Research On Fault Prognostics And Health Management Of Traction Motor For EMU

In recent years,with the rapid development of high-speed railway,the rapid increasing of road networks’ scale,and the expanding of traffic density,year by year,the safety problems have become increasingly prominent.Therefore,it is necessary to change the operation and maintenance mode of EMU,to improve the operation and maintenance efficiency and optimize the repair schedule.As a key component of traction drive system,traction motor holds the lifeline of train safety,so how to accurately evaluate the degradation state and predict the remaining life of traction motor is particularly important.The current operation and maintenance strategy is only a single fault and non-fault classification,which easily leads to "under-maintenance" and "over-maintenance".Moreover,it relies too much on expert experience,and the utilization rate of accumulated data during train operation is low.Fault prediction and health management technology can be applied to the operation and maintenance of EMU,which is also a major problem faced by railway operation and maintenance departments.Therefore the paper adopts a method of degradation state evaluation and remaining life prediction based on hidden semi-Markov model,and optimizes and improves it.By making full use of the value of operation data,mining its rules,and enhancing operation and maintenance decision-making,we can achieve the effects of improving operation and service productiveness.The main research contents are as follows:(1)The PHM system framework of EMU traction motor is proposed.Based on the research of PHM technology and the running characteristics of EMU traction motor,the PHM system architecture of EMU traction motor is put forward,which usually includes three subsystems,namely vehicle PHM system,ground PHM system and vehicle-ground transmission system.(2)Fault evolution mechanism analysis and HSMM modeling of traction motor in emu.The structure,composition,common fault modes,fault causes and fault levels of traction motors are analyzed,so as to determine characteristic parameters,generate fault evolution curves and divide specific degradation states.Finally,there’s proposed by HSMM modeling based on degradation states.(3)Improvement of HSMM model for degradation state assessment and remaining service life prediction.In view of the existing problems of the current model,this paper improves and optimizes from two aspects of feature transformation and model parameters,and uses PSO algorithm and redundant attribute projection to optimize and improve,and introduces their basic theories and algorithms in detail.(4)Evaluation of degradation state and prediction of remaining service life of traction motor based on improved HSMM model.Through dealing with the collected big data,and using data training from whole traction motor’s life,the full life HSMM mode is drew,and testing the problem by using it.In conclusions,this paper studies the key problems of hidden semi-Markov model in the degradation state evaluation and residual life prediction of traction motors.The research results will effectively promote the application of fault prediction and health management technology in EMU operation and maintenance.