Application of BP Neural Network in Prediction of Flow Temperature of Ash from Blended Coals

Abstract

Ash flow temperature is an important indicator of blended coals gasification, determined by its chemical composition. In view of nonlinear relationship between chemical compositions and flow temperatures of ashes from blended coals and individual coals, BP neural networks were built to forecast chemical compositions and flow temperatures of ashes. It was shown that relative errors for predicting contents of oxides, such as silicon oxide, alumina, iron oxide, calcium oxide, magnesium oxide, potassium oxide and sodium oxide, in ash from blended coals with BP neural network range from 0.01% to 4.14%, relative errors for predicting flow temperature of ash from blended coals by BP neural network based on prediction results of ash chemical composition locate in range of 0.20%-3.48%, illustrating good agreements with the measured flow temperatures. BP neural network can offer a shortcut for coal blending technology.