Abstract: A large number of non-rock failure signals are usually obtained by microseismic monitoring. At present, these kinds of signals are mainly identified and filtered by manual experience, which consumes much precious time and seriously affects the efficiency of disaster prevention and rescue. By analysing massive microseismic signals, we find that the STA/LTA algorithm can roughly characterise the amplitude and frequency of waveforms after the real-time triggering of signals. The R values of the rock failure signals and the non-rock failure signals are different in the delayed position. Therefore, a real-time recognition algorithm for the microseismic signal of rock failure is proposed. The new algorithm is applied to three projects, including the underground powerhouse of the Baihetan hydropower station, the deep stope of the Hongtoushan copper mine, and the deep stope of the Ashele copper mine. The accuracy rates of the identification of rock failure events are 85.98%, 92.45% and 91.06%, respectively. The accuracy rates of the filtering of non-rock failure events are 72.06%, 83.11% and 49.87%, respectively. It is of great significance that the algorithm makes it possible to automatically analyse and predict rock engineering disasters based on microseismic information of rock failure.

Key words: real time recognition algorithm, microseismic monitoring, STA/LTA, rock failure signal, rock burst, engineering application