大型电站锅炉混煤掺烧的NOx排放特性预测与运行优化

在某700 MW四角切圆燃煤电站锅炉的NOx排放特性及锅炉效率多工况热态测试的基础上，应用人工神经网络建立大型四角切圆电站锅炉NOx特性及锅炉效率模型并进行预测，检验样本NOx排放浓度和锅炉效率预测值与实测值的平均相对误差分别为3.63%和0.23%，证实模型的可行性。并在此基础上，结合遗传算法对锅炉运行参数进行优化。优化后NOx排放浓度为421.44、255.05和215.40 mg×m-3，分别降低了37.56%、29.43%和30.56%；锅炉效率为94.56%、94.13%和94.80%，分别提高了0.09%、0.42%和0.88%。该模型可在掺烧非设计煤种情况下寻找出最优运行参数，降低锅炉NOx排放浓度并提高锅炉效率；同时随掺混比的增大，NOx排放浓度降低；掺烧D磨和E磨有利于降低NOx。

Prediction and Operation Optimization for NOx Emission Property of Large-scale Mixed Coal-fired Utility Boiler

The NOx emission property and boiler efficiency of a 700MW utility tangentially firing coal burned boiler were experimentally investigated, an artificial neural network model on NOx emission property and boiler efficiency of large-scale boiler were developed to predict the NOx emission, and the predicted result indicates the mean relative error of NOx emission and boiler efficiency is 3.63% and 0.23% between experimental value and the calculated value, respectively, which proves the feasibility of the model. Using genetic algorithm based on the neural network model to optimize the operation parameters, and the optimization results of NOx emission are 421.44, 255.05 and 215.40 mg×m-3, which are reduced by 37.56%, 29.43% and 30.56%, respectively, and the relevant boiler efficiency are 94.56%, 94.13% and 94.80%, which are improved by 0.09%, 0.42% and 0.88%. The result shows the algorithm can get the optimal operating parameters, decrease the NOx emission and improve boiler efficiency when using the blending non-design coal; more the mixing ratio, less the NOx emission; Blending D and E coal mill is helpful to reduce NOx.