摘要

The rock failure process is associated with acoustic emission (AE). This paper presented a new waveform fractal dimension algorithm (WFD) to recognise AE source in rock failure process. According to chaotic features of the AE, WFD formula was deduced from BOX dimension. To be more precise for analysing AE waveform, three optimum parameters in WFD formula were calculated using Shuffled Frog Leaping Algorithm (SLFA), where the modified WFD was proposed (WFD-SLFA). The experimental AE data were recorded from the uniaxial and triaxial compression experiments in laboratory that were conducted coal and sandstone samples from Sanhejian mine in China, and then mapped into WFD-SLFA following SVM. Finally, recognition results have been figured out with different feature vectors at two different fracture modes. The results show that the proposed method can effectively recognise the samples in stable and critical instable at two different compression modes. It is a salient way to improve the practical coal rock stability prediction.