Adaptive Neural Network Control For Several Classes Of MIMO Nonlinear Systems With Non-Strict-Feedback Structure

In the early practical industrial applications,linear control theory played an important role.It can usually control the system behaviors to achieve the desired effects.With the development of society,the nonlinearities and uncertainties not only make the performance of the controlled system decline but also can cause the system to fail to run normally,which brings severe challenges to the design of the actual control systems.Therefore,both based on the development of control theory and the demand of practical application,the nonlinear control theory is in urgent need of further development.Based on the above discussions,this thesis studies the adaptive neural network control problem for several kinds of non-strict-feedback multiple-input/multiple-output(MIMO)nonlinear systems on the basis of deeply understanding of nonlinear system theory and intelligent control technique.The research contents of this paper are as follows:(1)The problem of tracking control for a class of non-strict-feedback MIMO nonlinear systems with input saturation is studied.In order to get the ideal controller,the radial basis neural networks(RBF NNs)are employed to approximate the unknown nonlinear functions and hyperbolic tangent functions are utilized to smooth the sharp corners of the input saturations.Then,Young's inequality is utilized to handle the nonlinear terms derived from the deducing process.Furthermore,a new neural networks controller is constructed by using the Lyapunov theory under the framework of Backstepping method.The controller can show that the whole closed-loop system is semi-global uniformly ultimately bounded(SGUUB)and the tracking error converges to a small neighborhood of the origin.Finally,a realistic example is provided to certify the effectiveness of the presented method.(2)The finite-time control and almost disturbance decoupling problems for a class of MIMO nonlinear systems with disturbances and non-strict-feedback structure is studied.As a classical function approximation tool,neural networks are employed to estimate the unknown nonlinearities and Young's inequality is utilized to cope with the disturbance terms derived from all subsystems.In order to realize finite time adaptive disturbance decoupling of nonlinear systems,a criterion named finite-time almost disturbance decoupling is first developed.Under this criterion,an new controller is designed via the Backstepping method and the appropriate selection of Lyapunov function.It is revealed that the proposed controller can guarantee all variables of the closed-loop system are bounded,and the performance of finite-time almost disturbance decoupling is realized.Finally,a practical example is employed to validate the effectiveness of the designed controller.