Energy balance error compensated extended-state Kalman filter-based model predictive control of a 300 MW boiler-turbine unit

Abstract

Coal-fired power plants are crucial for grid stability due to their dispatchability in response to the intermittency of renewable power generation. However, the frequent and rapid load variations pose great challenges to the flexibility and stability of operational control systems. To this end, this study proposes a novel energy balance error compensated extended-state Kalman filter-model predictive control (EBEC-ESKF-MPC) strategy for boiler-turbine units, inspired by the direct energy balance (DEB) framework. A linearized state-space model serves as the nominal model for MPC design, while an ESKF incorporating energy balance error compensation ensures accurate state estimates. The non-dominated sorting genetic algorithm II is applied for the multi-objective optimization of key weights. The wide-range simulations demonstrate that the proposed strategy achieves a significant improves pressure control performance, outperforming DEB-PI, DEB-active disturbance rejection control (ADRC), and ESKF-MPC by 58.9 %, 31.5 %, and 10.2 % respectively. Additionally, the EBEC-ESKF-MPC strategy achieves the shortest settling time and minimal control deviation under uncertain disturbance scenarios. Moreover, the proposed strategy's ability to reduce total absolute energy deviations by 92.3 %, 78.9 %, and 37.8 % compared to DEB-PI, DEB-ADRC, and ESKF-MPC, respectively, highlighting its superior disturbance rejection, control precision, and energy balance capabilities.