Acoustic emission (AE)/micro-seismicity (MS) signal obtained in laboratory or in-situ monitoring of the dynamic events is generally contaminated by various noises which result in strong spectral dispersion. Therefore, a precise locating of the main spectral components in conventional power/Fourier spectrum and deeper insights into the time dynamics carried by waveforms of the AE/MS signal would be a very hash work. The present research developed a-state-of-the-art technique for extracting precisely forewarning precursors in the spectral and time regimes from the noise-contained signal based on singular spectrum analysis (SSA). The extracted precursors, known as "eigenfrequency (EF)" and principal component waves are verified in a laboratory rock burst experiment and proved being robust in characterizing time dynamics and the frequency shift phenomenon consistently. The proposed AE/MS signal reduction approach and results have a potential being applied in the practical seismic event forecasting.
Acoustic emission (AE)/micro-seismicity (MS) signal, observed as time series in rock bursts carries considerable information regarding the seismic location, failure mechanisms and the constitutive behavior of the rock materials, providing a window into crack initiation and propagation till macroscopic failure (He et al., 2010). Major problem encountered for analyzing the AE/MS data lies in difficulties in extracting the inherent precursory wave components for spectral and time-dynamic analyses because of the complexity of the rock mass structure and boundary conditions.
Investigations have been carried out on rock burst forecasting based on extracting the spectral and time domain precursors from the AE signal. However, time-domain analyses were rarely seen on the internationally published academic journals as a result of the noise-containing nature of AE/MS signal. For spectral domain analyses, most efforts were made on extracting the main frequency which is the highest peak in the spectrum indicating a significant dynamic event (Michlmayr et al., 2012). Main frequencies will move from high-spectral band to the low band, termed "frequency shift", as the rock was brought to failure such as rock bursts (Cai et al.,2001; 2007; Young et al., 2004; Michlmayr et al., 2012) and earthquakes (Lei and Satoh, 2007). Due to the complexity of the rock structure and boundary, the frequency shift phenomenon does not always hold in the above-reviewed traditional spectral analysis. As the in-situ investigations revealed that the MS events are hybrid close to the monitored outbursts and low frequency components are only the necessary condition other than the sufficient condition for forecasting the onset of the dynamics events (Priestley, 1981; Lu et al., 2012).
