Application research of multiclass support vector machines in coal mine safety evaluation

Given the situation that there are many mine subsystems with varied impact factors under complex environment,and the difficulty to obtain a large number of coal samples,the support vector machine theory was introduced into the evaluation mechanism.The SVM have a higher processing capacity for nonlinear or small sample problems than others.On the basis of summarizing the principle of SVM multi-classification from 1-a-r to 1-a-1 to Decision tree model,establish a multi-level model for coal-mine safety evaluation in this paper.Introduce weighting factor and samples weighting factor by extracting characteristic parameters of factors that affect coal mine safety,solving the imbalance in the number of training samples and the data type of interference caused by the wrong sub-problems,to achieve a higher accuracy rate for coal mine safety and efficiency evaluation.