Adaptive Decoupling of Multivariable Systems Using Extremum-Seeking Approach

Decentralized control of MIMO system is still the dominating control strategy in industry for its ease in design and maintenance, despite its known shortcomings in performance. In this article, an online decoupling scheme of MIMO processes using an extremum-seeking approach is proposed. By applying the developed method, at first, the system's loop interactions are estimated in real-time using phasor extremum seeking, and then, a pre-compensator matrix is calculated and augmented to the process by minimizing the loop interactions. The presented online decoupling approach is a model-free method, and only the plant's inputs and outputs are analyzed to provide the decentralized control protocol. Numerical results provide the effectiveness of the developed method for a two-input, two-output (TITO) control system.