Modeling of boiler efficiency and NOx emission based on asymmetric PSO-BP

In the case of modern multiple power generation modes parallel, thermal power often needs to cooperate with new energy power generation for peak shaving operation, which leads to frequent load changes in thermal power plants, and it is necessary to establish new boiler models quickly and accurately. In this paper, an asymmetric PSO-BP neural network modeling method is proposed to solve this problem. According to the boiler mechanism principle, the non directly related network connection key is removed, and the network structure under different working conditions is unified, and the particle swarm optimization algorithm is adopted to optimize it. Compared with the traditional neural network, the asymmetric PSO-BP neural network model has smaller prediction error and higher stability, which can improve the accuracy of boiler modeling, and lay a foundation for the boiler online modeling when the load changes rapidly.