deeplearning资源-CSDN文库

deep

learning

需积分: 10 143 浏览量 2020-03-20 16:26:52 上传评论收藏 1.88MB PDF 举报

资源推荐

资源详情

资源评论

Smart Structures and Systems, Vol. 24, No. 5 (2019) 567-586

DOI: https://doi.org/10.12989/sss.2019.24.5.567 567

http://www.techno-press.com/journals/sss&subpage=7 ISSN: 1738-1584 (Print), 1738-1991 (Online)

1. Introduction

The structural health monitoring (SHM) of civil

infrastructures mainly aims to monitor the structural

condition, detect the structural damage/abnormality, and

evaluate the structural safety based on the long-term

monitoring data from a variety of sensors installed on the

structure. It is a cutting-edge and multi-disciplinary

technology acting as a powerful tool for improving and

upgrading the level of intelligent maintenance and

management of civil infrastructures (Ni et al. 2010, Ni et al.

2012, Ye et al. 2012, Hakim and Razak 2014). In addition,

the comprehensive understanding of in-service structural

performance and behavior under realistic environmental and

loading conditions will benefit from the long-term

monitoring of a civil engineering structure (Ye et al. 2013,

Ye et al. 2015, Ye et al. 2016a,b,c, Dong et al. 2018).

Although many kinds of SHM systems have been

designed and implemented on many kinds of civil

Corresponding author, Ph.D., Professor

E-mail: cexwye@zju.edu.cn

Ph.D. Candidate

Professor

infrastructures and a huge amount of monitoring data has

been obtained in the past two decades, a big gap still exists

between SHM and structural maintenance and management

(SMM). One of the main reasons is that the current data

processing methods are confronted with challenges from

environmental noise, the volume of measurement data, the

complexity of computation, etc., which severely constrains

the pervasive application of SHM technology (Kesavan et

al. 2005, Matos et al. 2009). Realization of the autonomous,

accurate and robust processing of the monitoring data has

been a great concern of the SHM community (Gao and

Spencer 2007, Jang et al. 2010, Min et al. 2010, Cho et al.

2015, Sony et al. 2019).

With the arrival of the fourth revolution of science and

technology, the technology of artificial intelligence (AI) is

subversively renovating the activities of human life and

social production (Weng et al. 2001). It has been deeply

integrated into the planning, design, construction,

maintenance and management of civil infrastructures (Onat

and Gul 2018, Salehi and Burgueno 2018). In the SHM

community, researchers have devoted efforts to analyzing

and processing the huge amount of monitoring data by the

use of machine learning methods, which are key

components of AI. Extracting and mining the patterns and

rules inherent in the original multi-source heterogeneous

field monitoring data will not only help us accurately and

A review on deep learning-based structural health monitoring

of civil infrastructures

X.W. Ye



, T. Jin

and C.B. Yun

Department of Civil Engineering, Zhejiang University, Hangzhou 310058, China

(Received June 23, 2019, Revised August 25, 2019, Accepted August 30, 2019)

Abstract. In the past two decades, structural health monitoring (SHM) systems have been widely installed on various civil

infrastructures for the tracking of the state of their structural health and the detection of structural damage or abnormality,

through long-term monitoring of environmental conditions as well as structural loadings and responses. In an SHM system,

there are plenty of sensors to acquire a huge number of monitoring data, which can factually reflect the in-service condition of

the target structure. In order to bridge the gap between SHM and structural maintenance and management (SMM), it is

necessary to employ advanced data processing methods to convert the original multi-source heterogeneous field monitoring data

into different types of specific physical indicators in order to make effective decisions regarding inspection, maintenance and

management. Conventional approaches to data analysis are confronted with challenges from environmental noise, the volume of

measurement data, the complexity of computation, etc., and they severely constrain the pervasive application of SHM

technology. In recent years, with the rapid progress of computing hardware and image acquisition equipment, the deep learning-

based data processing approach offers a new channel for excavating the massive data from an SHM system, towards

autonomous, accurate and robust processing of the monitoring data. Many researchers from the SHM community have made

efforts to explore the applications of deep learning-based approaches for structural damage detection and structural condition

assessment. This paper gives a review on the deep learning-based SHM of civil infrastructures with the main content, including

a brief summary of the history of the development of deep learning, the applications of deep learning-based data processing

approaches in the SHM of many kinds of civil infrastructures, and the key challenges and future trends of the strategy of deep

learning-based SHM.

