NIOSHTIC-2 Publications Search
Adaptive Learning Networks Applied to Coal Interface Detection and Resin Roof Bolt Bonding Integrity.
Proc 3rd Int'l Conf Innovative Mining Systems 11/2-4/87 Univ of Missouri-Rolla PP 160-174 :160-174
The Bureau of Mines is currently pursuing two possible approaches to the coal interface detection (cid) problem by monitoring mining machine signals (1) via accelerometers mounted directly on the mining machine itself (machine vibration) and (2) via accelerometers affixed to the roof, coal, and floor a distance from the mining machine (in-seam seismic). The Bureau is investigating the effectiveness and utilization of computer-based analysis and discrimination techniques to cid decision-making based on these signals. Both machine vibrational and in-seam seismic data are stored on a multichannel analog tape recorder and later digitized and analyzed using sophisticated signal processing, feature extraction, and adaptive learning network (aln) computer programs. It is this advanced data analysis that distinguishes this approach from those of prior researchers. The aln program is initially trained using a database of features extracted from known signals measured under conditions of interest (e.g., cutting coal, starting to cut roof, starting to cut floor, etc.). Afterwards it uses this database to determine whether new unknown signals (generated by the mining operation) belong to one or more defined "classes" or conditions. Assuming that the aln system correctly classifies these signals, such a system could have the potential for controlling a mining machine cutting drum, such that it always stays in the coal, via a suitable feedback loop.
Proc. 3rd Int'l Conf. Innovative Mining Systems, 11/2-4/87, Univ of Missouri-Rolla, PP. 160-174
Page last reviewed: September 2, 2020
Content source: National Institute for Occupational Safety and Health Education and Information Division