A feedback control scheme based on causal modelling

This paper presents a method for the feedback control of ill-defined systems based on a symbolic causal model of the system to be controlled. The motivation for such an approach is given, together with a description of the sources of knowledge that can be utilized to implement the controller design. In contrast to previous attempts at Knowledge-Based Control Systems based on rule-based methods, this approach allows the internal mechanism of the system to be represented, thereby allowing more accurate control to be achieved. The system is implemented in a real-time KBS shell, written in Prolog, and running on mini-computers. Experimental results are given, comparing the performance of the causal model scheme with that of a conventional microprocessor-based three-term controller. These show that although comparable performance is achieved, the KBS method has the distinct advantage of being easily understood by operating personnel and, hence, permitting meaningful interaction. © 1989.