Simulation-Driven Prediction of Operator Behaviour to Improve Safety and Productivity in Opencast Mining

The rapid expansion of opencast metal mining operations has led to increased reliance on Heavy Earth Moving Machinery (HEMM), thereby intensifying safety risks and operational challenges. Accident investigations consistently identify unsafe operator behaviour and inadequate skill levels as major contributors to excavation and haulage-related incidents. This study presents a structured, simulation-driven framework designed to improve safety and productivity through behavioural assessment and predictive analytics. High-fidelity mining simulators were utilised to replicate realistic excavation scenarios and capture performance indicators such as overspeed events, braking irregularities, load deviations, and error frequency. Derived behavioural metrics, including Improvement Ratio (IR), Error Velocity (EV), and Residual Risk Index (RRI), were integrated with operational parameters to develop a Real-Time Prediction Model (RTPM) using supervised machine learning techniques. The model was validated using confusion matrix analysis, Receiver Operating Characteristic (ROC) curve evaluation, and performance comparison metrics to ensure predictive reliability and classification stability. Major Findings: The findings demonstrated that integrating simulator-derived behavioural analytics with predictive modelling enables early identification of unsafe trends, supports targeted retraining, and contributes to improved operational stability. The proposed framework offers a data-driven decision-support approach for enhancing safety management and productivity in opencast mining environments.

View article page

Download PDF