Identification of mine water sources using a multi-dimensional ion-causative nonlinear algorithmic model

Area of Science:

• Hydrogeology
• Geochemistry
• Data Science

Background:

• Mine water inrushes pose significant risks in underground mining operations.
• Accurate identification of water source types is crucial for effective risk management and prevention strategies.
• Existing methods for water source discrimination may lack the precision required for complex hydrogeological conditions.

Purpose of the Study:

• To develop and validate a robust model for discriminating mine water sources based on ionic composition.
• To establish a rapid and accurate method for identifying different types of water inrush.
• To provide a tool that supports decision-making in water damage prevention and control within mining areas.

Main Methods:

• Application of nonlinear algorithmic theory to establish the R-SVM (Resilient Support Vector Machine) model.
• Selection of six key ionic components (Na+, Ca2+, Mg2+, Cl-, SO42-, HCO3-) as discrimination factors based on water sample analysis.
• Utilizing SPSS statistics and MATLAB for model development and discriminant analysis with training and prediction samples.

Main Results:

• The R-SVM model successfully classified mine water sources into four categories: goaf water, Ordovician carbonate water, and two types of sandstone fracture water.
• The model achieved a high classification accuracy of 90.90% in the case study of Zhaogezhuang Coal Mine.
• The developed model demonstrated strong applicability and discriminant ability for mine water source identification.

Conclusions:

• The R-SVM model provides an accurate and efficient method for discriminating mine water sources.
• This approach offers significant guiding value for water damage prevention and control in mining engineering.
• The study highlights the potential of data-driven models in addressing complex hydrogeological challenges in mining.