in the last preliminary results from Alaska. 8th International Conference on Environment
Science and Engineering (ICESE2018) 167 (1), 12018. https://doi.org/10.1088/
decade, combining with the advents of cluster computing environment
1755-1315/167/1/012018.
and more powerful personal computers, immediately offers a potential Nabian, M.A., Meidani, H., 2018. Deep learning for accelerated seismic reliability analysis
solution to the fields requested to address massive seismology data. of transportation networks. Comput. Aided Civ. Infrastruct. Eng. 33, 443–458.
Aiming at developing seismic prediction models that identify seismic Perol, T., Gharbi, M., Denolle, M., 2018. Convolutional neural network for earthquake
detection and location. Sci.Adv. 4, e1700578.
response from noisy data (i.e., effective seismic data) while reveal unseen Ross, Z.E., Meier, M., Hauksson, E., Heaton, T.H., 2018. Generalized seismic phase
patterns and features from detected seismic data (i.e., undetected earth- detection with deep learning. Bull. Seismol. Soc. Am. 180 (5A), 2894–2901.
quake), preliminary efforts have been dedicated to deploying DL to Ross, Z.E., Yue, Y., Meier, M., Hauksson, E., Heaton, T.H., 2019. PhaseLink: a deep
learning approach to seismic phase association. J. Geophys. Res.: Solid Earth 124,
earthquake analysis. Here, we move a step forward to envision the future 856–869.
development trends of the DL-enhanced seismology in IoT platform. Other Salehi, H., Burgueno, R., 2018. Emerging artificial intelligence methods in structural
than the fact that DL seismic analysis is still in its infancy, IoT has just engineering. Eng. Struct. 171, 170–189.
Schmidhuber, J., 2015. Deep learning in neural networks: an overview. Neural Netw. 61,
reached the peak at the Gartner’s hype cycle. Integrating the cutting-edge 85–117.
technologies of DL and IoT techniques and applying them to seismic data Shahin, M.A., 2016. State-of-the-art review of some artificial intelligence applications in
can lead us to the great-leap-forward development of seismology. pile foundations. Geosci. Front. 7, 33–44.
Sick, B., Guggenmos, M., Joswig, M., 2015. Chances and limits of single-station seismic
event clustering by unsupervised pattern recognition. Geophys. J. Int. 201,
Declaration of competing interest 1801–1813.
USGS, 2019. Seismicity of the Earth Maps 1900-2013. US Geological Survey. Available at:
https://earthquake.usgs.gov/earthquakes/byregion/.
The authors declare that they have no known competing financial
Wang, K., Ellsworth, W.L., Beroza, G.C., Williams, G., Zhang, M., Schroeder, D.,
interests or personal relationships that could have appeared to influence Rubinstein, J., 2019. Seismology with dark data: image-based processing of analog
the work reported in this paper. records using machine learning for the Rangely earthquake control experiment.
Seismol. Res. Lett. 90 (2A), 553–562.
Woo, T.H., 2019. Analysis of earthquake management design for nuclear power plants
Acknowledgements (NPPs) incorporated with artificial intelligence (AI) method. Energy Sources, Part A
Recovery, Util. Environ. Eff. 41 (17), 2104–2113.
A.H. Alavi acknowledges the startup fund from the Swanson School of

743
P. Jiao, A.H. Alavi Geoscience Frontiers 11 (2020) 739–744

Pengcheng Jiao earned his MSc and PhD degrees in Civil En- Amir H. Alavi is an Assistant Professor in the Department of
gineering from West Virginia University and Michigan State Civil and Environmental Engineering, and holds a courtesy
University in 2012 and 2017, respectively. Prior to joining the appointment in the Department of Bioengineering at the Uni-
Ocean College at Zhejiang University as a Research Professor in versity of Pittsburgh. Prior to joining the University of Pitts-
2018, Dr. Jiao was working as a Postdoctoral Research Fellow burgh, he was an Assistant Professor of Civil Engineering at the
in the Department of Mechanical Engineering and Applied University of Missouri. Dr. Alavi’s research interests include
Mechanics at University of Pennsylvania, USA. Dr. Jiao’s structural health monitoring, smart civil infrastructure systems,
research interests include advanced metastructures, mechanical deployment of advanced sensors, energy harvesting, and engi-
metamaterials, multiscale structural stability analysis, artificial neering information systems. At the University of Pittsburgh,
intelligence in engineering, structural health monitoring and his Intelligent Structural Monitoring and Response Testing
energy harvesting. Particular interests are focused on: (1) con- (iSMaRT) Lab focuses on advancing the knowledge and tech-
ceptual development and investigation of structural compo- nology required to create self-sustained and multifunctional
nents and systems to expand advanced features; (2) micro/ sensing and monitoring systems that are enhanced by engi-
nanoscale fabrication and testing of mechanical metamaterials; neering system informatics. Dr. Alavi has authored five books
(3) applications of artificial intelligence algorithms in design and over 170 publications in archival journals, book chapters,
and optimization of structures; and (4) deployment of meta- and conference proceedings. He has received a number of
structures for multifunctional applications. award certificates for his journal articles. Dr. Alavi is among the
Google Scholar 200 Most Cited Authors in Civil Engineering, as
well as Web of Science ESI’s Top 1% of Scientific Minds in the
World. Dr. Alavi received his Ph.D. degree in Civil Engineering
from Michigan State University.

744

You might also like