Inspection of exterior substance on high-speed train bottom based on improved deep learning method

Yao, Zikai; He, Deqiang; Chen, Yanjun; Liu, Bin; Miao, Jianyin; Deng, Jianxin; Shan, Sheng

doi:10.1016/j.measurement.2020.108013

Cited by 35 publications

(20 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These traditional methods are often limited to a certain class of object in the 3D reconstruction of a single image, or it is difficult to generate a 3D object with better precision. With the continuous development of deep learning technology, the technology has been widely used in recent years [6][7][8][9][10][11][12][13][14], such as video analysis [8], image processing [9][10][11], medical diagnosis and service [12,13], and target recognition [14]. Applying these to actual scenarios will encounter problems of large energy consumption and long response time.…”

Section: Introductionmentioning

confidence: 99%

Improved Multiview Decomposition for Single-Image High-Resolution 3D Object Reconstruction

Peng

Wei

et al. 2020

Wireless Communications and Mobile Computing

View full text Add to dashboard Cite

As a representative technology of artificial intelligence, 3D reconstruction based on deep learning can be integrated into the edge computing framework to form an intelligent edge and then realize the intelligent processing of the edge. Recently, high-resolution representation of 3D objects using multiview decomposition (MVD) architecture is a fast reconstruction method for generating objects with realistic details from a single RGB image. The results of high-resolution 3D object reconstruction are related to two aspects. On the one hand, a low-resolution reconstruction network represents a good 3D object from a single RGB image. On the other hand, a high-resolution reconstruction network maximizes fine low-resolution 3D objects. To improve these two aspects and further enhance the high-resolution reconstruction capabilities of the 3D object generation network, we study and improve the low-resolution 3D generation network and the depth map superresolution network. Eventually, we get an improved multiview decomposition (IMVD) network. First, we use a 2D image encoder with multifeature fusion (MFF) to enhance the feature extraction capability of the model. Second, a 3D decoder using an effective subpixel convolutional neural network (3D ESPCN) improves the decoding speed in the decoding stage. Moreover, we design a multiresidual dense block (MRDB) to optimize the depth map superresolution network, which allows the model to capture more object details and reduce the model parameters by approximately 25% when the number of network layers is doubled. The experimental results show that the proposed IMVD is better than the original MVD in the 3D object superresolution experiment and the high-resolution 3D reconstruction experiment of a single image.

show abstract

Section: Introductionmentioning

confidence: 99%

Improved Multiview Decomposition for Single-Image High-Resolution 3D Object Reconstruction

Peng

Wei

et al. 2020

Wireless Communications and Mobile Computing

View full text Add to dashboard Cite

show abstract

“…Deep learning methods enable neural networks that are composed of multiple nonlinear neural modules to solve many complex pattern recognition problems [44]. Multilayer perceptron (MLP) is a kind of feedforward artificial neural network in deep learning.…”

Section: Establishment Of the General Forward Modelmentioning

confidence: 99%

A New High-Precision General Inverse Model Based on Deep Learning for Magnetorheological Damper

2021

View full text Add to dashboard Cite

There are several main factors affect damping characteristics of a magnetorheological damper, the influence laws of some ones are eager to change with varying working conditions. Therefore, it is challengable to establish the relationship between the two: influence factors and damping characteristics of the magnetorheological damper. This paper proposed a new deep learning-based general inverse model to predict the required current value of a magnetorheological damper under changing complex working conditions. Specifically, a fully-connected multilayer perceptron was chosen to train the general forward model. The complex and time-varying relationships between the main factors and damping characteristics were accurately characterized on use of the high nonlinear mapping ability of the multilayer perceptron. Compared to most of general forward models, the proposed model was more excellent with an accuracy of 99.73% based on ensuring generality. Besides, the network inversion method was used in the inverse solution of the multilayer perceptron general forward model, which simplified the process of inverse solution without requiring a mass of training data. Furthermore, a genetic algorithm was used to replace the gradient descent method which was commonly used in network inversion, with the benefit of solving the problem of selecting the initial value and step size, as well as improving the accuracy. Simultaneously, the experiments were conducted on the proposed general inverse model, the experimental results, being consistent with the simulation results, demonstrated that the proposed model can accurately predict the control current value and track the required damping force in changing complex working conditions.INDEX TERMS General forward model, general inverse model, genetic algorithm, multilayer perceptron, network inversion.

show abstract

“…In [14], 2D images are transformed into 1D signals by Gabor filter, and then, the multiple signal classification (MUSIC) algorithm is used to detect the 1D signals, which can classify the signals produced by different track components. Yao et al [15] used an artificial neural network (ANN) and a long short-term memory (LSTM) network to predict the frost heave deformation of a railway subgrade with four sections of data. Wei et al [16] used Dense-SIFT, CNN, and R-CNN to detect defects in fasteners.…”

Section: Introductionmentioning

confidence: 99%

Semi-Supervised Deep Learning in High-Speed Railway Track Detection Based on Distributed Fiber Acoustic Sensing

Wang

Liu

2022

Sensors

Self Cite

View full text Add to dashboard Cite

High deployment costs, safety risks, and time delays restrict traditional track detection methods in high-speed railways. Therefore, approaches based on optical sensors have become the most remarkable strategy in terms of deployment cost and real-time performance. Owing to the large amount of data obtained by sensors, it has been proven that deep learning, as a powerful data-driven approach, can perform effectively in the field of track detection. However, it is difficult and expensive to obtain labeled data from railways during operation. In this study, we used a segment of a high-speed railway track as the experimental object and deployed a distributed optical fiber acoustic system (DAS). We propose a track detection method that innovatively leverages semi-supervised deep learning based on image recognition, with a particular pre-processing for the dataset and a greedy algorithm for the selection of hyper-parameters. The superiority of the method was verified in both experiments and actual applications.

show abstract

Inspection of exterior substance on high-speed train bottom based on improved deep learning method

Cited by 35 publications

References 18 publications

Improved Multiview Decomposition for Single-Image High-Resolution 3D Object Reconstruction

Improved Multiview Decomposition for Single-Image High-Resolution 3D Object Reconstruction

A New High-Precision General Inverse Model Based on Deep Learning for Magnetorheological Damper

Semi-Supervised Deep Learning in High-Speed Railway Track Detection Based on Distributed Fiber Acoustic Sensing

Contact Info

Product

Resources

About