Research on Intelligent Maintenance Mode of Train Running Gears Based on Edge-Cloud Collaborative Computing

Abstract

To solve small coverage, uncomplete detection, low efficiency, low accuracy and high missed and mistaken detection rate of traditional detection against defects on wheelsets, one of the critical moving parts for safe, reliable and comfortable operations of railway vehicles, a domain knowledge-driven based intelligent maintenance and repair mode in the combination of edge/client computing and cloud computing is proposed. Fixed and scattered multi-machine-vision detection is carried out to acquire the images of wheelsets treads in an online and offline manner. After image optimization and processing, feature extraction and integration at the device client, the results are uploaded to the cloud, triggering online distributed optimization, information integration and intelligent perception of scratches, peeling and other major defects of multiple wheelsets of multiple railway vehicles, and providing decision support for active generation of maintenance, repair and remanufacturing strategies. Example of visual inspection for wheelset tread defects shows that the proposed maintenance and repair mode is reasonable and feasible, and that the intelligent machine-vision detection algorithm boasts high detection rate and high accuracy.