An efficient method to design a series cascade controller using multi-objective optimization

Multi-objective optimization (MOO) technique, elitist Non-dominated Sorting Genetic Algorithm (NSGA-II), has been used for the design of a Proportional — Integral (PI)/Proportional (P) controller for a series cascade system. Unlike the earlier method of single objective function optimization based controller design which gives the single best controller, the present method generates multiple sets of controller parameters. The present method is shown to exhibit better closed loop performance than the existing methods for the cascade controller. This proposed method is also a single step tuning procedure in that the controller parameters in both loops are simultaneously calculated. Application of the MOO concept has also shown how the chosen objective functions are related to each other. This apriori knowledge of the interdependence of objective functions is important in the design of controllers since it constrains the best performance achievable. This also helps in anticipating the tradeoffs in the objective function behavior.