Keywords: structural health monitoring; deep learning; convolutional neural network; structural damage detection;

structural condition assessment; artificial intelligence; machine learning; computer vision

X.W. Ye, T. Jin and C.B. Yun

effectively grasp the structural service condition and the

characteristics of the long-term deterioration of the target

structure, it will also promptly issue warning information as

well as make decisions regarding inspection, repair and

strengthening (Min et al. 2015, Feng and Feng 2018).

The artificial neural network (ANN) algorithm is a

classical machine learning method, and has been applied to

civil engineering since 1989 (Adeli and Yeh 1989). Early

ANNs were perceptrons with one or two hidden layers, and

had a limited capacity for non-linearity abstraction (Wu et

al. 1992, Szewczyk and Hajela 1994, Yun and Bahng 2000).

Meanwhile, the application frameworks were realized based

on the general-purpose computing languages such as

FORTRAN, MATLAB or C language (Adeli 2001, Ni et al.

2002). Later studies employed the hand-crafted algorithms

to extract features from the data and applied ANNs with a

limited number of hidden layers for classification (Ceylan et

al. 2014); while the capacity of autonomous feature

learning from raw data was not available before the training

of a deep neural network (DNN) (Hinton et al. 2006). In

recent years, along with the significant improvement of

network architecture and computing capacity, deep learning

algorithms, e.g., convolutional neural networks (CNNs),

recurrent neural networks (RNNs), etc., have experienced

rapid growth, and have been applied to automatically

process all kinds of data, especially image data (Dong et al.

2016). Many kinds of DNN frameworks and datasets have

been developed to deal with various data processing

scenarios and to satisfy different types of industrial

demands (Vodrahalli and Bhowmik 2017).

Much research has been carried out to explore the

application of deep learning-based approaches in the field

of the SHM of civil infrastructures (Spencer et al. 2019).

This paper aims to address a review on deep learning-based

SHM of civil infrastructures, and is organized as follows:

Section 2 briefly summarizes the history of the development

of deep learning with incidents of milestones. Section 3

presents the applications of deep learning-based approaches

for SHM on various kinds of civil infrastructures. Section 4

discusses the current key challenges and future trends of the

deep learning-based SHM strategy. Section 5 gives some

conclusions of issues dealt with in the paper.

2. A brief history of deep learning research

2.1 Significant contributions to deep learning

Nowadays, deep learning-based approaches have played

an increasingly important role in the field of image

recognition, natural language processing, recommendation

systems, etc., to execute automated, time-saving and low-

cost operations (Schmidhuber 2015, Goodfellow et al.

2016, Silver et al. 2016). Deep learning is a kind of

representational learning method, which enables a network

architecture to autonomously learn highly abstract features

from raw data to fulfill recognition or classification tasks

(Hinton and Salakhutdinov 2006, LeCun et al. 2015). It is a

branch of machine learning, which belongs to a part of AI.

Machine learning is a process of enabling a computer to

Fig. 1 Relationship of AI, machine learning & deep

learning

learn hidden patterns among extracted features and targets

for classification or prediction (Lake et al. 2015). Machine

learning algorithms contain ANNs, support vector machines

(SVMs), random forests, decision trees, Bayesian inference,

etc. (Bishop 2006). AI is a system that is able to

demonstrate the intelligence by machines, similar to but not

the same as the natural intelligence of human beings

(Russell and Norvig 2016, Silver et al. 2017), which

contains computer vision, machine learning, robotics,

speech recognition, expert systems, etc. The relationship

among AI, machine learning and deep learning is shown in

Fig. 1.

The development of deep learning has mainly evolved

from the ANN. The basic element of an ANN was called the

neural cell, and has not been changed much since the first

neural cell model, i.e., the MP model, was proposed in 1943

by McCulloch and Pitts (1943). A neural cell with three

input elements and one output element is shown in Fig. 2.

The input elements, i.e., x

, x

and x

, are multiplied by

weights, i.e., w

, w

and w

, for summation, and a bias, b, is

added for modification. An activation function, f(x),

implements nonlinear transformation to generate an output.

Rosenblatt (1958) proposed a single layer perceptron

structure that consisted of multiple neural cells, which could

learn through perceptron convergence algorithms to

improve the capacity for classification. Rumelhart et al.

