In this paper we propose an adaptation of an algorithm based on biological evolution to obtain optimal control for the long run average cost problem for Markov jump linear systems. There is no in the literature a method that provides, proven, the optimal control of the problem, nor comparatives studies of different methods. The algorithm employed differs from the genetic algorithms to replace the basic operators for rolling a drawing according to a probability distribution. Comparing the proposed algorithm with a widely used method for this class of problem, leading into account the cost obtained, CPU time and amount of problems in which the stopping criterion set has been reached.
Markov jump systems; evolutive algorithm; control problem
