A hybrid model of residual dilated convolutional neural network and bidirectional recurrent neural network for the outlier detection of rail profile

Xu Wang; Yanfu Li; Cheng Zhang

doi:10.1117/12.3025528

28 February 2024 A hybrid model of residual dilated convolutional neural network and bidirectional recurrent neural network for the outlier detection of rail profile

Xu Wang, Yanfu Li, Cheng Zhang

Author Affiliations +

Proceedings Volume 13071, International Conference on Mechatronic Engineering and Artificial Intelligence (MEAI 2023); 130713T (2024) https://doi.org/10.1117/12.3025528
Event: International Conference on Mechatronic Engineering and Artificial Intelligence (MEAI 2023), 2023, Shenyang, China

Abstract

The accurate measurement of rail wear is critical in track quality inspection. Usually, the wear is obtained by matching the measured profile with the standard one, and comparing their railhead differences. However, in the complex field environment, the process becomes difficult with lots of outliers mixed in the profile. The outliers not only mislead the location of rail waist, also could cause serious measurement errors. Considering that the global geometry of rail profile is prior and stable, a hybrid model including a residual dilated Convolutional Neural Network (CNN) and a bidirectional Recurrent Neural Network (RNN) is constructed in this paper. The former is used for extracting the local features to detect sparse outliers, and the latter is used for extracting the global geometry to detect dense outlier segments. The efficiency and superiority of the proposed method were verified by numerous experiments. The results show that the average F1-score for outlier detection reaches 99.54%, which outperforms some classical neutral network models obviously. Meanwhile, it omits the registration process, and the real-time performance is also improved obviously.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Xu Wang, Yanfu Li, and Cheng Zhang "A hybrid model of residual dilated convolutional neural network and bidirectional recurrent neural network for the outlier detection of rail profile", Proc. SPIE 13071, International Conference on Mechatronic Engineering and Artificial Intelligence (MEAI 2023), 130713T (28 February 2024); https://doi.org/10.1117/12.3025528

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
8 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Convolution

Feature extraction

Neural networks

Convolutional neural networks

Performance modeling

Binary data

Statistical modeling

Show All Keywords

Keywords/Phrases

Search In:

Publication Years