Study on recognition of mine water sources based on statistical analysis

Abstract

Models for recognizing the source of underground mine water is the basis for real-time monitoring of the water source in active underground mines. Real-time monitoring of the water source is of great significance for predicting water inrush and controlling flooding disasters. This study used the hydrochemistry data of the main aquifers of the Ningtiaota Coal Mine, Jiaozuo Coal Mine, Xieyi Coal Mine, and Huacheng Coal Mine. Through contrast experiments, the influence of dimension reduction processing of discriminant indexes and the selection of discriminant analysis methods in the recognition of the ability of the mine water source model were analyzed for different sample sizes for hydrochemistry data. The research shows that when the sample size is small, the dimension reduction processing using stepwise discriminant analysis and principal component analysis can significantly improve the model’s ability to recognize the mine water source. The recognition accuracy of the model using the distance discriminant method is significantly lower than that of the other two methods, and the recognition ability of the Bayes model is slightly better than that of the Fisher model. When the sample size is large enough, the dimension reduction processing can no longer improve the recognition ability of the model, and all three methods have the same recognition ability. This study can provide an important reference for the establishment and optimization of mine water source recognition, and it is expected to be significant for the prediction and prevention of coal mine water inrush events.