Globally Stable Neural Network Control For Uncertain Output Feedback Systems

Most of the dynamics models of the actual system are nonlinear and inevitably affected by system uncertainties and external disturbances,so it is difficult to establish an accurate system model.In addition,the uncertainty of the system has a significant impact on the control performance and closed-loop stability of the whole system.Adaptive control can identify the model parameters of the system and adjust them online,which has a good control effect on the uncertain system.On the other hand,neural network can approximate any unknown nonlinear function with arbitrary accuracy in the approximation domain,which is a very promising modeling tool for unknown uncertainties.In recent years,adaptive control methods for uncertain nonlinear systems based on approximation have been widely studied,and a lot of valuable research results have been obtained.However,there are still many problems that need to be further studied and solved.Based on this,the global stable adaptive neural network control problem of uncertain output feedback system is studied in this paper.The main contents are as follows:1.Globally stable adaptive neural network tracking control for uncertain output-feedback systemsThe problem of globally stable adaptive neural network tracking control for a class of uncertain output feedback systems under disturbances with unknown bounds be investigated.Compared with the existing adaptive neural network control method for uncertain outputfeedback systems,the differences of the proposed scheme are as follows.The designed actual controller consists of neural network controller working in the approximate domain and robust controller working outside the approximate domain,in addition,a new smooth switching function is designed to achieve the smooth switching between the two controllers,so as to ensure the globally uniformly ultimately bounded(GUUB)of all closed-loop signals.Lyapunov analysis method is used to strictly prove the global stability of the whole closed-loop system under the combined action of unmeasured states and system uncertainties,and the output tracking error is guaranteed to converge to an arbitrarily small neighborhood through reasonable selection of design parameters.Finally,two simulation examples are provided to verify the effectiveness of the proposed control method.2.Neural network-based globally stable adaptive control for uncertain output-feedback systems with prior tracking accuracyA globally stable neural network-based adaptive control scheme is proposed for a class of uncertain output feedback systems with prior tracking accuracy and time-varying bounded disturbances.A pre-specified flat zone in the neighbourhood of the origin is introduced into Lyapunov function to achieve the estimation of unknown parameters.A novel symbolic function is designed to avoid the dead zone problem and ensure the differentiability of virtual control.To reconstruct the unmeasured states,a reduced order filter driven by the control input is designed,and the state estimation error is bounded by the dynamic signal driven by the system output.Barbalat's lemma is used to strictly prove that all variables of the closed-loop control system are globally uniformity ultimately bounded(GUUB)and the output tracking error converges to the specified small neighborhood of the origin.Finally,the effectiveness of the proposed control method is verified by providing two simulation examples.