为提高煤灰熔点预测精度，采用支持向量机算法，分别建立了以8种煤灰主要成分为输入量的8输入参数模型以及以8种灰成分加5种煤灰熔融性参数为输入的13输入参数模型，用于预测煤灰软化温度，并利用误差补偿方法对13输入参数模型进行了改进。结果显示，13输入参数模型相比8输入参数模型可以缩小灰熔点预测误差区间，尤其使负误差减小；相比于改进前，改进后的13输入参数模型使均方差降低了59.5%。

In order to improve the prediction accuracy of coal ash melting point,the 8input parameter model with 8 kinds of main coal ash composition as the input and the 13input parameter model with 8 kinds of ash composition and 5 kinds of coal ash melting parameters as the input were established respectively, to predict coal ash softening temperature. And the 13input parameter model was improved by the method of error compensation. The results show that the 13input parameter model can reduce the error range of ash melting point prediction, compared with the 8input parameter model, especially to reduce the negative error. The improved 13input parameter model reduces the mean square error by 59.5%, compared with that before the improvement.

TQ536.4