Huainan Panbei Mine Water Chemical Characteristics Analysis And Discriminating Model Of Water Inrush Source

Water damage has long been one of the major disasters in coal mining and all kinds of mineral deposits,and is also a major factor restricting the sustainable exploitation of mineral resources.The primary task of water damage prevention is to quickly and accurately determine the source of mine water inrush,and then to carry out the next step of water damage prevention and control.The sedimentary environment and water control structure of Panbei minefield largely restricted the groundwater movement in the mining area.The main water filling sources in the mine are old water,coal series sandstone water,Cenozoic lower water bearing and too grey water.Because of the mining damage,the hydraulic connection between the aquifers is more closely,and the hydrogeological conditions of the mine have become more complicated.It is necessary to establish a comprehensive water inrush source discrimination model to quickly identify the source of the mine water inrush and ensure the effective prevention and control of mine water damage.Hydrogeologic data,test data of water sample and related geochemical data of main aquifer in Panbei mining area were collected and analyzed,through the Piper three line diagram,Durov diagram,Stiff diagram to analyze the main mine water filling aquifer hydrochemical characteristics,results show that the hydrochemical characteristics of the similarity of aquifer water is relatively high,difficult to distinguish.Using principal component analysis and the invalid information may cause interference information for subsequent model elimination in water,and with the help of SPSS and Matlab to build a Bayesian discriminant model based on principal component analysis and multinomial Logistic regression model and BP neural network model,the correct rate of cross validation the discriminant model,Logistic regression model to identify the overall number of samples the accuracy rate reaches 79.6%;Bayesian discriminant model of initial verification accuracy is 55.6%;the correct rate is 88.24% BP neural network comprehensive evaluation.The advantages and disadvantages of each discriminant model are compared and analyzed,so as to ensure fast and accurate identification of the aquifer of water inrush source and ensure mine safety production,and provide reliable reference for similar mine's water disaster prevention and control.