Low-Profile Capacitive Load Cells for Underground Mining Material and Wear Classification to Promote Worker Safety

Abstract

This work proposes a capacitive load cell for conical picks to enable underground continuous mining machine operators to perform their roles away from known hazardous regions near the machine. The load cell is embedded in commercially available flexible printed circuit board, integrates with the target tooling, and demonstrates in situ force sensing of vibration signatures for continuous mining cutting tools. Changes in material constitution, tool mass, and tool geometry cause modal variations in vibrational response measurable with force sensors at the cutting interface. Time-series measurements are captured during rock cutting tests using a linear cutting machine. These measurements are segmented into small windows, less than 0.25 s, and are preprocessed using the fast Fourier transform, which highlights the modal variations. The transformed measurements are then classified into different material and wear categories using support-vector machines with the radial basis function kernel. Different normalization schemes and Fourier transform methods are tested for signal preprocessing. Results show that the power spectral density measurements with normally distributed coefficients give good results for material classification, while the normalized time-domain measurements give better results for wear classification. Under laboratory conditions, this technique is shown to classify material and wear categories with F1 score above 0.85 out of 1.0 for our experiment. This technology could be used to assist operators in assessing material and wear conditions from a safer distance. It has applications in the coal mining industry as well as other applications which use conical picks such as road milling.