Intelligent Optimization of Combustion Process and NOx Emission Control of Power Plant Boiler Based on Deep Learning and Multi-objective Optimization

This study proposes an intelligent optimization and nitrogen oxide (NOx) emission control method for power plant boiler combustion processes by integrating deep learning and multi-objective optimization. While traditional empirical tuning and single-objective algorithms struggle with dynamic, multi-variable combustion environments and lack real-time adaptability and synergistic optimization of efficiency and emissions, this research addresses these gaps by establishing a rolling optimization model that considers load and emissions. By analyzing the relationship between boiler combustion efficiency and nitrogen oxides generation, a rolling optimization model considering load and emission is established. The study analyzes and predicts the operation data and optimizes the combustion strategy in real time by a dynamic multi-objective optimization evolutionary algorithm. Performance evaluation shows that the model achieves high prediction accuracy, with an average absolute error of 2.36×10-5 kW for boiler load, and outperforms existing models in key metrics such as ignition success rate (98.7%) and load adjustment accuracy (3.4 MW). The approach significantly improves combustion efficiency and tightens NOx control, reducing energy waste and improving power plant energy efficiency. These advances demonstrate their effectiveness in improving combustion efficiency, enhancing nitrogen oxide control, and reducing energy waste, providing a powerful solution for operating smart power plants that integrates real-time adaptability and multi-objective synergy, outperforming traditional methods.