Commande predictive par la theorie des intervalles flous et metaheuristiques

Abstract

In this work, a robust predictive controller was developed based on fuzzy intervals theory. The drawback of this method is that it is time consuming. To raise this problem, the applicability of metaheuristics to determine the online predictive control optimal solution was studied. For this, a comparison between three metaheuristics was carried out: ant colony optimization and particle swarm optimization and gravitational search algorithm. Results show that particle swarm optimization algorithm converges faster. The latter was applied (by an experimental study) for the control of two wheel mobile robot. After that, a simulation study was carried out on the applicability of multi objective metaheuristics for solving multi objective predictive control. The results show that the multi objective particle swarm optimization algorithm is encouraging for real-time applications and can be an alternative to other methods which are generally more difficult to implement