The study of building model to predict ash fusion temperature

Ash fusion temperature has important influence on boiler combustion, and is an important parameter of coal quality. The prediction abilities for ash fusion temperature of different models (support vector machine model and BP artificial neural network model) are studied in this paper. The compositions of coal ash are employed as inputs of these models, and the ash fusion temperature is used as output. The prediction of single coal and blended coal are studied, and the result shows that the optimized SVM model achieves more accurate prediction. At last, the SVM model is used to calculate ash fusion temperature of blend coal for real combustion test of a 300 MW boiler, and good result is achieved.