A Nonlinear Industrial Model Predictive Controller UsingIntegrated PLS and Neural Net State Space Model

摘要

Model predictive control (MPC) technology has been well developed and successfully applied in therefinery and petrochemical process industries over last 20 years. Recent development has beenfocused on nonlinear MPC and robust MPC technologies because new challenges have beenencountered in the polymer and chemical industries where many processes show strong nonlinearityand uncertainty. This paper presents a nonlinear industrial model predictive controller, recentlydeveloped by Aspen Technology, Inc. This MPC controller uses a nonlinear, state space, integratedPLS and Neural Net model (Zhao et al., 1998), and a multi-step, constrained, Newton-typeoptimization algorithm (Oliveira and Biegler, 1995). It results in a robust and cost-effective industrialnonlinear MPC controller. A pH reactor example and a successful industrial application in NOxemission control of a power plant are presented to demonstrate the capability of this controller.