Track Circuit Fault Diagnosis Based On BP Neural Network

The track circuit occupies an important position in the railway signal system,is one of the most important parts of the signal equipment.To improve the efficiency of the railway transport,ensure railway traffic and ensure railway traffic safety plays a vital role.In recent years,China's national economic development.China rail especially Chinese high-speed rail out of the country,into the world,to ensure road safety and smooth flow,ensure the normal operation of track circuit in the harsh environment,reduce the track circuit failure rate has become a problem unsolved,which is to maintain staff maintenance ability and work efficiency put forward higher to the old method of using experience to judge the fault is inefficient,and can not meet the current track circuit fault maintenance work?This paper selects high voltage asymmetric pulse track circuit,as the research object,the application of artificial neural network in checking the problems of the system.Using the matrix laboratory(MATLAB)a neural network model for neural network training,the fault diagnosis of high voltage pulse track circuit.Based on the network structure of BP neural network is simple.And the construction of track circuit fault diagnosis system model of high voltage pulse based on BP neural network.The paper analyses and summarizes the code track circuit voltage peak,the receiver head wave peak voltage,transmission frequency of three groups of track circuit parameters and red Light and fault shunt(working high voltage power supply,power supply voltage is too low,the ballast resistance is too small,send the fault terminal transformer)nonlinear mapping relationship,through the training of the neural network,to achieve the track circuit fault diagnosis recognition.At the same time,the recognition of the high voltage pulse track circuit fault diagnosis of BP neural network,neural network for more in-depth understanding of the advantages of solving the problem.The analysis shows that the BP neural network model to track circuit failure,compared to the traditional method of fault prediction is feasible and more efficient,through the training of BP neural network is accurate in the fault diagnosis of high voltage asymmetric pulse track circuit,the error of high precision,fast convergence speed,good function and fault diagnosis generalization ability.