Pattern Recognition Of Mine Blasts And Microseismic Events Based On Waveform Features

Abstract:This paper is focused on pattern recognition of mine blasts and microseismic events based on waveform features. Signal databases of mine blasts and microseismic events are established through calibration blasting and manual identification. Based on these two databases, seven waveform characteristic parameters were extracted. By applying the Fisher linear discriminant method to the waveform characteristic parameters extracted, a pattern recognition model that could recognize more than98%microseismic events is established. Specific findings and conclusions are as follows:(1) The characteristics of the noise signals including the drilling, orepass and operations are statistical analyzed.(2) By tracking the position and the recording time, an accurate blast signal database is established. Spectral characteristics of the waveform characteristics of the database within the blasts are analyzed. Through the FFT transform, the frequency of such blast wave which less than60～200Hz is determined.(3) According to the conclusions (1) and conclusion (2), a microseismic event signal database is built through the frequency distribution validation, the ES/EP validation and the Brune mode validation.(4) By comparing the starting-up features of the two types of signals, seven characteristic parameters that can identify microseismic events and blasts-the first peak, the first peak vibration velocity, the slope of the first peak starting-up trend line and the maximum peak, the maximum peak vibration velocity, the slope of maximum peak starting-up trend line and correlation coefficient-ware extracted:.(5) By combining the parameters of the waveform feature with Fisher linear discriminant analysis, a mathematical model that able to identify more than98%microseismic events is established.