基于神经网络的NO_x燃煤锅炉排放预测及优化

Neural network based prediction and optimization of NO_x emission from coal-fired boilers

应用Matlab神经网络工具箱对某燃煤电站锅炉NOx排放特性进行神经网络建模。仿真结果表明,该模型具有良好的准确性和泛化能力,模型平均相对误差为1.37%,具有较高的准确性。基于该NOx排放预测模型,结合遗传算法对燃煤锅炉的NOx排放进行优化,按照优化结果推荐的运行参数,在相同的运行负荷工况下,其NOx排放浓度由优化前的456.2mg/m3降为323.9mg/m3,下降幅度达到了29%,效果显著。

The Matlab neural network toolbox was applied to establish the prediction model for NOx emis- sion in a coal-fired power plant boiler. The simulation results show that this forecast model has well accura- cy and generalization ability,its average relative error is 1.37 %, which means the high accuracy. On the basis of this NOx emission prediction model,the NOx emission was optimized by genetic algorithm. According to the recommended operating parameters, after the optimization, the NOx emission decreased from 456.2 mg/m^3 to 323.9 mg/m^3 under the same operation load conditions,reduced by 29 % ,which is dramatic.