Distributed Model Predictive Control Based on New Cooperative Optimization Strategy for Nonlinear Large Scale Uncertain Systems

Page No.: 1-17 Volume: 13, Issue 1, 2020 ISSN: 1997-5422 International Journal of Systems Signal Control and Engineering Application Copy Right: Medwell Publications Abstract: In this study two linear cooperative Distributed constrained Model Predictive Control (DMPC) approaches are proposed to control the uncertain nonlinear interconnected large scale systems. In these approaches a proposed novel cooperative optimization strategy is employed that its advantage is to improve the centralized global cost function of each local controllers which decreases the control efforts, cost function values and convergence time compared to typical cooperative DMPCs which is demonstrated via. simulation results of a typical nonlinear large scale system. In proposed approaches two reconstructed linear distributed constrained model predictive controllers; Distributed Extended Dynamic Matrix Control (DEDMC) and Adaptive Generalized Predictive Control (DAGPC) are presented to control the uncertain nonlinear large scale systems by compensation of the mismatch between linearized and nonlinear models. The advantage of proposed controllers is their less complexity compared to fully nonlinear DMPCs. In DEDMC, the mismatch between linearized and nonlinear models is considered as a disturbance and in DAGPC this mismatch is compensated using online identification of the linearized model. The typical linear algorithms like distributed DMC leads to an unstable closed-loop response if the reference trajectory is a little far from the equilibrium point while this problem will be partially solved using the proposed DEDMC and will be completely solved using the proposed DAGPC even if the reference trajectory is too far from the equilibrium point. The performance of proposed approaches are demonstrated through simulation of a typical uncertain nonlinear large scale system.