Research On Singular Spectrum Analysis And Its Allpication In Rock Damage Detection

Natural disasters and construction safety accidents often occur in China from which Rock burst is one of the typical dynamic disasters that seriously affect the safety of underground engineering.Therefore,the extraction of precursory wave from rock damage AE signal and the study of frequency shift are important for revealing rock damage and preventing rock burst disaster.Up to now,the intern change in the process of rock damage can be detected by the AE sensor.Therefore,lots of researchers did research on the rock failure using rock damage AE signals.The main algorithms include spectrum analysis,time-frequency analysis and wavelet analysis.The most important finding is that the phenomenon,from high frequency to low frequency,occurring during rock failure can be found by the Fourier spectrum.However,it is so hard to quantify the frequency shift phenomenon for the rock damage signal,because it is very complex,which comprise lots of noise and non-periodic components.Thus,previous research about this issue can only find that there are frequency shift existed in the process of rock damage.Aiming at this problem,a singular spectrum algorithm(SSA)based on g statistics is proposed to extract the frequency characteristics of the signal,so as to realize the quantitative calculation of the frequency shift phenomenon,which will further study the frequency shift phenomenon of the rock failure process.By analyzing the time domain waveforms corresponding to the extracted frequency bands,the stress wave characteristics of precursory damage information with different rock damage are summarized.These rock damage precursors are important for predicting rockburst disasters.The traditional SSA need to select the important eigenvalues manually,which make the traditional SSA is very difficult to extract important eigenvalues from long signal.Because,long signal means it has more eigenvalues that need to be selected.Aiming at this problem,this paper proposes a new algorithm for frequency-frequency feature extraction based on SSA and Sl0 algorithm.This algorithm can accurately extract the main frequency band of the acoustic emission signal and remove the interference bands.Compared with traditional SSA,the modified algorithm can be adaptive to find the exact frequencies of the location,making this algorithm can be used for longer rock damage acoustic emission signal.In addition,this paper uses the proposed algorithm to reveal the main frequency shift characteristics from the process of unloading to rockburst,thus deepening the understanding of rockburst.