Adaptive decoupling control of the forced-circulation evaporation system using neural networks and multiple models

Abstract

Control objectives of the forced-circulation evaporation process of alumina production include maintaining the liquid level and fast tracking of the product density to its setpoint. Due to the strong coupling between the level control and product density control loops and high nonlinearities in the process, conventional control strategies can not achieve satisfactory control performance and meet production demands. Viewing the forced-circulation evaporation system and valves as a generalized plant, a nonlinear multi-model adaptive decoupling control strategy is proposed. The nonlinear adaptive decoupling controller includes a linear adaptive decoupling controller, a neural-network-based nonlinear adaptive decoupling controller, and a switching mechanism. The linear adaptive decoupling controller is used to reduce the coupling between the two loops. The neural-network-based nonlinear adaptive decoupling controller is employed to improve the transient performance and mitigate effects of the nonlinearities on the system, and the switching mechanism is introduced to guarantee the input-output stability of the closed-loop system. Simulation results show that the proposed method can decouple the loops effectively for the forced-circulation evaporation system and can improve the evaporation efficiency.