Smart Pick for Rock Cutting

In mining applications, conical-type rock picks are used in machines like continuous miners and long-wall shearers. These picks today are changed based on instincts, and these instincts cost the mining industry in particular an estimated 43 million dollars in lost revenue per year. This is because when one pick wears down, if it isn't replaced in time, adjacent picks are more likely to fail quicker because of uneven wearing, which causes cascading failures. The drum of the mining machine itself can be damaged in catastrophic cases, causing very long downtimes. In a coal mine, a minute of downtime could equate to a thousand dollars lost. Our team implemented an intelligent sensing system that can measure the loads experienced on the pick and the vibrational harmonics to both characterize the rock that is being cut into as well as predict the wear status of each pick. Rock characterization serves two benefits for mine efficiency; first, by cutting only your desired mineral, you save the processors of the raw material lots of time sifting out unvaluable materials. Also, cutting into abrasive, harder rocks can reduce the time each rock pick lasts, meaning more downtime to replace them. Preliminary implementation of a 43-feature neural network proved to be effective in differentiating rock-type coal simulants vs hard rock simulants with 86 percent accuracy using 0.1s windows of data from the smart picks outputs (acceleration and capcitance readings from load cell).