基于 BP 神经网络的燃煤机组 NOx 排放浓度预测系统

摘要

The prediction model of NOx discharge concentration is proposed by establish the BP neural network to unit from coal-fired power plant,to explore the feasibility of BP neural network system for the pol utant concen-tration prediction.Through the train for cur ent time different unit load,stock imported smoke temperature,con-centration of NOx and O2 import,export NOx concentration of certain power plant 660 MW unit,it is concluded that the training model,after training the BP network to forecast the unknown concentration of NOx emission,predic-tion accuracy above 93 .48%.That method can meet the actual forecast of demand completely.Conclusion:the BP neural network system of coal-fired power plant NOx emission concentrations of real-time prediction is technical feasible,can monitor the quality of NOx emission concentration effectively.%建立了BP神经网络对燃煤电厂机组NOx排放浓度的预测模型，探索BP神经网络系统对污染物排放浓度预测的可行性。通过对某电厂660MW机组实时的不同机组负荷、脱销进口烟温、进口NOx 浓度、进口O2浓度、出口NOx 浓度的样本进行训练，得出训练模型。通过训练后的BP神经网络模型对未知的NOx排放浓度进行预测，预测精度达到93．48％以上，完全能够满足实际中的预测需求。结论：BP神经网络系统对燃煤电厂实时的NOx 排放浓度的预测具有技术上的可行性，能有效的对NOx 排放浓度进行监测。