Nonlinear multiple-input multiple-output (MIMO) processes which are common in industrial plants are characterized by significant interactions and nonlinearities among their variables. Thus, the tuning of several controllers in complex industrial plants is a challenge for process engineers and operators. This paper addresses the problem of simultaneously tuning n proportional-integral-derivative (PI/PID) controllers in a coupled multivariable process as a multi-criteria optimization problem. A multi objective genetic algorithm modified by a niching technique with castes formation is proposed to solve this problem. The optimization is carried out in two levels. In the first level a local function that considers both the integral time squared error (ITSE) and the minimum variance criteria is computed to separately evaluate the performance of each closed loop. Thereafter, a global cost function that considers all the loops is used to compute a set of solutions (a set of PI/PID parameters) to the optimization problem. The proposed system was applied to the control of a fluid catalytic cracking (FCC) unit, and its performance was compared to dynamic matrix control (DMC). The results show the applicability and effectiveness of the proposed method.
MIMO control; PI/PID control; Multi objective genetic algorithm; niching; castes
