Real-time Fault Diagnosis Of Catenary System Based On Pantograph Vibration Performance

The pantograph-catenary system is the main structure for electrified railways to obtain electrical energy from traction substations.The reliability of the pantograph-catenary system limits and determines the maximum operating speed and operational safety of the electrified railway.Therefore,ensuring the safety of the pantograph-catenary system is of great significance to railway transportation.With the rapid development of electrified railway and the increase of running speed and density,catenary system failures occur frequently.To this end,this thesis uses fiber Bragg grating sensor technology to design and develop a real-time condition-based monitoring system for catenary system.Real-time safety state evaluation and fault diagnosis of catenary are realized by monitoring pantograph vibration.The main research contents are as follows:In view of the shortcomings of existing contact force detection methods and the engineering requirements of catenary detection,a simplified beam model of pantograph head is established.A new contact force measuring method was proposed by analyzing the relation of the strain responses under the effect of concentrated force acted on the pantograph head,which achieves the detection of the contact force based on the fiber grating strain sensors without breaking the load transfer path of pantograph head.At the same time,considering the low measurement accuracy caused by the cross-sensitivity of temperature and strain of fiber grating strain sensor,a temperature compensation method based on adaptive noise canceling system was proposed.It based on the correlation between reference temperature and interference temperature and the orthogonality between real strain,taken the Least Mean Squares as the optimization objective,and the filter parameters are adaptively adjusted to eliminate the temperature-related components in the measured strain signal,which greatly improves the measurement accuracy,and verified to be reliable through the line test.In order to study the fault characteristics of catenary,the fault of expansion joint,overlap and support device and compound fault of catenary are simulated based on finite element method.Through the analysis of catenary fault characteristics and fault evolution law,it provides theoretical and data support for the follow-up study of catenary diagnosis methods.Considering the practical difficulties and shortcomings of the existing methods,local outlier factor(LOF)and minimum entropy deconvolution(MED)algorithm are introduced to detect catenary defect.Square envelope and feature fusion are used to optimize MED analysis results,further enhance fault features and improve detection accuracy.Considering the realtime requirement of cateanry detection,K-S test and QQ chart were used to evaluate the normality of strain data,and then a real-time detection method of catenary abnormal state based on sliding window was proposed — MMSTD,which takes pantograph head strain as characteristic parameter and adaptively adjusts evaluation index according to train operation environment and speed.The calculation speed of MMSTD is fast and the detection accuracy is high.In order to select reasonable sliding window parameters,the influence of speed and fault degree on window length is analyzed by fault simulation data.Compared with the catenary detection method based on LOF and MED-feature fusion,the MMSTD has better detection accuracy and real-time performance,which is more in line with the real-time condition monitoring requirements of catenary.According to the difference of catenary fault characterization in strain signals,based on the fluctuation characteristics of strain data,the enhanced sample entropy and catenary structure location coefficient are defined to extract fault features and fault location information effectively.Linear Discriminant Analysis(LDA)is used to project fault features into the optimal discriminant subspace to maximize the separability and identify fault types.By training LDA fault classification model with existing fault samples,intelligent fault diagnosis can be realized.Based on the above research,a real-time monitoring and fault diagnosis system for catenary based on pantograph vibration is established.On this basis,a real-time monitoring device for the pantograph-catenary system was developed.At the same time,the sensor parameters and installation location of the monitoring system were determined based on the actual operating environment of the pantograph-catenary system.The feasibility and reliability of the system in engineering application are verified by the analysis of test data and monitoring results.The engineering of real-time fault diagnosis system for catenary based on Pantograph vibration is basically realized.