On acoustically determining ore types and particle size in tumbling ball mills

We investigate the correlation between audio noise emissions from a laboratory-sized tumbling ball mill and its operating parameters, with focus on particle size and the type of ore inside the mill. We show that the average particle size is directly linked to the average energy in the signal. Where the mill is loaded with a mixture of ores, we demonstrate that a neural network can be trained to determine the ratio of the ores present