Expert System Architecture for Control System Design

This paper describes the development of an expert planning framework for design of control systems. The expert planner incorporates the knowledge of a control system designer in selecting and applying mathematical algorithms from several branches of control theory such as Kalman filter design, proportional/integral/ derivative control, and optimal state feedback. The framework has been developed directly in Lisp rather than using any of the available domain-independent expert system frameworks. The planning is done at two levels: a high level that uses general rules about control system design and a low level that uses rules about the use of a computer aided control system design (CACSD) package. The system has been implemented and demonstrated on a Xerox 1100 Lisp workstation controlling a CACSD package running on a VAX 11/780, designing a servo (command tracking) control system for aircraft.