Real-Time Acoustic Smart Monitoring of Lifter Wear in SAG Mills

Accurate detection of lifter wear in Semi-Autogenous Grinding (SAG) mills is essential for improving operational efficiency, enhancing safety, and reducing costly downtime in intelligent mining operations. Existing monitoring approaches, often based on external vibration or acoustic measurements, face limitations in reliably distinguishing overlapping impacts and are not easily adaptable to real-time predictive maintenance. To address this gap, this study presents a novel real-time acoustic monitoring methodology using high-sensitivity, low-cost audio sensors integrated with an ESP32 microcontroller. Through a laboratory setup replicating authentic metal-to-metal impacts between grinding media and worn lifters retrieved from active mining operations, we captured representative acoustic signals and systematically augmented them with realistic perturbations including Gaussian background noise, overlapping impacts, and motor-induced vibrations to emulate harsh mill environments. Signal analysis was performed using time-domain (Root Mean Square [RMS], energy), frequency-domain (Fast Fourier Transform [FFT]), and time–frequency domain (Short-Time Fourier Transform [STFT], Continuous Wavelet Transform [CWT]) techniques. The results demonstrate that RMS and energy metrics successfully identified more than 90% of impact events even under noisy conditions, while FFT revealed dominant 50 Hz motor interference and broadband noise components. Time–frequency analyses (STFT and CWT) provided high-resolution localization of impact signatures across scales, offering robust indicators of abnormal lifter wear. This work highlights the potential of low-cost IoT-enabled acoustic sensing for predictive maintenance in SAG mills. Beyond technical contributions, the findings carry strong managerial implications by reducing reliance on manual inspections, improving workforce safety, and enabling condition-based maintenance strategies that minimize unplanned downtime.