Rule-based control of slowly varying systems using compensator segmentation determined by simulated annealing

A novel compensator segmentation scheme for rule-based control of a linear system characterized by structured uncertainties is presented. It is assumed that the design of a segmented compensator becomes necessary because the parameter variation region is too wide to be handled by a single robust compensator. The segmentation scheme is based on the simulated annealing technique. Starting with an initial estimate of the number of segments, the offline segmentation scheme attempts to minimize it while assuring coverage of the entire parameter variation region. The information about the segment boundaries and the appropriate compensators is then utilized as the database of a rule-based switching controller. The results of simulation studies that demonstrate the performance of the proposed scheme are included