Performance of optimal and standard controllers for disturbance rejection in industrial processes

Abstract

The performances of standard (proportional plus integral) and of optimal controllers, designed according to IMC (internal model control) and LQOC (linear quadratic optimal control) methods, are compared at equal robustness, for disturbance rejection. Optimal controllers, which account for the specific type of input, give a much superior performance in the case of delay-dominant processes, while for lag-dominant processes they may be found to give a very sluggish return to steady state. Their performance can be improved by considering that a large time constant acts as an integrator on the input disturbance, and then a higher order filter must be used. The LQOC method gives optimal results also in this case, though at the expense of a much more complex algorithm and design procedure. The much simpler design makes the IMC methodology preferable in the case of delay-dominant processes, as the design for robustness depends only on filter parameters and not on process and disturbance. >