Abstract: Since there are multiple peaks and waveform intensive problems by using direct cross-correlation for locating the rock acoustic emission (AE) sources, no solution occurs in the solution process and the positioning accuracy is insufficient. In this study, an algorithm of AE source localisation was proposed to solve these problems, based on all-phase FFT analysis and multiple cross-correlation. The experiments were carried out on granite specimens with the size of 50×100×50 mm3. The wavelet transform was used to pre-process the spectrum of cross-correlation. Meanwhile, the phase invariance of all-phase FFT was also applied to find the waveform peak and obtain the time difference. Finally, the initial point of positioning was acquired by the least square method, and then the final positioning coordinates were obtained through the Geiger iteration. Eight measuring points were selected and monitored during the experiments. Results indicate that this method is easier to find the wave peak than the direct cross-correlation, and can obtain clear waveforms. The selection of initial points can effectively meet the strict requirement on the initial value of the Geiger iterative method. It is found that the obtained results are more accurate than those by the direct cross-correlation. The mean absolute error of this study was 2.40 mm lower, compared with Geiger positioning result by the PCI–2 AE system from the United States. The result was 3.31 mm lower, compared with the result of AE source localisation based on multiple cross-correlation and Geiger algorithm. The result was 1.74 mm lower, compared with time-delay positioning result by the all-phase phase difference method for AE sources.

Key words: acoustic emission (AE), wavelet transform, all-phase FFT, multiple cross-correlation, Geiger, source localisation