Automated Slope Failure Prediction in Surface Mines: Integrating Onset Detection with Time-to-Failure Forecasting

Abstract

Slope failures in surface mining environments present substantial hazards to human safety, infrastructure, and operational continuity. Accurate and timely prediction of such events is essential for effective risk mitigation and early warning. This study introduces a fully automated framework that integrates real-time detection of the onset of critical acceleration with time-to-failure forecasting, leveraging high-resolution displacement monitoring data. The proposed approach employs a multi-stage algorithm to objectively identify the transition from stable or creep behaviour to rapid acceleration, minimizing subjectivity and enabling prompt intervention. Upon detecting this transition, the system applies inverse velocity analysis to predict the impending failure time, continuously refining its forecasts as new data become available. Validation using both real-time datasets from Indian coal mines and historical records from Australian sites demonstrates the framework’s adaptability across diverse geological conditions. By combining automated onset detection with dynamic failure prediction, this methodology significantly enhances early warning capabilities and supports proactive slope management in surface mining operations.