The Bayes Recognition Model for Mine Water Inrush Source Based on Multiple Logistic Regression Analysis

摘要

Accurate recognition of water inrush sources is important in mine water hazard control. In this study, 118 water samples of four water types in the Qinan coal mine were analysed by multiple logistic regression, and the 12 water samples that did not meet the requirements were removed. The remaining 106 were used as training samples to establish a Bayes recognition model (BRM). In addition, the BRM was used to complement multiple logistic regression analysis (MLRA) to discriminate other water samples in the mining area. The recognition accuracy of the combined model was 95.28%. The results from the model were consistent with the field water samples and showed that the combined MLRA-BRM approach fully considers the hydraulic relationships between different aquifers and a mixed water inrush source. Moreover, the MLRA-BRM combination improved water inrush source recognition accuracy and was more reliable than using MLRA or BRM alone.