Prediction of water inrush source of coal seam floor based on Fisher discriminant model

In order to solve the problems of low accuracy mine water inrush source discriminant method for mine floor water inrush source discrimination, taking the second level coal seam of suburban coal mine as an example, the Fisher mine floor water inrush source discriminant model is established. The aquifers with the threat of water inrush in the second level coal seam of suburban coal mine are the coal-measure sandstone aquifer and the karst fractured aquifer of the Taiyuan Formation in the floor. Considering the importance of hydrochemical ions and the validity of the data, three kinds of water quality analysis data of sandstone water, limestone water and mixed water with water inrush threat in the coal seam floor are used as samples. The content and mineralization of six kinds of ions, Ca²⁺, Mg²⁺, Na⁺+K⁺, HCO₃⁻, Cl⁻ and SO₄²⁻, are selected as the discriminant analysis variables for the identification of mine inrush water sources. Two typical Fisher discriminant functions (the first and the second discriminant functions) are obtained by SPSS software. The central values of the typical discriminant functions in the three water quality groups are calculated. By comparing the distance between the function values of the water samples to be discriminated and the central values of the three water quality groups, it is able to determine which group the samples belong to. The back substitution estimation method is used to test the Fisher mine floor water inrush source discriminant model. The results show that the discriminant accuracy rate of the model is 93.3%, and the discriminant results are highly reliable. The model is used to classify 10 known water samples in the second level of suburban coal mine. The results show that the discriminant effect of 10 water samples is consistent with the actual situation, and the discriminant accuracy rate is 100%.