MULTIVARIABLE PID-CONTROLLER DESIGN FOR A GASIFIER PLANT USING PENALTY-BASED-MULTI-OBJECTIVE GA

This paper presents a new technique for designing a multivariable PID controller and its application to a Gasifier benchmark problem. The design procedure allows the optimization of both controller parameters and controller structure. Simulation results confirm the superiority of this Penalty-Based-MultiObjective Genetic Algorithm approach over standard GradientBased searching techniques. By carrying out the design in several stages, the computation time required by the overall algorithm is considerably reduced. Applying the technique to the Gasifier problem has led to a robust controller that meets all given design objectives under all operating conditions, and satisfies all input/output constraints.