Robust Optimal Control Of Uncertain Nonlinear System Based On Adaptive Dynamic Programming

Recently,the optimal control problems of uncertain nonlinear systems have attracted wide attention.On the one hand,uncertainty and nonlinearity generally exist in the actual engineering systems.On the other hand,in addition to stabilizing the systems,the control tasks often require some performance indexes to be optimal.Therefore,it is of great practical engineering significance to study the robust optimal control of uncertain nonlinear systems.This dissertation adopts adaptive dynamic programming(ADP)method which combines with neural network and adaptive critic to propose several robust optimal control strategies for uncertain nonlinear systems.The main work of this dissertation is as follows:The first part studies robust optimal control for a class of matched uncertain continuous time nonlinear systems.First,assuming the uncertain terms are bounded and the bounded information is known.The bounded information is used to transform the robust control problems of controlled systems into optimal control problems of the nominal systems.And the robust optimal control of the controlled systems is given.Then,assuming the uncertain terms are continuous but the bounded information is unknown.The neural network is designed to estimate the uncertain terms and using ADP and adaptive critic method is to compensate for the impact of the disturbances on the systems as well as making the cost function optimal.Finally,the simulation results verify the correctness and effectiveness of the control strategy.The second part studies robust optimal control for a class of non-matched uncertain continuous time nonlinear systems and the relevant conclusions of the matched case are generalized to the non-matched case.First,assuming the uncertain terms are bounded and the bounded information is known.The robust control problems of the controlled systems can be transformed into optimal control problems of the auxiliary systems by constructing the auxiliary systems.And the robust optimal control of the controlled systems is given.Then,assuming the uncertain terms are continuous but the bounded information is unknown.The combination optimal controller,included a disturbance compensation controller and a stabilization controller,is designed to compensate the impact of disturbance influence on systems and make the cost function optimal respectively.Finally,the simulation results verify the correctness and effectiveness of the control strategy.The third part studies robust optimal control based on data-driven for a class of matched uncertain nonlinear systems with unknown dynamic information.First,assuming the uncertain terms are bounded and the bounded information is known.The robust optimal control problems of the controlled systems are transformed into the optimal control problems of the nominal systems by using the bounded information.The unknown dynamic information and the optimal cost function of the nominal systems are identified via constructing a neural network,so as to acquire the robust optimal control of the controlled systems.Then,assuming the uncertain terms are continuous but the bounded information is unknown.The neural network is constructed to identify the unknown dynamic information and the optimal cost function.And the data-driven control strategy is proposed,which can not only compensate the impact of disturbance on systems but also make the cost function optimal.Finally,the simulation results verify the correctness and effectiveness of the control strategy.