(1986) applied a back-propagation algorithm to train multi-

layer neural networks, enabling the hidden layers to

construct useful features for classification.

Fig. 2 The MP model (McCulloch and Pitts 1943)

A review on deep learning-based structural health monitoring of civil infrastructures

LeCun et al. (1989) developed the first deep CNN,

trained by a back-propagation algorithm, to recognize

handwritten zip codes. Later, they proposed LeNet-5 for the

recognition of handwritten characters with over 99.65%

accuracy (LeCun et al. 1998), as shown in Fig. 3. The RNN

is an important DNN for the processing of time-series data.

Hopfield (1982) proposed a network with a circular

structure, which was considered to be the rudiment of the

RNN. Unlike the previous feed-forward neural networks,

the processing of input elements in this network architecture

had backward paths. Elman (1990) proposed a fully-

connected RNN with local memory units and feedback

connections to deal with time-series data. Hochreiter and

Schmidhuber (1997) developed the long short-term memory

network (LSTM) with gate units to solve the problem of

long-term dependence. An LSTM cell has a forgetting gate

and an input gate to filter input data, and an output gate to

generate output data. However, due to the issues of gradient

vanishing or explosion, it is difficult to train a DNN. This

challenge prevented the development of deep learning until

the deep belief network (DBN) was developed by Hinton et

al. (2006). They trained the DBN by unsupervised greedy

training for each layer and then fine-tuning by a supervised

back-propagation algorithm.

Furthermore, the training of a DNN requires the

processing of a massive amount of data with the help of a

great computing power. The improvement of the efficiency

of training is critical to the practical application of a DNN.

To accelerate the training process, Chellapilla et al. (2006)

proposed a graphics processing unit (GPU)-accelerated

convolutional network and produced a 3.1X-4.1X speedup.

Raina et al. (2009) constructed a GPU-aided deep

unsupervised learning network which was simple to

program and needed less time for training. Ciresan et al.

(2010) presented a GPU-accelerated approach to efficiently

train the multi-layer perceptron (MLP). With the progress of

GPU-based training methods, the efficiency of training a

DNN has been drastically improved. However, when the

neural networks become deeper, the number of parameters

grows explosively and this generates the problem of

overfitting.

Krizhevsky et al. (2012) won the 2012 ImageNet

challenge by the proposal of AlexNet with proper treatment

of the overfitting issue. To reduce the overfitting effect,

relu, dropout, and data augmentation were jointly adopted

to train the network architecture with about 60 million

parameters. Also, two GPUs were applied to speed up the

training process of the CNN. The joint application of these

techniques enabled AlexNet to obtain a 15.3% top-5 error

rate in the image classification for 1000 different categories.

Fig. 3 The architecture of LeNet-5 (LeCun et al. 1998)

Fig. 4 Historical development of deep learning

剩余18页未读，继续阅读

评论收藏

内容反馈

weixin_38094471

粉丝: 0
资源: 6

deep learning

deeplearning4j-core-1.0.0-M1.1-API文档-中文版.zip

Deep Learning: Practical Neural Networks with Java 完整高清英文azw3版

MATLAB Deep Learning: With Machine Learning, Neural Networks and A I

《Deep Learning》Yann LeCun, YoshuaBengio, Geoffery Hinton

deeplearning4j-nn-1.0.0-M1.1-API文档-中文版.zip

Deep Learning with Python: A Hands-on Introduction

Better Deep Learning by Jason Brownlee

deep learning(Adaptive Computation and Machine Learning series)

Python Deep Learning: Exploring deep learning techniques, neural network

深度学习(Deep Learning) 三本 Part II 【共 5 个分卷】

Deep Learning in Python

Deep Learning for Beginners

Deep Learning, Vol. 1: From Basics to Practice

Deep Learning 教程中文版

Deep Learning for Time Series Forecasting - by Jason Brownlee

deep learning经典论文30篇

deep learning with theano

SolidWorks-Flexnet-Server.zip

(完整版)ST语言编程手册.pdf

EPLAN部件库、EPLAN宏，近100G

工业触摸屏、组态等图库资源

2018功能安全（ISO26262）中英双语版.zip

FANUC机器人Socket Message通讯设置及使用手册（中文版）.pdf

eplan自带部件库ESS_part001.mdb

西门子S7-1200 SCL编程指令手册.pdf

2020SolidWorks入门教程(很全面).ppt

开关电源明纬部件库和宏.edz

最新资源