Multiobjective optimal controller design with genetic algorithms

Finding a controller for a given plant in order to achieve a number of design objectives is a common control design problem. As well as closed loop plant stability, design objectives often include measures such as rise time, settling time, overshoot, asymptotic tracking, decoupling and regulation, gain and phase margins, small disturbance response and bounds on frequency response magnitudes. Genetic algorithms have previously been shown to be useful in addressing ill-behaved optimization problems, being able to cope with discontinuities, multimodality and uncertain function evaluations, and their single objective formulation has been extended by the authors to include multiple objectives. The paper shows how genetic search can be interactively used to design controllers of given complexity, in a multiobjective sense, while learning about the trade-off between the design objectives.