Fault Diagnosis System Of Track Circuit Based On RBF Neural Network And Information Fusion

As one of the infrastructures in railway train operation control system,the running state of track circuit system directly affects the safety and efficiency of train operation.At present,the fault rate of the track circuit in the practical application is in the forefront of the signal equipment,but it still depends on the experience of the staff in the process of fault diagnosis most of the time.It is still an important subject to improve the accuracy,efficiency and intelligence of the fault diagnosis of railway track circuit.This paper was combined with the subject called Railway Electrical Equipment Fault Diagnosis and Early Warning System Based on Deep Learning and Fault Tree.It put forward a kind of track circuit fault diagnosis model based on RBF neural network and decision level information fusion,and the contrast experiment of model performance was proceeded.Finally,the prototype system of track circuit fault diagnosis was realized.The main research contents include:(1)Track circuit data was analyzed and reprocessed.The original track circuit voltage data of the electricity section was analyzed.The data characteristics of circuit fault were analyzed by circuit voltage graph.The Pan Ta anomaly detection method was used to extract the fault data.In order to solve the problem that the dimension of fault data was inconsistent,Cubic Spline Interpolation algorithm was used to unify the voltage characteristic values.Finally,the track circuit fault data set was obtained.(2)It was specific to research and comparative analysis of the single fault model.The RBF neural network algorithm was mainly studied.The K-means clustering algorithm was used to determine the center value of the network.After that,the number of hidden layer nodes in the network was determined by several experimental methods and the sample data set was used to test.At the same time,the BP Neural Network model and the Support Vector Machine model were studied.The model was constructed by MATLAB,and the model performance test was carried out by using the data set.The three models were compared and analyzed.(3)To improve the accuracy of fault diagnosis,a fault diagnosis model of track circuit combined with RBF and information fusion was proposed based on the initial diagnosis and the information fusion method.Decision level fusion based on RBF.Neural network was on the basis of the preliminary diagnosis results in three different classifiers: the RBF neural network,SVM and BP neural network.The model was trained by the training set and test set was used to verify.The experimental results showed that the model could effectively improve the accuracy of fault detection.(4)The railway track circuit fault diagnosis prototype system was designed and implemented by Visual Studio,Matlab,SQLServer.The system realized the function of monitoring station,the real-time monitoring of the track voltage fault and the intelligent learning of the circuit fault.After testing,the system was stable,the results were correct,indicating the feasibility of the model.