Adaptive Neural Control For Nonlinear Systems With Unmodeled Dynamics

This article is concerned with adaptive neural networks(NN) control for a class of strict-feedback nonlinear plants. Firstly, the plants are dealt in the ways of indirect and direct NN-based adaptive control approaches, respectively. Then, we focus on the nonlinear plants with dead-zone and unknown control directions. Based upon the Lyapunov theory, the developed controllers guarantee the desired control performances for the closed-loop plants.The effectiveness of the developed schemes are respectively verified by simulation results.The compendious description is shown as follows:1. For a class of nonlinear plants with unmodeled dynamics, we develop an indirect adaptive neural control scheme. To deal with the unmodeled dynamics, a dynamic signal is introduced. Radical basis function(RBF) NN are employed to model the packaged unknown nonlinearities. It should be noted that the number of adaptive parameters is not more than the order of the considered plant. As a result, it makes our control design more suitable in the point of practical applications.2. The stabilization problem for nonlinear strict-feedback systems with unmodeled dynamics is considered. In the control design, an adaptive neural controller is constructed by combining the property of hyperbolic tangent function with RBF NN based on small gain theorem. Only one online updating equation is required for n-order systems, which makes the computational burden alleviated.3. For a class of nonlinear systems with unknown control directions, unmodeled dynamics and dead-zone, we propose an control scheme. During the control design, the dead-zone output is expressed as a simple linear system and bounded disturbance. Nussbaum gain functions are employed to handle the unknown control directions.