Research On State Estimation And Fault Diagnosis Of Nonlinear Control System Based On Real-time Switching

The actual industrial process control is generally affected by unfavorable factors such as nonlinear links,undetectable signals,equipment failures,unknown input disturbances,etc.Therefore,studying the state estimation and fault diagnosis of nonlinear control systems is a prerequisite and important to realize the closed-loop control function of the system.Guarantee has important theoretical and practical application value.Considering that traditional nonlinear control system state estimation and fault diagnosis methods based on Takagi-Sugeno fuzzy model are limited in application due to their excessive conservativeness,this thesis studies the state estimation and fault diagnosis problems of nonlinear control systems based on real-time switching in depth.By using the normalized fuzzy weighting function updated in real time by the fuzzy control system to construct different nonlinear control system state estimation/fault estimation working modes,effectively reducing the conservativeness of the corresponding state estimation/fault estimation design conditions,and obtainingvarious performance indicators that are better than those in the existing literature.Specifically,the research of this article is divided into the following three parts:Aiming at the relaxed state estimation problem of a class of discrete-time Takagi-Sugeno fuzzy systems,a switching observer based on the idea of fuzzy partition is proposed.By introducing a pair of weighted scalars,the entire interval formed by the normalized fuzzy weighting function updated at the current sampling time can be divided into a set of non-overlapping sub-intervals,and then based on the above-mentioned sub-intervals to define the switchingmode,and design a corresponding set of fuzzy observer gain matrix for each switching mode.The simulation results show that the obtained method is less conservative than the existing methods given in recent references.Furthermore,the scope of the fuzzy partition idea is extended from the normalized fuzzy weighting function at the current sampling time to the normalized fuzzy weighting function at the multi-sampling time,and a more effective multi-sampling time switching observer is proposed.In the design process,we introduced a set of variable weights for different sampling moments,and defined more sets of switching modes by categorizing the characteristics of the fuzzy control system shown by the normalized fuzzy weighting function changes at different sampling moments.And design more groups of fuzzy observer gain matrices in switching mode.The simulation results show that the proposed multi-sampling time switching observer further reduces the conservativeness of the state estimation based on the discrete-time Takagi-Sugeno fuzzy system.Aiming at the fault estimation problem of the discrete-time Takagi-Sugeno fuzzy system,a new type of multi-sampling time real-time scheduling fault observer is proposed.First,a new framework of fuzzy fault estimation observer is constructed.The information of the normalized fuzzy weighting function between adjacent sampling moments can be updated in real time at each sampling time,and it is used for real-time scheduling to determine the fuzzy fault estimation observer.The current activation mode.For each activation mode,by introducing the relevant free matrix in the design conditions,a special set of fault estimation observer gain matrix corresponding to the specific activation mode is obtained.In addition,it can be proved that the existing literature method belongs to a special case of the method proposed in this article.The simulation results show that our proposed method has certain advantages over existing related methods.