Distributed model predictive control for reactor-separator

Abstract

For the complex non-linear industrial process with many variables and coupled in chemical production process, such as reactor-seperator process, some traditional control methods, for example, Centralized Model Predictive Control(CMPC) and PID have limitation for effective manipulating those process, CMPC has poor real-time performance, PID has limitation for multi-variable coupled process. Considering these deficiencies, the Distributed Model Predictive Control(DMPC) strategy is proposed to effectively control the nonlinear and multi-variable process system for improving the real-time performance. The nonlinear system is divided into several subsystems, and after the subsystem is linearized, the model predictive controller is designed respectively, through setting the parameters of the controller reasonably, the control requirements are achieved, and the error is reduced, at the same time, the real-time performance is improved. The simulation of the reactor-separator process shows that the DMPC strategy can effectively control multivariable industrial processes. Meanwhile, compared with the CMPC strategy, the output errores can better reduced by using the proposed DMPC even if the adverse factors exist in the system, meanwhile the real-time performance is improved.