Research On Fault Diagnosis Algorithm For Auxiliary Inverter Of Rail Transit Train

In recent years,rail transit trains have become the first choice for more and more people to travel due to their advantages of green energy saving,safety,comfort,and punctuality.During the operation of rail transit trains,whether the auxiliary inverter can work normally will directly affect the operation safety of the train.Therefore,a detailed study on the fault diagnosis of the auxiliary inverter of rail transit trains has important practical significance for ensuring the safe operation of trains.In this paper,based on the fault diagnosis of the auxiliary inverter of the rail transit train,this paper conducts a fault diagnosis study on the possible open circuit conditions of various IGBTs,and designed a simulation experiment.This article is divided into five chapters.First,it introduces the background of the subject and the research status of fault diagnosis technology.Secondly,study the working principle of the auxiliary inverter of the rail transit train and the cause and type of the fault,and build the auxiliary inverter simulation model in the SIMULINK environment to simulate the 21 modes of the auxiliary inverter IGBT open circuit failure;the relevant programs were written in Matlab,the EMD algorithm and the MEEMD algorithm were compared,and the original fault signal was separately decomposed using these two methods.The MEEMD algorithm with better decomposition accuracy was selected for feature extraction.This method can effectively suppress the modal confusion caused by EMD decomposition;obtain the fault feature vector by calculating the energy moment as the input data of the fault diagnosis algorithm;for the fault feature vector,two methods are used to identify the fault type,one is based on the GRNN neural network method,which has a simple network model structure and few parameters.The diagnostic accuracy rate is 98.8%,which is higher than the BP neural network 77.0% accuracy rate.The other is the least squares support vector machine(LSSVM)algorithm optimized based on the artificial bee colony algorithm(ABC),and compared with the LSSVM standard algorithm without optimization.the diagnostic accuracy rate is 98.88%,which is higher than the simple use of LSSVM algorithm.Diagnostic rate.Finally,the experiment proves that the two algorithms proposed in this paper can indeed improve the accuracy of fault diagnosis.