Model Predictive Control for MIMO Systems Integrated with Active Disturbance Rejection

Abstract

This paper presents a proposal for a novel predictive control scheme for uncertain and disturbed multi-input, multi-output (MIMO) systems. The proposed method is a fusion of a well-known and effective linear Model Predictive Control (MPC) and a modern concept, Active Disturbance Rejection Control (ADRC). The approach is to introduce a fictitious state that includes all possible uncertainties and disturbances originating from the plant that are unknown and not accounted for in the plant model and its estimation in real-time using an Extended State Observer (ESO). The estimated state is then used to generate control that supports the MPC controller with a simplified model of the plant. This leads to an easily designed, robust predictive control structure similar to the well-known tube-based method. It seems that the proposed control approach has potentially great advantages, is simple to implement, and is easy to tune. Due to this, it can be used in many control systems of difficult and inaccurately identified plants.