Coordinated model predictive control methods for 1000 MW thermal power units based on multi-objective trade-off mechanism

To address the multi-objective dynamic conflict challenges during deep peak shaving of 1000 MW ultra-supercritical thermal power units under high renewable energy integration, including rapid load response, pressure stability, economic operation, and low emissions, this study proposes a Model Predictive Control (MPC) method based on multi-objective trade-off mechanisms. The approach first constructs a simplified linear predictive model for control purposes to capture key unit dynamic characteristics. A comprehensive quadratic objective function integrating four objectives is then established. A fuzzy logic dynamic weight adjustment strategy is introduced to adaptively adjust control priorities according to operating conditions. Within the MPC framework, a rolling optimization problem incorporating input/output amplitude and rate constraints is designed and solved using an efficient quadratic programming solver. A feedback correction mechanism based on actual output continuously refines model prediction errors. Results demonstrate that the proposed method achieves a load tracking root mean square error (RMSE) of 0.093 and a regulation time of 12.29 s, representing 55.9% and 43.2% reductions compared to traditional MPC methods, respectively. Mean main steam pressure fluctuations are reduced by over 29.8% compared to coordinated MPC. Power supply coal consumption rate decreases to 282.7 g/kWh, with nitrogen oxide (NOx) emission concentration reaching 106.5 mg/m 3 . The method effectively resolves multi-objective dynamic conflicts while significantly enhancing unit flexibility, economic efficiency, and environmental performance. • Proposes a coordinated MPC strategy for 1000 MW ultra-supercritical power units with high renewable integration. • Balances load tracking, steam pressure stability, coal consumption, and NOx emissions via a multi-objective model. • Uses fuzzy logic for dynamic weight adjustment and quadratic programming for constrained optimization. • Achieves fast tracking (RMSE 0.093), stable pressure, lower coal use, and reduced NOx emissions.