Rapid identification model of mine water inrush source using random forest optimized by multi-strategy improved...

Area of Science:

• Geosciences
• Mining Engineering
• Artificial Intelligence

Background:

• Mine water inrush accidents pose significant threats to coal mining operations.
• Accurate identification of water inrush sources is crucial for preventing disasters.

Purpose of the Study:

• To develop a fast and accurate method for identifying mine water inrush sources.
• To enhance the performance of the Random Forest algorithm using an improved optimization technique.

Main Methods:

• Kernel Principal Component Analysis (KPCA) was used to extract key factors from mine data.
• An Improved Sparrow Search Algorithm (ISSA) was developed and validated.
• The ISSA was employed to optimize hyperparameters of the Random Forest (RF) model.

Main Results:

• The ISSA-optimized RF model demonstrated superior predictive performance over other models.
• Application to a mine in Shandong province showed high accuracy, precision, recall, and F1 index.
• The proposed method confirmed reliability and stability in identifying water inrush sources.

Conclusions:

• The KPCA-based, ISSA-optimized RF model provides a robust and accurate approach for mine water inrush source identification.
• This method enhances safety by offering a reliable guarantee for mining production.
• The study validates the effectiveness of integrating advanced AI algorithms into mining safety protocols.