Research On Fault Diagnosis Of On-Board Equipment For Train Control System Based On BN Optimized By CS-PSO

The train control system is used as the control system for the automatic operation of high-speed trains.Its reliability and safety are essential to the smooth operation of trains.As a key component of the train control system,the on-board equipment is the speed running protection,which is the key to ensure the safe operation of the train.Its working performance and operating status are important indicators for efficiency and protection of train driving.If the on-board equipment fails,the train will be delayed and the driving efficiency will be reduced.In serious cases,it will cause collisions and other safety accidents.Therefore,it is important to quickly and accurately detect and eliminate faults to ensure driving safety,which is of great significance for the development of fault diagnosis technology for on-board equipment.At present,the problems of on-board equipment include that the system structure is complicated,the fault diagnosis characteristics are uncertain,the traditional fault diagnosis method relies too much on expert and maintenance experience,the degree of fault diagnosis is low,and so on.In order to shorten the fault processing time and improve the maintenance efficiency,it is necessary to establish an intelligent fault diagnosis method,which can quickly and accurately locate the cause of the fault and timely and effectively eliminate the fault.Therefore,in view of the above problems,this paper deeply studies the structural composition and fault characteristics of on-board equipment for train control system,and selects Cuckoo Search(CS)and Particle Swarm Optimization(PSO)intelligent algorithms to optimize Bayesian networks(BN).Based on the model construction,combined with expert prior knowledge and fault data perfect model,a BN fault diagnosis method based on CS-PSO was proposed.Through the diagnosis and analysis of the actual fault data of the CTCS3-300 T on-board equipment,the intelligent fault diagnosis of the on-board equipment is realized,and the efficiency and accuracy of the method are verified.The specific work of this paper includes:Firstly,the structure and function characteristics of the CTCS3-300 T on-board equipment are studied in detail,and the characteristics of fault diagnosis are analyzed.The fault appearance and cause of the fault for different fault modules are analyzed.The irregular fault tracking table recorded in natural language is used to conduct data mining of faults by using data mining technology,extract fault characteristics and fault triggering rules,and find out the potential correlation between fault feature points.The fault features are classified and coded,and a computer-identifiable coding fault diagnosis rule is established.Based on the expert prior knowledge and fault diagnosis features,a fault diagnosis prior knowledge database is established to provide a priori diagnostic rules for fault diagnosis of on-board equipment.Secondly,for the fault characteristics of on-board equipment,the BN is taken as the core algorithm,which is best for dealing with uncertain knowledge representations,and the CS and PSO intelligent algorithms are introduced to optimize the BN for the model construction.The basic flow of CS and PSO algorithms are analyzed,and the factors affecting the convergence of the algorithm are found out.Combined with the advantages of CS and PSO algorithms,the CS-PSO algorithm is improved.Depending on the structural characteristics of BN,the CS-PSO algorithm is applied to the structure learning of BN,and a structure learning algorithm of BN optimized by CS-PSO is proposed.Chest Clinic,Credit,Car Diagnosis classic network is used as the simulation model,and the modeling and simulation comparison of greedy algorithm,K2 algorithm,CS algorithm and CS-PSO algorithm are carried out.It is verified that the CS-PSO algorithm has fast convergence,high convergence precision and excellent stability in the structure learning of BN,and can obtain accurate Bayesian network structure model faster and better.Finally,the CS-PSO-BN structure learning algorithm is applied to the fault diagnosis of CTCS3-300 T on-board equipment.The fault diagnosis model of BN optimized by CS-PSO for on-board equipment is established,and the MLE algorithm is utilized to learn the parameters of the model to further optimize the model structure of BN.Through integration with the prior knowledge base,the optimal BN model structure for fault diagnosis of on-board equipment is constructed.Diagnostic output of the CS-PSO-BN model can be obtained by using on-board equipment fault training sample to simulate its training.Through the comparison of different models,it is concluded that the CS-PSO-BN model has fast convergence time,short learning time and excellent diagnostic model,which effectively improve the accuracy of fault diagnosis and provides a new idea for the efficient diagnosis of on-board equipment.