Extended-state Kalman filter-based model predictive control and energy-saving performance analysis of a coal-fired power plant

Faced with the high penetration of renewable energy sources, coal-fired power plants (CFPPs) are confronted with flexible requirements and the challenge of energy-saving transformation. To this end, this study introduces an extended-state Kalman filter (ESKF)-based model predictive control (MPC) strategy to optimize control and energy-saving in a CFPP. A nonlinear model is developed and its accuracy is validated using field data. The modelling uncertainties and unquantifiable disturbances beyond the nominal are modeled as a constrained extended state, estimated by ESKF. The MPC design is crafted to incorporate ESKF's joint estimation, effectively addressing uncertainties and constraints. In addition, a dynamic exergy method is developed to introduce energy-saving performance indicators. Case simulation results demonstrate ESKF-MPC's substantial improvement over the field controller in the IAE index for power and pressure control by 45.3 % and 59.3 %. Additionally, ESKF-MPC enhances the average exergy efficiency by 1.8 % across various loads, highlighting its energy-saving capabilities. Under parameter perturbation, the ESKF-MPC significantly outperforms the PID controller and KF-MPC in dual-loop control, achieving an 81.0 % and 85.2 % improvement in the IAE index. Simulation of varying ramp scenarios reveal that higher ramp rates worsen energy-saving performance during load ascents, but unexpectedly improve it during load reductions, reflecting the asynchronous trend of control and energy-saving performance.