Research On Fault Detection And Diagnosis Method For Running Gears Of High-speed Trains Via Slow Feature Analysis

Running gear systems,regarded as the key system for operation and braking of high-speed trains,are of complex mechanical structure and precise electrical control unit,which bears and transmits various loads from the body and the track.As a complex electromechanical system,the key component of running gear systems will wear and age after long-term operation under complex working conditions,which will affect the running state of trains.Considering the research background of running gear systems,this paper solves the problem of health state assessment and fault detection and diagnosis via a data-driven approach.Based on the slow feature analysis method,the feature extraction and incipient fault detection for running gear systems are studied.Therefore,the main research of this paper focuses on the following two aspects:(1)Aiming at the health status assessment of running gear systems in high-speed trains,an assessment model based on slow feature analysis and support tensor machine is proposed.Firstly,a slow feature analysis model is used to extract the slowest feature in the monitoring data that reflects the general trend of system changes.Secondly,the slowness data is constructed into tensor form,and accurate state evaluation is achieved by support tensor machine model.Finally,based on the actual monitoring data of a running system in high-speed trains,the validity and accuracy of the design scheme are verified.(2)Aiming at the incipient fault detection and diagnosis of running gear systems in high-speed trains,fault detection and diagnosis model based on Hellinger distance and slow feature analysis is proposed.Firstly,a data-driven method named Hellinger distance and slow feature analysis is studied to realize the task of incipient fault detection for running gear systems in high-speed trains.Secondly,the fault diagnosis task is realized with the help of Hidden Markov Model.Finally,the effectiveness of the proposed method is verified by a numerical case and a running gear system in high-speed trains under actual conditions.