GA Based Predictive Controller

This paper presents a new solution to the model based predictive control problem whose main innovative characteristic is the use of a genetic algorithm to minimize the objective function which expresses the desired behavior for the control loop. This is a highly versatile approach because it permits the use of an objective function that is not necessarily quadratic and also the efficient introduction of constraints in both the control and controlled variables. Some preliminary results of this approach, obtained via a simulation study, are discussed.