Experimental Study on the Differences in Acoustic Emission Precursor Between Rockburst and Brittle Failure Based on the Data Mining Method

Rockburst and brittle failure are two common failure forms in deep-rock engineering. In this study, comparative experiments were conducted to understand the differences in the failure process and precursor characteristics of rockburst and brittle failure. The experimental results revealed a severe dynamic failure process during rockburst and a static failure feature during brittle failure. Additionally, it was revealed that the violent debris ejection during rockburst is due to the instantaneous instability of the plate-like rock structure driven by excess energy, whereas brittle failure begins with rapid main shear crack penetration. Additionally, we designed a self-organizing mapping neural network array to extract acoustic emission (AE) precursor waveforms for rockburst and brittle failure. A unique feature of this method is the continuous AE signal pointing set with the highest contribution rate. The extraction results indicated that the AE precursor waveforms exhibit shape characteristics with an uncountable number of envelope peaks. The parameter distribution feature highlights the disparity in the dominant frequency range, which has been proven to be a manifestation of AE precursor waveform differences between rockburst and brittle failure. Based on these differences, AE precursor waveforms can be categorized into two types: Types I and II, both with long duration, high amplitude, low average frequency, and high absolute energy, but with high and low main frequencies, respectively. Specifically, Type-I behavior is recommended as precursor signals for brittle failure, whereas Type-II AE waveforms are recommended as precursor signals for rockburst. Finally, considering the impending mechanical response to the peak stress, the precursor waveform was largely driven by the formation of the main fracture near the failure point.