Neural Network And Fuzzy-Based Adaptive Control For Switched Nonlinear Systems

With the expansion of the industrial system complexity,more and more prac-tical systems that cannot be simply described as the single model systems,and the multiple model systems can well explain their characteristics.Switched systems,as a class of typical multiple model systems,have received extensive attention of re-searchers,and a large number of results about switched systems have been reported.In these results,most of them focus on the switched systems without uncertainties.However,during the actual production process,the controlled systems are usually in a complicated environment.Influenced by the uncertain factors and the changes of themselves,the uncertainties often exist in the controlled systems.These un-certainties make the structure of switched systems more complex and increase the difficulty of system analysis and design.This dissertation studies some neural network and fuzzy-based adaptive con-trol problems by combing the intelligent approximator with adaptive technology for switched nonlinear systems with unknown functions.The prescribed performance control,fault detection and fault tolerant control are studied.The main contents are listed as follows:(1)The prescribed performance control problem is investigated for uncertain switched nonlinear systems in general form,in which the case that all the subsystems have common constraint is considered.We use the radial basis function neural networks(RBFNNs)to approximate the unknown terms.And by using the common Lyapunov method,we design the controller and adaptive laws to guarantee the stability of the transformed systems,which implies that the original systems are also stable and the tracking errors satisfy the prescribed performance constraints.(2)The prescribed performance control problem is studied for uncertain switched nonlinear systems in general form with different constraints.The different performance constraints always lead to different coordinate transformations.This will bring great difficulty in analysis and design.In order to avoid the difficulty caused by different coordinate transformations,we construct a common coordinate transformation.Based on the dwell time technique,the subsystem adaptive con-trollers are designed such that all the signals in the closed-loop system are bounded,and the tracking errors satisfy the multiple prescribed performance constraints.(3)The fault detection problem is studied for switched nonlinear systems with unknown functions.Firstly,an adaptive switched nonlinear observer is designed for the nominal system.Then,an observer-based fault detection scheme is established by using the average dwell time method.And the switched threshold is proposed,which means that each subsystem has its own threshold.This kind of threshold can reduce the conservativeness caused by the common threshold.(4)Based on the average dwell time method and the backstepping technique,the adaptive fault tolerant control problem is investigated for a class of switched nonlinear systems with unknown functions and unmeasurable system states.Firstly,the robust fault tolerant control problem of switched systems is studied.Both the actuator loss of effectiveness and bias are considered.The robust fault tolerant controllers for subsystems are constructed such that all the signals in the closed-loop system are bounded in spite of faults.Secondly,the active fault tolerant control problem is investigated for the switched systems with process faults.Due to the existence of faults,some unstable subsystems are allowed to exist.The designed fault tolerant controllers can guarantee the boundedness of all the signals in the closed-loop system.(5)The adaptive tracking control problem is studied for a class of discrete-time switched nonlinear systems with dead-zone input.The fuzzy logic systems are used to approximate the unknown terms.The parameters of dead-zone are unknown but bounded.In order to reduce the conservativeness cause by that all the subsystem parameters are required to satisfy some certain conditions,some subsystem parameters are allowed to not satisfy these condition.That is to say,some unstable subsystems are allowed to exist.The designed controllers guarantee that the closed-loop system is stable under a class of switching signals with average dwell time.(6)The optimal control is studied for discrete-time switched nonlinear systems with unknown backlash-like hysteresis.Firstly,the nonswitched case is investigated.In order to avoid the noncausal problem,which means that the future information is used in current step,the original system is transformed into a prediction model.Based on the Lyapunov stability analysis,the designed controller and adaptive laws not only guarantee the stabilities of the closed-loop system,but also minimize the performance index.Then,based on this result,the considered system is extended to the discrete-time switched nonlinear system.The optimal control problem of discrete-time switched nonlinear system with unknown backlash-like hysteresis is solved.Finally,the conclusions and further research topics are presented.