Rule based control of an uncertain system using segmentation by simulated annealing and pattern recognition techniques

As the uncertainty in the parameter values in a linear system becomes large, the design of a single robust controller to achieve the desired dynamic response becomes more and more difficult. In the present work, the concept of segmentation of the parameter variation region is implemented using an optimization technique based on a simulated annealing algorithm. A rule-based system that consists of an online system identifier and a rule-based switching scheme is utilized to control the system. The results of a simulation study are presented. It is noted that this segmentation technique is particularly import ant in performance-related controller designs. The advantage in terms of a reduction in the number of segments and subsequent complexity of the rule-based controller can be seen from the example presented