This report documents the current status of the U.S. Bureau of Mines ongoing investigation of the use of adaptive signal discrimination (ASD) systems to distinguish between cutting coal and cutting rock. Cutting-tool forces and vibrations were measured in the laboratory using both conical bits and roller cutters in a linear-cutting apparatus for several material samples and two cutting directions. A number of ASD systems consisting of one or more signal classifiers were trained and tested to study how data window size, type of signal feature, and combining (polling) of classifier results influence system performance. The results show that ASD system recognition rates can be improved by increasing data window size, removing air-cutting portions from the signal data, overlapping data windows, and combining (fusing) information at various levels of ASD system operation.

National Institute for Occupational Safety and Health