Classification of Coal Bursting Liability Based on Support Vector Machine and Imbalanced Sample Set

As an inherent property of the accumulation elastic energy and sudden instability failure coal, coal bursting liability (CBL) is basis research on early warning prevention burst. To accurately classify CBL level, support-vector-machine (SVM) method was introduced in this paper, dynamic time (DT), index (WET), impact (KE) uniaxial compressive strength (RC) were selected as classification indexes. An imbalanced sample set, containing 95 groups measured data CBL, established, eight SVM models constructed, based different kernel functions swarm-intelligence-optimization algorithms. Focusing problem imbalance, accuracy, A, F1-score kappa coefficient used to comprehensively evaluate performance models, grey-wolf-optimizer (GWO-SVM) model best reaching highest accuracy 98.9%. The GWO-SVM applied identify level 4# seam Xiaozhuang Coal Mine 1# Wanfeng Mine. results engineering application are consistent with those from field, show that proposed scientific practical, can be a new for classification.