Research on the classification model of coal’s bursting liability based on database with large samples

Abstract

In order to evaluate the intensity of the coal’s bursting liability scientifically and accurately, the index relevance problems existing in the current classification methods were focused on, and the Mahalanobis distance discriminant analysis (DDA) method was introduced on the basis of establishing a measured database with the large sample of the coal’s bursting liability. Besides, the DDA model of coal’s bursting liability classification was also established. Then, with the duration of dynamic fracture, elastic strain energy index, bursting energy index, and uniaxial compressive strength selected as evaluation indicators, three grade-discrimination functions of the bursting liability were established through training the data of 95 groups of bursting liability of different coal seams that were collected extensively. After training, the accuracy rate of the DDA model reached 96%. The results of the application of DDA model to the classification of coal samples from 10 coal mines exhibited remarkable agreement with the actual situation, which also solved the difficult problem that the fuzzy comprehensive evaluation method could not distinguish 8 kinds of samples. Its application to the engineering project shows that the classification result of coal’s bursting liability based on the DDA method is both accurate and easy to calculate, and the DDA model has good engineering application prospect.