Evolved finite state controller for hybrid system in reduced search space

This paper presents an evolutionary methodology to automatically generate finite state automata (FSA) controllers to control hybrid systems. The proposed approach reduces the search space using an invariant analysis of the system. FSA controllers for a case study of two-tank system have been successfully obtained using the proposed evolutionary approach. Experimental results show that these controllers have good performance on the set of training targets as well as on a randomly generated set of validation targets.