Boiler Oxygen Optimization Based on Double-hidden-layer BP Neural Network and Improved PSO Algorithm

摘要

Oxygen content is an important parameter of air supply system control for a large coal-fired boiler unit. The amount of oxygen is closely related to the ratio of air to coal, which directly reflects whether the air is enough to ensure complete combustion of pulverized coal in the combustion process, and whether the set value is reasonable has a direct influence on the efficiency of the boiler. Therefore, it is of great significance to optimize the set value of boiler oxygen content. Based on the historical operation data of a 1000MW thermal power unit, a double-hidden layer BP neural network model with the main parameters of the unit and the oxygen content of flue gas as input and the boiler efficiency as output is established, and a simplified particle swarm optimization algorithm with high efficiency was adopted. With the boiler efficiency as the optimization objective, the oxygen content of the boiler was optimized in order to obtain the optimal oxygen setting value. The experimental results show that the efficiency of the optimized boiler is obviously improved. This method can predict the optimal value of oxygen content in different working conditions efficiently and accurately, and provide guidance for the optimal operation of the boiler.