Automatic Identification System for Rock Microseismic Signals Based on Signal Eigenvalues

The microseismic signals of rock fractures indicate that the rock mass in a particular area is changing slowly, and the microseismic signals of rock blasting indicate that the rock mass in a particular area is changing violently. It is of great significance to accurately distinguish rock fracture signals and rock microseismic signals for analyzing the changes in the rock mass in the area where the signal occurs. Considering the microseismic signals of the Dahongshan Iron Mine, the time domain, frequency domain, energy characteristic distribution, and fractal features of each signal were analyzed after noise reduction of the original signal. The results demonstrate that the signal duration and maximum amplitude of the signal could not accurately distinguish the two types of signals. However, the main frequency of the rock fracture signal after noise reduction is distributed above 500 HZ, and the main frequency of the rock blasting signal is mainly distributed below 500 HZ. After the denoised signal is decomposed by the ensemble empirical simulation decomposition, the energy of the IMF1 frequency band of the rock fracture signal occupies an absolute dominant position, and the sum of the energy of the IMF2–IMF4 frequency bands of the rock blasting signal occupies a dominant position. The fractal box dimension of the rock fracture signal is mainly below 1.1, and the fractal box dimension of the rock blasting signal is mainly above 1.25. According to the above research results, an automatic signal recognition system based on the BP neural network is established, and the recognition accuracy of the rock blasting and rock fracture signals reached 93% and 94% respectively, when this system was used.