Groundwater Hydrochemistry Field Analysis And Discriminating Model Of Water-inrush Source In Pansan Mine Based On Matlab

Discriminating the source of coal mine water-inrush quickly and correctly is the precondition of water control. Matlab is a large software for scientific computing environment and easy to use. With its numerical calculation and simulation analysis power, it is significant for timely and effective prevention and treatment of mine water damage to research on methods of discriminating water-inrush source and to build a scientific and effective discriminating model. This paper chooses the main water-inrush aquifer system in Pansan Mine as the research area, collects and analyzes the groundwater level and hydrogeological data of the past years, studies the groundwater level dynamics, water chemical types, characteristic ions, hydrochemistry field of different aquifers in the mine and discriminating model of water-inrush source (water-inrush source from different aquifers or from different mine sandstone aquifers). Results show:(1) The groundwater in the 4th Cenozoic is a single type of Cl-Na+K; Water-types in the sandstone aquifers are Cl-Na+K, Cl-HCO₃^-Na+K and HCO₃^-Cl-Na+K. Through the contrast map ion concentration, it is found that the 4th Cenozoic water has higher Cl- and SO₄^2- contents than coal water; On the contrary, the coal water has higher HCO₃^- content, and the Ca²⁺ and Mg²⁺ contents between the 4th Cenozoic water and the coal water are not much different. After drawing box charts, classification of principal component analysis and analysis of variance, it is found that HCO₃^- and SO₄^2- play more important roles in identifying the 4th Cenozoic water and coal water.(2) By visual analyzing the spatial distribution of characteristic ions(SO₄^2-, HCO₃^-)and TDS in the main pouring water aquifers, it is found that with the 4th Cenozoic water level decreasing, the concentrations of characteristic ions(SO₄^2-, HCO₃^-)and TDS increase and the total direction of the 4th Cenozoic water flow is from the southeast to the northwest; the concentration of SO₄^2- in the coal water changes complexly, the rule is not obvious; the total direction of HCO₃^- and TDS changing are similar: increasing from the northeast to the southwest in Pansan coalfield; with the coal water level decreasing, the concentrations of HCO₃^- and TDS increase and the total direction of the coal water flow is from the northeast to the southwest.(3)Discussing the application of Bayes Stepwise Discrimination method, Fuzzy Comprehensive method, Degree of Grey Incidence method and BP neural networks in discriminating water-inrush source from two main aquifers (the 4th Cenozoic aquifer and the mine sandstone aquifer). In discrimination accuracy, the determining effect of neural network model (95.65%) is better than the others, followed by Fuzzy Comprehensive method based on the deviation weighing and Degree of Grey Incidence method (82.61%), and Fuzzy Comprehensive method based on biasing weighting is the worst (65.22%).(4) Discussing methods of discriminating water-inrush source from different mine sandstone aquifers, and trying to establish the water inrush discrimination (B8 & C13) model of SVM, it is right to identify 10 unknown water samples. Then comparing it with other methods, the results show the classification function of SVM algorithm has faster operation and higher discrimination accuracy, so it can distinguish the water on the level of B8 & C13 quickly and effectively.(5) The groundwater levels in the 4th Cenozoic aquifer in Pansan Mine is affected by precipitation and change with season. The dynamic changes of the groundwater levels in the mine sandstone aquifer in the Pansan Mine need to be further studied.