Reliability Analysis And Fault Prediction Of JTC Compensation Capacitor Based On Operation And Maintenance Data

Compensation capacitor is one of the important equipment of ZPW-2000(UM)type jointless track circuits(JTC)in China,which is used to ensure the effective transmission of track circuit signals on steel rails.The failure of the compensation capacitor will directly affect the normal operation of the track circuit,affect the driving efficiency and even endanger the driving safety,so it is necessary to replace the failed capacitor in time.Compensation capacitors are characterized by a large number and wide distribution,and their failure maintenance requires a lot of manpower and material resources.At present,a large amount of compensation capacitor operation and maintenance data has been accumulated on site.Through the mining of compensation capacitor operation and maintenance data,it is helpful for us to summarize the potential law of compensation capacitor failure,so as to guide the operation and maintenance of compensation capacitor more targeted and save maintenance cost.Therefore,it is of great practical significance and research value to analyze the fault law of compensation capacitor and predict its faults based on operation and maintenance data.The main research contents of the paper are as follows:1.Based on the non-parametric test method,the influencing factors of the compensation capacitor fault are analyzed.Using the non-parametric difference test method of samples,the difference analysis of the compensation capacitor fault samples of different carrier frequency track circuits is carried out to eliminate the influence of the track circuit signal carrier frequency on the compensation capacitor fault.On this basis,the time-space distribution of the fault capacitors are further analyzed,and the time factor and the external environment factor are determined to be some of the main influencing factors for compensating the capacitor fault.2.Based on the Weibull distribution,the reliability of the compensation capacitor is analyzed.First,The trend of compensation capacitor failure over time is analyzed;second,modeling the cumulative failure probability of compensation capacitors based on Weibull distribution,using the method of matching curves,the cumulative failure probability model is first transformed into a linear model,and then using weighted least squares estimation to solve the model parameters,and calculating the Mean Time to Failure(MTTF)of the compensation capacitor;finally,the rationality of the above model is verified.3.Based on K-means clustering,the influence of weather factors on the reliability of the compensation capacitor is analyzed.First,constructing the outdoor working environment characteristics of the compensation capacitor;then,using K-means clustering method to cluster the outdoor working environment of the compensation capacitor to determine the areas with large environmental differences;finally,analyzing the failure of compensation capacitors according to the characteristics of different types of environments.The results show that rainy weather has no significant impact on the service life of the compensation capacitor,an environment with a low outdoor temperature and a large temperature difference between day and night is more likely to adversely affect the service life of the compensation capacitor.4.The number of compensation capacitor faults is predicted based on Long ShortTerm Memory(LSTM).First,based on the compensation capacitor fault records,statistically construct a sequence of the number of compensation capacitor faults with time at a certain interval;then,build a prediction model of the number of faults based on LSTM,and use the Ada Delta algorithm to train model parameters;next,for the problem that the initial hyperparameters of LSTM have a great impact on the performance of the prediction model,the firefly algorithm is used to optimize the hyperparameters,so as to mine the change law of the number of compensation capacitor faults with time,and realize the prediction of the number of failures of the compensation capacitor in a period of time;finally,the experiment verifies that the method can well fit the variation of the number of compensation capacitor faults with time,and can accurately predict the number of compensation capacitor faults.5.The software for predicting the number of compensation capacitor faults is designed and implemented.According to the characteristics of on-site compensation capacitor operation and maintenance data,the software for predicting the number of compensation capacitor faults is designed and implemented,and the software functions are verified.This paper contains 41 figures,7 tables and 70 references.