Abstract: Aiming at the characteristics of large inertia, large lag and strong coupling of coal-fired boiler systems of thermal power units, a multi-objective cooperative combustion optimization control strategy for thermal power units based on implicit generalized predictive control (IGPC) was proposed. Taking main steam pressure, furnace negative pressure and flue gas oxygen content as the controlled quantities, and taking fuel quantity, induced air volume and air supply volume as control quantities, a multi-input multi-output combustion optimization control system based on IGPC decoupling control was designed. The PID neural network decouples the multi-input multi-output prediction model, uses rolling optimization to optimize the objective function in real time, and introduces an error-based feedback correction algorithm to make the output of each control variable reach the set value. The simulation results show that compared with the conventional PID and IGPC undecoupled control strategies, the IGPC decoupling control strategy can reduce the adjustment time of main steam pressure, flue gas oxygen content and furnace negative pressure by up to 141s, 210s and 162s respectively under the condition of disturbance. The overshoot can be reduced by up to 18.2%, 4% and 9.3% respectively. The engineering application shows that the control deviation of main steam pressure, flue gas oxygen content and furnace negative pressure are lower than ±0.15MPa, 3% and ±0.7Pa respectively; and the fluctuation range of each controlled quantity is small, which has a good control effect.