Feature extraction and classification of mining microseismic waveforms via multi-channels analysis.

In this paper, the author presented a strategy for classifying local multi-channels MS waveform, triggering by a single event. There were three steps to achieve the goal. Firstly, based on STA/LTA method, the first arrival and terminated time had been picked up, using the MS signal preprocessed; secondly, the author extracted the waveform features, time-frequency (L,f), amplitude (A), statistics of amplitude distribution (AD) and threshold algorithm (TS), and correlation coefficient (R); thirdly, before establishing an effective judgment mechanism, this method employed a hierarchical recognition framework with 3 layers, which integrated preliminary judgment, combined recognition and optimization judgment. This method was validated through analyzing a coal mine visual event in Shandong Province, and the result shows that it is successfully used to classify electromagnetic interference wave, background noise and MS events. The result can basically meet the requirements of classification accuracy. [ABSTRACT FROM AUTHOR]