Differential Evolution algorithm applied to sparse PI control synthesis for non-square multivariable systems

This paper presents a sparse PI control synthesis methodology for non-square multivariable systems aiming tracking response performance and decoupling among control loops. We formulate the control problem as a non-linear optimization problem in the space of the sparse PI control parameters. To formulate the control problem, we consider the H-infinity reference model approximation approach. We apply the differential evolution algorithm to solve the proposed optimization problem. We illustrate the efficiency of the proposed synthesis methodology considering a crude distillation unit that is non-square, with four process outputs and five manipulated variables, with multiple delay times, and high interactions among the control loops.