Fault Diagnosis And Fault Tolerant Control For Non-Gaussian Nonlinear Stochastic Distribution Systems Using Neural Network

With the rapid development of science and technology,the industrial system becomes more and more complex,and the probability of fault occurrence is more and more big.Thus,the research on fault diagnosis and fault tolerant control has received more and more attention.In the past thirty years,this research has been developed very quickly.So far,the method based on experience knowledge,signal processing or analytical model is adopted to solve the related problems.After the fault occurs,the optimal control,PI control,sliding mode control is used to maintain the system stable and satisfy some performance indexes.The above process is called fault tolerant control.There are various kinds of random disturbances such as random parameters or sensor noise in the actual industrial production.The traditional research of stochastic systems is based on the assumption that stochastic variables obey Gaussian distribution.However,in practical applications,this assumption is not always true.However,the fault,noise,input and output of most stochastic systems do not obey the Gaussian distribution,so it is necessary to study the ralated problems non-Gaussian stochastic systems.Prof.Hong Wang proposed the stochastic distribution control(SDC)theory,in which the crisp control input is directly designed to make the output probability density function(PDF)track the given distribution.The actual system is essentially nonlinear system,however,this research of non-Gaussian nonlinear SDC system is not perfect.The specific contents in this thesis are given as follows:(1)For the non-Gaussian nonlinear SDC system using the Takagi-Sugeno fuzzy model,a complete set scheme of fault diagnosis and fault tolerant control is given.A new fault diagnosis algorithm basing on RBF neural network is adopted to diagnose the fault that occurred in the system.Using the DE algorithm,the parameter of fault diagnosis is optimized.Based on the fault diagnosis information,the sliding mode control method is adopted to construct the fault tolerant controller.Integral switching surface and sliding mode control law can ensure that the system is asymptotically stable.The simulation results have further confirmed the fault diagnosis and fault tolerant control results.(2)For the non-Gaussian nonlinear SDC system,for the nonlinear term,the Lipschitz condition does not need to be satisfied.A RBF neural network is used to approximate the nonlinear part,so the nonlinear observer is established.Another RBF neural network is used to diagnose the fault.Based on the fault diagnosis information,a PI fault-tolerant controller which is based on fuzzy control is designed.The control parameters can be online self-tuned.The simulation results have further confirmed the fault diagnosis and fault tolerant control results.(3)For the non-Gaussian nonlinear singular SDC system,an adaptive fault diagnosis observer is designed to estimate the fault.According to the relevant data,the corresponding linear matrix inequality is solved,and the gain vector of the fault adaptive control law is obtained.Using the fault estimation information,a PI active fault tolerant controller based on BP neural network is designed.The control parameters can be online self-tuned.The simulation results have further confirmed the fault diagnosis and fault tolerant control results.