Distributed model-free adaptive control for output-coupled interconnected processes

Abstract

A novel data-driven distributed control strategy for output-coupled interconnected industrial processes with unknown models is proposed in this paper. The method derives a Distributed Output-coupled Dynamic Linearization (DOCDL) data model that effectively captures the complex output coupling relationships between the subsystems, overcoming the limitations of traditional model-free adaptive control frameworks in handling output coupling terms. An Output-coupled Interconnected Processes Disturbance Observer based Distributed Data-driven Adaptive Control (OC-DO-DDAC) is developed to address the computational burden and accuracy deterioration associated with centralized control approaches. The controller incorporates both rate terms in its multi-factor objective function and dedicated disturbance observers to effectively suppress the propagation of the disturbances through the coupling channels, thereby enhancing system stability and robustness. The parameter vectorization approach employed in the distributed design circumvents the control accuracy issues arising from matrix norm simplification while maintaining computational efficiency as system dimensions increase. Theoretical analysis demonstrates the boundedness of the tracking errors through an innovative dimension-extended system approach combined with Gerschgorin’s circle theorem. The effectiveness of the proposed method is validated through comprehensive simulations and experimental studies on output-coupled systems.