A Waveform Image Method for Discriminating Micro-Seismic Events and Blasts in Underground Mines.

Area of Science:

• Geophysics
• Mining Engineering
• Data Science

Background:

• Accurate discrimination between micro-seismic events and blasts is crucial for understanding underground mine seismicity.
• Conventional methods face limitations, necessitating advanced techniques for reliable classification.

Purpose of the Study:

• To develop and evaluate a waveform image discriminant method for classifying micro-seismic events and blasts.
• To improve the efficiency and accuracy of micro-seismicity monitoring in underground mines.

Main Methods:

• Utilized Principal Component Analysis (PCA) on waveform images of seismic data to extract key features.
• Employed Support Vector Machine (SVM) algorithm for binary classification based on PCA-derived features.
• Introduced Computation Efficiency per Accuracy (CEA) to assess the impact of feature contribution rate.

Main Results:

• Achieved optimal classification accuracies of 97.1% for events, 93.8% for blasts, and 93.60% overall.
• The quality evaluation parameter MCC reached 0.8723.
• Determined an optimal contribution rate of 0.90 for balancing accuracy and computational efficiency.

Conclusions:

• The waveform image discriminant method offers an automated and effective solution for classifying seismic events in mines.
• This method facilitates the creation of high-quality micro-seismic databases for subsequent seismicity analysis.