Research On Fault Diagnosis For ZPW-2000A Jointless Track Circuit Based On Neural Network

Track circuit is one of the most important equipments in the railway signal system. It plays an important role in ensuring safe and efficient operation of the train operation.If the track fails,it is most likely to cause major accidents.At present, in the way of diagnosis of track circuit,we can not reach the stage of complete state of repairation.The failure detection of the track circuit is mainly reflected in the test of field staff and the computer monitor. However, these methods are not comprehensive nor intelligent. Therefore, we need to explore a better diagnostic method.This dissertation takes ZPW-2000A track circuit as the study object.Then,it analysises the structure and the working principle of the track circuit.Then,Summarizing the common failure,analyzing the reasons for the failure and the emergence of the phenomenon.Taking several failures of track circuit for example,establish the BP neural network model.Based on the Neural network,in order to perfect the track circuit fault diagnosis model we use genetic algorithms to optimize the neural network algorithm.Realize the detection and diagnosis of the track circuit.This dissertation mainly complete the following work.Firstly,according to the characteristics of ZPW-2000A track circuit, summarize and analysise its working principles. Then analyzes the causes of common track circuit failures.Researching in the Railway Bureau,based on field research summarize and conclude the process of track circuit fault diagnosis, and according to the diagnostic process to extract the fault characteristics and the fault types of the track circuit.Secondly,Taking several common failures of track circuit for example,uses BP neural network to create the fault diagnosis model of the track circuit, on this basis, as a result of the BP neural network’s prone to local optima and the slow convergence, we use a genetic algorithms to optimize the weights and threshold of neural network.So,we use a GA-BP method as as the core technology to achieve fault diagnosis of track circuit.Thirdly, using Matlab to simulate the neural network model and compare the simulation results.From the diagnostic results we can see all the methods are applicable to the track circuit fault diagnosis.And the method of using genetic algorithm to optimizate neural network is better than the BP neural network.