Fault Diagnosis Of Hybrid System Based On Particle Filter

Hybrid system is a kind of system which includes both continuous and discrete dynamic characteristics,and the interaction of two dynamic characteristics.The automatic control system whose size is relatively large has hybrid characteristics,so result in system complexity and various fault.In order to ensure the normal operation and safety of the system,there must be efficient and accurate fault diagnosis method for hybrid system;for nonlinear system,particle filter algorithm has more advantages than the Kalman filter.The thesis studies the fault diagnosis problem of hybrid systems based on particle filter algorithm,the starting points are the model establishment scheme of hybrid systems,fault classification and specific method of fault diagnosis and its performance evaluation method.Fault diagnosis of nonlinear systems are studied.The thesis emphasized on several aspects as follows:(1)The Sample degeneracy problem and the selection of importance density function problem are analyzed,and the advantages and disadvantages of SIR and UPF are compared.Because the classical logic system fault model is simple,and its fault injection does not have the random characteristic,according to stochastic hybrid system theory,a stochastic hybrid automaton model is established to simulate more complicated fault.(2)A stochastic hybrid automaton water tank level control system model is established,summed up abstraction and classification of the faults of the system:the implementation fault,observation fault,and switching fault.In order to save the problem that the common mode variable hybrid system particle filter estimation method can be used in complex logic model,a fault diagnosis statistical switching method of hybrid system based on particle filtering is proposed and verified.(3)In order to make the model more flexibility,the hierarchical hybrid system theory is introduced,specific method to realize hierarchical hybrid system theory based on the theory and the concept of the upper mode and sub mode is proposed.In order to save the problem that there is switch fault problem in the specific process of modeling the upper mode,put forward the concept of transfer modal and its methods.Hierarchical hybrid automata model is established.(4)A hybrid system with fault controlled migration model is established,SIR particle filtering method and smoothing method were used to study the fault diagnosis method of the upper mode switch fault on hierarchical hybrid system,and its performance index are analyzed and compared.The model of level conversion operation scene for CTCS-3 level train control system is introduced is used to implement the fault diagnosis method.