Abstract: The effect of heat adding rates of blast furnace gas, coke oven gas and coal on flue gas volume was theoretically analyzed in a 300MW boiler co-firing coal with gases. Moreover, the optimization test of NO_x emissions was conducted under the condition of heat adding rates and air staging, and the prediction models of NO_x emissions was established based on the experimental data. The result showed that the theoretical flue gas volume of coal co-fired with gases was equal to that of pure coal when the ratio of blast furnace gas and coke oven gas was 0.3. The co-combustion of blast furnace gas can decrease the difference of furnace temperature, being beneficial to control the NO_x emissions. The NO_x emissions concentration decreased and the unburned carbon in fly ash increased gradually with the increase of heat adding rate of blast furnace gas and the ratio of separated overfire air. By comprehensive measurement of the NO_x emission and carbon content of fly ash, the optimal ratio of heat adding rate between blast furnace gas and coke oven gas was 1.3, and the optimal ratio of separated overfire air was 24%. The comparative analysis of the three models indicated that the GABP model had the highest prediction accuracy and could accurately describe the nonlinear relationship between the input and output parameters of boiler.

Key words: coal combustion, air staging, NO_x emissions, model, neural networks

摘要： 以300MW煤/气混燃锅炉为研究对象，理论计算了高炉煤气、焦炉煤气及煤粉的热量混燃比对烟气量的影响特性，对混燃煤气协同分级配风条件下的NO_x排放特性进行了优化实验研究，利用实验数据建立了NO_x排放浓度的预测模型。结果表明：当高炉煤气与焦炉煤气配比为0.3时产生的理论烟气量与纯然煤工况相当；混燃高炉煤气使得炉膛温度温差减小，有利于控制NO_x排放；随着高炉煤气热量混燃比和分离燃尽风率的增加NO_x排放浓度逐渐降低，而飞灰含碳量逐渐增加。综合衡量NO_x排放浓度和飞灰含碳量，高炉煤气（BFG）与焦炉煤气（COG）的最佳热量混燃比配比应低于1.3，最佳的分离燃尽风率为24%。通过对比所建立的3种NO_x排放预测模型，发现遗传算法优化的BP神经网络模型具有较高的预测精度，能够准确表达锅炉输入参数和输出参数的非线性关系。

关键词: 煤燃烧, 分级配风, NO_x排放, 模型, 神经网络