Abstract: The acoustic emission(AE) signals are obtained through uniaxial load tests on sandstones. Performing the fast Fourier transform(FFT) on AE signals, the analysis result shows that the peak frequency of AE signals reduces from 60 kHz to 40 kHz as load process proceeding, and the average peak frequency is 1 kHz lower than that of in natural situation. By analyzing the relationship between energy and acoustic emission events of rock, it is found that the bigger ratio of energy to acoustic emission events means the energy comes from the deeper degree rock fracture, the smaller ratio(order of magnitude <104 ) means that energy is accumulated by some lighter degree rock fractures. The power spectral density (PSD) of AE signals is done with Welch method; it is found that the power spectrum of AE signals can be divided generally into two classes(Type A and Type B), the phase of Type B has obvious discontinuity. Some larger-energy signals are chosen, probability of Type B power spectrum before these larger-energy signals is computed; it is found that the closer to the rupture time, the greater the probability of the type B. The research results provide a new way to analyze the AE properties of the whole rupture procedure of rock, and lay a certain working foundation for the application of acoustic emission in the prediction of rupture and instability.

Key words: acoustic emission signal, dominant frequency, fracture, energy, Welch power spectrum