Fault Diagnosis Of Hybrid Vehicles Based On FA Optimized SOM-BP Neural Network

With the rapid development of global automobile industry has brought air pollution and energy consumption problems,and hybrid vehicles have become one of the effective methods to solve this problem.The safety and reliability of hybrid vehicle is an important factor to ensure the safety of life and property of drivers and passengers.Fault diagnosis technology judges whether there is a fault by analyzing the parameters of hybrid vehicle.Therefore,it is necessary to study the fault diagnosis technology of hybrid vehicles.In this paper,the composition of hybrid vehicle system,fault types and fault causes are analyzed,and intelligent algorithm is applied to the fault diagnosis of hybrid vehicle.The specific research contents are as follows:(1)Firstly,the structure of hybrid vehicle is introduced,and the common faults and causes are summarized.The advantages and limitations of traditional fault diagnosis methods and intelligent algorithm based fault diagnosis methods are studied.(2)Aiming at the redundancy of hybrid vehicle information,PCA(Principal component analysis)is used to reduce the dimension of original information;aiming at the limitation of traditional fault diagnosis methods and the nonlinear mapping problem of hybrid vehicle fault information and type,neural network is used for fault diagnosis by its strong nonlinear processing ability.The algorithm flow of fault diagnosis based on SOM-BP(Series network of SOM neural network and BP neural network)neural network is described,and the feasibility of the method is proved by simulation.(3)Aiming at the problem that the network performance of SOM-BP neural network is affected by the connection weight and threshold value,this paper proposes to use FA(Firefly algorithm)to optimize SOM-BP neural network,analyzes the basic idea and algorithm flow of FA,constructs the hybrid power vehicle fault diagnosis model based on FA optimized SOM-BP neural network,and carries out simulation,the simulation results verify that the method has higher accuracy Accuracy and fast convergence.