Discriminator of Mining Blasts and Microseismic Events Based on Multi-Scale Discrete Wavelet Transform

The event database, based on manual identification, was set up to find discriminating features in seismograms for purpose of auto-classifying mining blasts and microseismic events. Since the signals of rock rupture and blast are characterized by time-dependence, non-stationary, unpredictability and transience, the Wavelet Transform was proposed to extract the recognition features. The logarithms of the maximum absolute value in each decomposition layer's coefficients were selected as the characteristic parameters as the coefficients with the maximum absolute value not only have the greatest energy in the frequency domain but also play the biggest role in the process of signal reconstruction. A mathematical model, making it possible for accurately discriminating 98.59% mining blasts and microseismic events, was established by use of Fisher discriminant analysis. The causes of misclassified cases were also investigated.