Application of principal component-Fisher discrimination model in grade prediction of coal and gas outburst

Abstract

In view of problems of complicated calculation process, strong subjectivity and low accuracy in existing prediction methods of coal and gas outburst, a principal component-Fisher discriminant model was constructed and applied to the prediction of coal and gas outburst grade in a coal mine. Based on analysis of gas factors, coal structure and geological structure, the factors that affect coal and gas outburst of the coal mine included gas pressure, gas content and initial velocity of gas release and so on were obtained. On the basis of 23 groups measured data of coal and gas outburst of the coal mine, firstly, the principal component analysis model was used to do dimension reduction of influencing factors of the mine coal and gas outburst, 5 principal components with high index correlation were extracted. Then the 5 principal components were input into Fisher discriminant model, and the grade of coal and gas outburst of samples was predicted according to discriminant function. The application results show that the principal component-Fisher discriminant model has high credibility, and can accurately predict coal and gas outburst grade, the training sample accuracy is 100%, the predicted results of the tested sample are also consistent with the actual situation of coal and gas outburst of the coal mine, misjudgment rate of 0, which provides a new method of accurate prediction of coal and gas outburst.

Keywords

• coal and gas outburst
• principal component-fisher model
• grade prediction
• cumulative contribution rate
• clustering
• discriminatio