On-line multi-objective optimization of controllers for electric drives

This paper describes an automatic approach to design discrete-time linear controllers for electric drives. The design procedure is based on a genetic algorithm, which iteratively searches for the controller with the highest satisfaction of a predefined objective function. The fitness of each controller is measured directly on the hardware, and takes into account multiple and conflicting design aspects, such as settling times, overshoots and ripples on the applied current. As confirmed by the experimental results, the proposed approach exhibits interesting potentialities.