Impacts of Human Fatigue Impairments on Mine Equipment Operation Performance - SME Annual Meeting 2026

Fatigue is a major contributor to haul-truck incidents in mining, with up to 65% of fatigue-related events associated with operator drowsiness or microsleep episodes, Williamson et al. (2011) and Caterpillar (2007). This study examines whether operator fatigue produces measurable and repeatable signatures in haul-truck telemetry. Eight confirmed fatigue-related disengagement events were analyzed using 20-minute telemetry windows centered around the incident. Statistical descriptors, Z-score normalization, first- and second-order derivatives, sliding-window variability, and cross-signal correlations were extracted from engine speed, throttle position, fuel rate, ground speed, and brake indicators. A logistic regression classifier was trained and evaluated using leave-one-event-out cross-validation. The model achieved a receiver operating characteristic (ROC) area under the curve (AUC) scores of 0.75–0.79, demonstrating consistent discriminative capability. A characteristic shutdown sequence characterized by sudden throttle collapse, fuel-rate decay, RPM reduction, and unbraked coasting was observed across all events, indicating operator disengagement rather than mechanical malfunction. Earlywarning indicators appeared 90–500 seconds before shutdown in the events analyzed, suggesting practical potential for real-time fatigue prediction. These findings provide initial evidence that telemetry-based fatigue detection is feasible and could enhance safety in mining operations.