Multistage control of a fuzzy system using a genetic algorithm

We consider multistage control of a fuzzy system, given by a fuzzy state transition equation, under fuzzy constraints and fuzzy goals. First, we briefly survey previous basic solution methods of dynamic programming and branch-and-bound, which basically require some "trickery", and are plagued by low numerical efficiency, and then sketch Kacprzyk's (1993) approach based on possibilistic interpolative reasoning aimed at enhancing the numerical efficiency but requiring a solution of a simplified auxiliary problem, and then some "readjusting" of the solution obtained. Then, we propose the use of a genetic algorithm. The real coding and specially defined operations of crossover, mutation, etc. are employed. The results obtained seem to be promising.<>