Research On Fault Dagnosis Method For Signal Based On VMD And Grey Comprehensive Relational Analysis

Railway signal belongs to the trackside basic equipment on the railway,which is of great significance in ensuring the safe operation of trains and improving the efficiency of railway transportation.In the current railway signal system,the Centralized Signalling Monitoring system(CSM)can collect the signal lighting loop circuit current in real time and dynamically,and send out an alarm in time when it is found that the current value is higher or lower than the preset limit.This paper obtains a large number of signal lighting loop circuit current data from CSM,and summarizes four typical fault curves and one normal current curve,the key problem of this paper is how to correctly diagnose the five signals.In order to solve this problem and realize the intelligent diagnosis of railway signal equipment state,a railway signal fault diagnosis method based on Variational Mode Decomposition(VMD)hybrid feature extraction and grey comprehensive correlation model is proposed in this paper.The main contents are as follows:Firstly,through the analysis of the current acquisition principle of railway signal lighting circuit and signal structure,the curve of signal lighting loop circuit current collected by CSM is summarized into four typical fault modes and one normal mode,the curve characteristics and causes are summarized,and the sample data set is established.Secondly,according to the characteristics of signal lighting loop current curve,the wavelet packet energy spectrum feature extraction method,and information entropy plus root mean square hybrid feature extraction method based on VMD,are used to extract the sample feature sequence set of signal lighting loop circuit current curve,which is used as the input of subsequent grey comprehensive correlation model.Then,the basic theory of grey comprehensive correlation analysis is summarized.Firstly,the sample for dimensionless processing is building,and divided into reference sequence and comparison sequence.After the grey comprehensive correlation model is established,the dimensionless multiple groups of reference sequences are processed through the grey correlation matrix to obtain the unique feature vector of the reference sequence.At this time,the grey comprehensive correlation model is used to calculate the grey comprehensive correlation degree between the comparison sequence and the reference sequence,and the reference sequence corresponding to the maximum value is the fault category.Finally,in the MATLAB simulation environment,there are five models:grey comprehensive correlation model based on wavelet packet energy spectrum,grey comprehensive correlation model based on VMD hybrid feature extraction,k-nearest neighbor(KNN)model based on wavelet packet energy spectrum,KNN model based on VMD hybrid feature extraction and dynamic time warping(DTW)model,the fault diagnosis simulation experiments are carried out on these five models respectively.The performance evaluation index of fault diagnosis method is selected to compare the advantages and disadvantages of the model.The results show that the grey comprehensive correlation fault diagnosis method based on VMD hybrid feature extraction is the most effective for signal fault diagnosis.Finally,based on the grey comprehensive correlation fault diagnosis method of VMD hybrid feature extraction,with the help of the CSM environment of laboratory,the modular software system is realized by using C# programming language and My SQL database.The software includes the functions of signal circuit current query and display,signal fault diagnosis,signal fault maintenance prompt and follow-up correction,fault sample set and diagnosis model update.After testing the software function,it shows that the system can diagnose the signal fault quickly and effectively.There are 56 figures,21 tables and 62 references.