Design of a Knowledge-Based Controller for Intelligent Control Systems

Abstract

A hierarchical knowledge-based controller is proposed to improve the performance of complex control systems, such as robots. Unlike parameter- and performance- adaptive controllers, this controller is designed only to modify the reference input of a low-level servo controller. Because the internal parameters and structure of the low-level controller are not affected, commercial servo controllers can be made to perform more sophisticated tasks than originally intended. The principle of the knowledge-based controller, modification of the reference input, knowledge representation, existence of the solution, and analyses of the controller's stability and tracking error are described in detail. A self-tuning multiple-step predictor is designed as a part of the controller to eliminate the undesirable effects of system time delay. Both linear and nonlinear example control systems are tested via extensive simulations and have all shown promising performances.