Research on Object Detection and Combination Clustering for Railway Switch Machine Gap Detection  

作　　者：Qingsheng Feng Shuai Xiao Wangyang Liu Hong Li 

机构地区：[1]School of Automation and Electrical Engineering,Dalian Jiaotong University,Dalian 116028,China [2]School of Software,Dalian Jiaotong University,Dalian 116028,China

出　　处：《Chinese Journal of Electronics》2025年第1期186-203,共18页电子学报(英文版)

基　　金：supported by the Traffic Science and Technology Project of Liaoning Provincial Department of Education(Grant No.202243);the Natural Science Foundation of Liaoning Province(Grant No.2019-ZD0094)。

摘　　要：Turnouts and switch machines play a crucial role in facilitating train line operations and establishing routes,making them vital for ensuring the safety and efficiency of railway transportation.Through the gap detection system of switch machines,the real-time working status of turnouts and switch machines on railway sites can be quickly known.However,due to the challenging working environment and demanding conversion tasks of switch machines,the current gap detection system has often experienced the issues of fault detection.To address this,this study proposes an automatic gap detection method for railway switch machines based on object detection and combination clustering.Firstly,a lightweight object detection network,specifically the MobileNetV3-YOLOv5s model,is used to accurately locate and extract the focal area.Subsequently,the extracted image undergoes preprocessing and is then fed into a combination clustering algorithm to achieve precise segmentation of the gap area and background,the algorithm consists of simple linear iterative clustering,Canopy and kernel fuzzy c-means clustering.Finally,the Fisher optimal segmentation criterion is utilized to divide the data sequence of pixel values,determine the classification nodes and calculate the gap size.The experimental results obtained from switch machine gap images captured in various scenes demonstrate that the proposed method is capable of accurately locating focal areas,efficiently completing gap image segmentation with a segmentation accuracy of 93.55%,and swiftly calculating the gap size with a correct rate of 98.57%.Notably,the method achieves precise detection of gap sizes even after slight deflection of the acquisition camera,aligning it more closely with the actual conditions encountered on railway sites.

关 键 词：Gap detection Railway switch machine Object detection Combination clustering Fisher optimal segmentation 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U284.92[自动化与计算机技术—计算机科学与技术]