Research On Fault Diagnosis Of Hybrid System Based On Improved Particle Filter Algorithm

Hybrid system refers to a complex system in which continuous and discrete states exist simultaneously,and the two interact and couple with each other.Many large-scale complex industrial control systems are hybrid systems,and the hybrid characteristics of these systems increase the complexity of fault,which also pose challenges for the fault diagnosis.However,existing fault diagnosis methods for hybrid systems still have many shortcomings,and more accurate and effective methods need to be proposed.In this thesis,for the various faults that may occur in the hybrid system,the problem of fault diagnosis based on the particle filter algorithm is studied.Based on the built hybrid system model,the faults are classified.Combined with the improved particle filter algorithm,the fault diagnosis method of the hybrid system is studied.Through simulation experiment analysis,the feasibility of method is verified,which has important theoretical significance for improving the safety and reliability of hybrid systems.The main contents of this thesis include:(1)Taking hybrid systems such as the maneuvering target tracking system as the research object,hybrid automata is used to build the system model.The advantages of particle filter algorithm in the hybrid system are analyzed,and the state estimation method of hybrid system based on particle filter is derived.Through simulation,the effectiveness of the above method is verified.(2)Based on the state space model of hybrid system,the possible faults in the hybrid system are abstracted and classified.The calculation method of the estimated fault value is derived,and the particle filter algorithm is improved.Combined with the hypothesis test theory,a fault detection method is proposed.Combined with the fault signature matrix,the fault separation and identification are realized.Through fault injection,different types of hybrid system faults are simulated,and the simulation results show the feasibility of the fault diagnosis method.(3)The CTCS-3 train movement hybrid model is built by using hybrid automata,and the train movement simulation model is built by using Simulink,which realized the basic functions of the train control system.The causes of faults that may cause the train overspeed are analyzed,and the speed and position measurement faults are set as the main research objects.Using the proposed improved particle filter algorithm,fault diagnosis of three kinds of speed and position measurement faults is carried out.The simulation results show the applicability of the method in train control systems.