Statistical analysis in classification of bituminous coal

Hundreds of samples and 17 variables collected from coalfields of major coal-bearing strata over China except for Tibet and Taiwan, were used in this study. The dry, ash-free basis volatile matter (V ^r) and caking index (G _(RI)) were chosen by means of correlation analysis and stepwise discriminatory analysis as major indices of a new classification. By means of the optimum section, the boundary value of the axis of ordinate (G _(RI)) and axis of abscissas (V ^r) can be determined in the classification system. Thus, aV ^r –G _(RI) classification scheme diagram was formed and bituminous coal was divided into nine classes. Use of correspondence analysis reduced dimensions of sample-expressive space without losing initial information. The trend on the factor surface of samples shows that the classification obtained from correspondence analysis conforms to theV ^r –G _(RI) classification result and further verified the dependability of classification by two indices. At the same time, a certain relationship between the properties of a great variety of coal and their attributes can be explained. Hence, bituminous coal classification becomes more scientific, reasonable, and practical than before.