Research On Adaptive Control Methods For A Class Of Constrained Uncertain Nolinear Systems

There always exist system uncertainties in practical nonlinear control plants,such as mechanical manipulators,mobile robots,aerospace devices,motor devices,and process control industries.Thus,improvement of control robustness of uncertain nonlinear control systems has always been a hot research topic.Besides system uncertainties,the nonlinear control systems are usually constrained by states or control inputs constraints.Ignoring the constraints will degrade system performance,resulting into instability of the closed-loop control system and even disasters.Therefore,study of adaptive control for constrained uncertain nonlinear systems is of great significance.Based on barrier Lyapunov functions(BLFs),backstepping method,dynamic surface control,and neural networks,this thesis studies adaptive control of constrained uncertain nonlinear systems,such that the desired tracking performance can be obtained with satisfaction of the constraints.The main contribution and innovations are summarized as follows:1)A locally weighted learning adaptive control strategy is proposed for a class of uncertain nonlinear systems with control input constraints and state constraints based on BLFs and a backstepping method.By considering the control input as an extended state,the system constraints are tackled by introducing BLFs into the backstepping method.The system uncertainties are estimated and compensated by piecewise linear approximations based on locally weighted learning neural networks.By theoretical analysis and simulation results,it is proved that the system can track the desired trajectory without violation of the system constraints under the proposed control.2)A composite locally weighted learning adaptive control strategy is proposed for a class of uncertain nonlinear high-order system models with control input constraints and state constraints based on BLFs.To eliminate the differential expansion of the backstepping control,the dynamic surface control technique is applied.The time derivative of virtual control is estimated by use of one-order low pass filters,which makes the controller and parameterdesign simpler and avoids the error bounded assumption and the circular argument.To improve the control performance,the neural network weight estimator is updated by composite errors of a tracking error and a prediction error.By theoretical analysis and simulation results,it is proved that the proposed composite adaptive locally weighted learning control can improve tracking accuracy and the approximation accuracy.3)A robust adaptive control strategy is proposed for a class of state constrained uncertain nonlinear systems with prescribed transient and steady-state behavior.The prescribed tracking performance can be characterized by the constraints on the output tracking error.Thus,the design of the prescribed performance control is translated into a robust control design of the constrained nonlinear system.Satisfaction of the system constraints,including the state constraints and the output tracking error constraints,is achieved by bounding integral barrier functions.The robust adaptive term is applied to compress the system uncertainties and ensure the stability of the closed-loop control system.The feasibility and effectiveness of the proposed robust adaptive control strategy are verified by theoretical analysis and illustrated by simulation results.