Distributed Estimation with Decentralized Control for Quadruple-Tank Process

Abstract

This paper presents a unified modeling, control, and estimation framework for the quadruple-tank process, a benchmark multivariable system that exhibits either minimum phase or nonminimum phase behavior depending on valve flow ratios. A decentralized PI control strategy is employed to regulate water levels, while a distributed state estimation scheme is developed using local Luenberger observers and inter-agent communication. Each observer uses only local output measurements and exchanges information with neighboring nodes over a strongly connected communication graph. To address the limitations of partial observability, the observer design incorporates an observability decomposition and consensus-based coupling that ensures convergence to the true system state. Simulation results validate the effectiveness of the proposed framework, demonstrating accurate state reconstruction and stable closed loop performance under both minimum-phase and nonminimum phase configurations. These results highlight the potential of combining decentralized control with distributed estimation for scalable, networked control of complex multivariable systems.