Comparison of linear and nonlinear model predictive control of benchmark drum boiler

Abstract

This paper deals with linear and nonlinear model predictive control (MPC) applied to a benchmark nonlinear boiler. The motivation for this research is the necessity to achieve tighter power plant control in the wide range load following operation. Since the boiler exhibits considerable nonlinearity, nonlinear MPC (NMPC) is the natural choice. On the other hand, linear MPC is much simpler to design and operate. For these reasons, three variants of MPC algorithm were designed (linear MPC, multiple MPC and NMPC) and compared in simulation in the terms of control performance versus complexity trade-off.