【摘要】 首先针对秸秆、污泥和中药渣与贫煤掺烧时的NO和SO₂排放特性进行了研究，然后对掺混样品灰的熔融特性进行了分析，进而结合机器学习算法完成了灰软化温度预测。结果表明：单独燃烧时，贫煤的NO和SO₂的排放曲线呈双峰结构，完全排放时间较长；生物质燃烧时NO和SO₂的排放曲线呈单峰结构，排放时间短；而生物质的掺混使得混合物NO和SO₂的转化率较贫煤单独燃烧时有所降低。贫煤和中药渣的结渣风险较低，秸秆和污泥的结渣风险较高。秸秆对混合物软化温度的影响较小，结渣风险小；中药渣的掺混会使混合物的软化温度显著下降，结渣风险增大。在所选算法中，决策树回归模型对灰的软化温度预测较为准确。更多还原

【Abstract】 The NO and SO₂ emission characteristics of straw, sludge and Chinese medicine slag mixed with lean coal were studied. Then, the ash fusion characteristics of the blended samples were analyzed, and the softening temperature was predicted by machine learning algorithm. The results show that when the raw material is burned alone, the emission curves of NO and SO₂ in lean coal show a bimodal structure, and the complete discharge time is longer. However, the emission curves of NO and SO₂ in biomass combustion shows a single peak structure and the emission time is short. The conversion rate of NO and SO₂ in the mixture was reduced by biomass mixing compared with that of lean coal combustion alone. The slagging risk of lean coal and Chinese medicine slag is low, while the slagging risk of straw and sludge is high. The straw has little effect on the softening temperature of the mixture, the slagging risk is low, while the mixing of Chinese medicine residue will significantly decrease the softening temperature of the mixture, the risk of slagging is increased. Among the selected algorithms, decision tree regression model is more accurate in predicting the softening temperature of ash.更多还原