Research On Fault Diagnosis Of Shunt Malfunction For ZPW-2000A Track Circuit Based On Improved VMD And Feature Selection

Track circuit is important for railway signaling system and plays an important role in supervising the occupation of the track,checking the integrity of the train and transmitting information to train.However,Shunt malfunction of track circuit is one of the potential safety hazard.And it can directly cause interlock failure,which can lead to serious safety accidents,then cause massive property and life loss.But,due to wide distribution of track circuits,poor maintenance environment,plenty of maintenance tasks,the maintenance of shunt malfunction is not timely and accurately.Based on these situation,realization of intelligent diagnosis of shunt malfunction for track circuit can shorten the time of fault diagnosis and improve the classification accuracy.Previous studies on fault diagnosis of shunt malfunction of track circuit focus on the design of classifier system.This paper concentrates on fault signal's processing and feature selection so as to improve the accuracy of fault diagnosis of shunt malfunction.And this paper mainly completes the following aspects.(1)Based on the two-port network theory and transmission line theory,the equivalent model of ZPW-2000 A jointless track circuit is established.And the correctness of model is verified by the field measured data.By using the model,this paper simulates signal of normal situation and four types of signal when the track circuit is under the situation of shunt malfunction.This paper also analyzes the influence of shunt malfunction on the amplitude envelope of train's induced current.(2)Based on fault diagnosis theory,this paper introduces variational mode decomposition(VMD)as signal processing method.According to the decomposition results of shunt malfunction signal,VMD can not only reduce the interference brought by noise and regular change,but also highlight mutation feature of shunt malfunction signal in the form of singular waveform.Besides,singular waveform happens at the demarcation point of shunt malfunction section.By using signal of shunt malfunction as sample signal and setting different K parameter and ? parameter of VMD,the results show that the K parameter and?parameter have a significant impact on the performance of VMD.(3)Aiming at solving the parameter selection problem of VMD,a parameter optimization method named GWO-VMD based on grey wolf optimization(GWO)algorithm is proposed.And this paper compares GWO-VMD with existing methods such as GA-VMD,PSO-VMD and BA-VMD.The results show that the optimization speed and precision of the GWO algorithm are better.And the signal decomposition effect of the parameter-optimized VMD is better.(4)this paper introduces theory of fault feature type and fault feature selection.Then twelve-dimensional time domain features and four-dimensional energy entropy features are extracted from the signal dealt by GWO-VMD of shunt malfunction.When using Relief_F algorithm to weigh the twelve-dimensional time domain features,the results show that the kurtosis coefficient,the margin factor,the pulse factor,and the amplitude are sensitive to the mutation feature in signal of shunt malfunction.(5)in order to realize fault diagnosis,this paper introduces Support Vector Machine which is widely used on fault diagnosis and designs fault diagnosis model of shunt malfunction of track circuit.Using five types of feature set in the fourth part as the input of SVM,this paper realizes the fault diagnosis of shunt malfunction.It can be seen from the diagnostic accuracy and distance index that the feature set including the four-dimensional selected time domain features and four-dimensional energy entropy features has the best diagnostic effect and the highest quality.Finally,this paper summarizes the research content and look forward to the future research work.