The PCA-KD-KNN-based water chemistry identification model of water inrush source type in mine and its application

Abstract

The rapid and accurate identification of water inrush source plays an important role in preventing and controlling mine water inrush disaster. According to the difference in composition and content of the chemical component of groundwater in different aquifers in the mine, the 9 water chemical components of total hardness, Ph value, Na++K+, Ca2+, Mg2+, Cl-, SO4-, HCO3-, and mineralization are selected as the identification indicators of water source type. On this basis, combined with the hydrogeological data of typical coal mines, the principal component analysis (PCA), K-dimension tree (KD), and K-nearest neighbor (KNN) algorithm were used to establish a water source identification model for mine water inrush and compared with the traditional water source identification model. The research results show that the water source identification model based on PCA-KD-KNN reduces the complexity of calculation, overcomes the influence of information overlap between indicators on the identification results, and effectively improves the identification accuracy. The research can provide a certain basis and reference for the identification of water inrush sources in mines.