Abstract：In order to obtain an accurate and reliable prediction model for caking index of blended coal, two different coal mines in Hebei province were used as research objects. Prediction models of the caking index based on the coal petrographic blending utilizing nonlinear regression analysis were established. The macerals were effectively separated by the gravity separation method. The coal petrographic blending was carried out in low density products. The contents of different macerals were measured by a polarizing microscope. The results show that the molecular structure of the active component is moderate and contains a large number of active structures, which determines the caking index. The inert component forms the core and influences the caking index. With the increasing of the active component content and the decreasing of the inert component content, the caking index gradually increases. The S model, inverse model and progressive regression model based on the ratio between the reactive and the inert component had a higher fitting effect and accuracy. The regression model had the highest fitting degree, but the structure was complex. The inverse model had a simple structure and a good reliability, which could be used to predict the caking index of blended coal.
王成勇，门东坡，陈鹏，李子文. 基于煤岩学的黏结指数非线性预测模型[J]. 钢铁, 2019, 54(9): 22-26.
WANG Chengyong, MEN Dongpo, CHEN Peng, LI Ziwen. Nonlinear prediction model for caking index based on coal petrology. Iron and Steel, 2019, 54(9): 22-26.