DECENTRALIZED LINEAR MODEL PREDICTIVE CONTROL FOR BOILER-TURBINE UNIT

Abstract

Boiler - turbine unit is an essential part in thermal power plants. To control and ensure stability of this nonlinear MIMO system, we can use a nonlinear model predictive controller (NMPC) with object function using penalty function, however it is difficult to select the parameter of penalty function and the solving online optimal problem affects real-time features of the system because of the volume of calculation. This paper proposes a decentralized model predictive control (DMPC) algorithm based on linear model boiler-turbine units. The interconnections between the subsystem are considered as perturbation terms and predicted by using the predictive information from local predictive controllers. Moreover, the input-to-state stability (ISS) of both subsystems and the overall closed–loop system is guaranteed and the performance of the proposed approach is demonstrated by the simulation results on the boiler-turbine system.