Study On 25Hz Phase Sensitive Track Circuit Fault Diagnosis Of Fuzzy Neural Network Based On Bat Algorithm

As one of the main modes of transportation,railway transportation has received widespread attention for safety and reliability.As basics part of railway transportation equipment,The 25 Hz phase-sensitive track circuit is used in most railway stations.If it occurs failure,train operation will be delayed,operational efficiency will be reduced,and even security incidents will be occered.Therefore,its working status and performance are crucial to the operational efficiency of the railway and the safety of personnel.At present,the fault diagnosis of 25 Hz phase-sensitive track circuit equipment is inefficient and inaccurate,and experience of depends on maintenance personnel is make judgments and positioning faults.With the rapid development of railway transportation,its maintenance methods and diagnostic methods cannot be improved effectively.Therefore,in order to shorten the maintenance time of the equipment and locate the cause of the failure quickly and accurately,and it becomes increasingly important to perform maintenance in a timely manner.it is necessary to establish an intelligent fault diagnosis method for quickly and accurately find the cause of equipment failure and improve maintenance efficiency.Thence,this thesis proposes an Improved Bat Algorithm-Fuzzy Neural Network(IBA-FNN)model to make deeply study on the fault diagnosis of 25 Hz phase-sensitive track circuits.This thesis mainly completes the following contents:Firstly,based on the analyzes of the working principle of 25 Hz phase-sensitive track circuit,this thesis analyzes the common faults of 25 Hz phase-sensitive track circuit in the station and the causes of faults for the different working states.Through the simulation analysis of the track section characteristics,the calculated value of the critical state of the voltage is obtained and applied to the alarm output of the system.Then,combined with the main research of this thesis,the overall block diagram of the fault diagnosis process for a 25 Hz phase-sensitive track circuit is obtained.Secondly,Bat Algorithm(BA)is introduced,and then the standard BA is briefly described,including its basic idea,supere echo localization behavior,characteristics and basic steps.In order to verify the standard BA has a good optimization performance,the comparison tests and analysis of iterative functions such as Genetic Algorithm(GA),Particle Swarm Optimization(PSO),and BA are performed through standard test functions.Finally,BA is improved,an Improved Bat Algorithm(IBA)is proposed,and its effectiveness is verified by standard test functions.Finally,according to the flight process of simulated bats,an IBA is used to optimize the fuzzy BP neural network,an IBA-FNN model for fault diagnosis of 25 Hz phase-sensitive track circuits is constructed,and using the track circuit to train the sample for its simulation training,the diagnostic output of the IBA-FNN model can be obtained.The comparison between different models shows that the IBA-FNN model can not only reduce the training number,shorten the training time,but also improved the accuracy of the search,and effectively improve the accuracy of the fault diagnosis of 25 Hz phase-sensitive track circuits.