Research On Fault Prediction Of Compensation Capacitor For JTC Based On Particle Filter Algorithm

Jointless Track Circuit(JTC)is the key equipment to ensure the safety of the train,which can detect the occupation of the line automatically and continuously,and can also control the signal device and switch device.Compensation capacitor can reduce the influence of inductance,which is an important equipment to ensure the good transmission performance of track circuit.The failure of compensation capacitor may cause the failure of track circuit,which brings great hidden danger to the driving safety and efficiency of train operation.At present,our country adopts the method of regular detection after compensation capacitor failure.With the rapid development of high-speed railway,the timeliness and efficiency of detection are far from meeting the needs.In this thesis,based on the simulation of short-circuit current amplitude curve,the influence of compensation capacitor failure on track circuit is analyzed,and the fault prediction of compensation capacitor is studied on the basis of fault feature extraction.The main work is as follows:First of all,the signal transmission process model of JTC is established based on the transmission line theory,and the calculation formula of receiving voltage and short-circuit current is obtained.After the simulation of short-circuit current amplitude curve,the influence of compensation capacitor failure at different positions on short-circuit current amplitude curve is analyzed and summarized,and the position of compensation capacitor which has the greatest influence on track circuit after the fault is found.The results show that the failure of the second compensation capacitor at the receiving end,the fault of two adjacent capacitors,or the distance between two fault capacitors is three steps has serious impact on JTC.So the maintenance personnel need to pay more attention to these kinds of faults.The analysis process provides a theoretical basis for the field maintenance.Secondly,the short-circuit current amplitude curve in each state is taken as the research object,and Modified Ensemble Empirical Mode Decomposition(MEEMD),Hilbert Transform(HT)and Teager Energy Operators(TEO)are used to analyze fault characteristics and complete feature extraction.The results show that the feature extraction scheme can identify the fault of multiple compensation capacitors breaking at the same time and the fault of single compensation capacitor falling in capacitance,and can extract the number of broken compensation capacitors and the falling degree of capacitance.Finally,the decrease of the capacitance value in feature matrix obtained by feature extraction is taken as the input of fault prediction.Based on the particle filter algorithm,the capacitor degradation model is used to predict the decline degree of compensation capacitance,further the trend extension method is used to predict the remaining service life,and the field data is combined to verify the feasibility and correctness of the algorithm.The results show that the root-mean-square error of the prediction scheme adopted in this thesis is 1.23%,and the accuracy meets the prediction demand.