Аннотация

In the past decades, both optimal control and adaptive control have been widely investigated to solve control problems of nonlinear systems arising from engineering applications. Optimal control aims at f i nding a control law to drive a control system to a desired state while optimizing certain performance index with or without con-straints. Adaptive control is a tool to handle parameter uncertainty or structure uncertainty of control systems. In most works, the two types of control methods are separated. Reinforcement learning and, in particular, deep reinforcement learning have attracted more and more research interest in recent years. Such type of learning methods could be a powerful tool for the control of nonlinear systems.
In this book, we present our systematic investigations on a near-optimal adaptive control method based on the Taylor expansion, neural networks, estimator design approaches, and the idea of sliding mode control. We mainly handle the tracking control problem of nonlinear systems under different scenarios, for which the output of a controlled nonlinear system is expected to track a desired reference output trajectory with respect to time t. The performance of the presented approach is theoretically guaranteed. Specif i cally, the performance index is asymptotically convergent to the optimal and the tracking error is asymptotically convergent to 0.
Issues like actuator saturations are also considered in this book. By combining the design method of the near-optimal adaptive control method and the novel ideas in the developments of methods for the redundancy resolution of redundant manip-ulators, two new redundancy resolution methods are also presented.
To make the contents clear and easy to follow, in this book, each part (and even each chapter) is written in a relatively self-contained manner.