Multivariable Disturbance Observer Based Advanced Feedback Control Design and Its Application to a Grinding Circuit

Design of advanced feedback control (AFC) for optimal process operation has been mostly based on a hierarchical structure for many years. Since many advanced control algorithms (such as the model predictive control) do not handle disturbances directly in their design phase, it is difficult to achieve satisfactory performance in controlling complex process operations in the presence of heavy disturbances and large uncertainties. Focused on this practical challenge, in this paper we propose a novel multivariable disturbance observer (MDOB) to improve the disturbance rejection performance of conventional AFCs. The MDOB formulation is based on the approximate inversion of the multivariable generalized system, which consists of the multivariable operation process and the lower level basic feedback control system. All the stable and realizable MDOBs are characterized in terms of time delays and nonminimum phase zeros of the open-loop generalized systems. In the proposed MDOB-based AFC, the advanced feedback controller acts as a presetting controller to generate the proper pre-setpoint for the lower level basic feedback control (BFC) system such that a desired setpoint tracking is achieved. The MDOB acts as a compensator to enhance the operational performance of the process by dynamically adjusting the setpoints of the BFC according to the observed disturbances and plant uncertainties. Theoretical analysis, simulation comparisons, and experimental evaluation using a hardware-in-loop simulation platform of a grinding circuit are given, showing the effectiveness, validity, and advantages of the proposed approach.