Different Rock Acoustic Emission Frequency Characteristics And Signal Recognition Technology Research

The rock acoustic emission (AE) technology has achieved many valuableachievements on ground pressure monitoring in Mining and geotechnical engineering.However, the monitoring results of distortion or a miscalculation happened occasionallydue to the complexity of the practical engineering conditions. The theoretical researchstill obviously lagged behind the needs of engineering practices, mainly including thedeeper study of AE characteristic in different types of rocks and identification technologytheory in characteristics of noise signals. In this thesis, the AE serials of different rocks(like granulite, granite, limestone, siltstone, etc) were investigated by uniaxialcompression test. The complete stress-strain curves, the AE characteristics and the AEsignals were obtained. The different rock failure processes of AE time domin feature wereanalyzed by using fractal theory. The Fourier transform and the wavelet transformationwere employed to analysis the frequency spectrum characteristics of rock AE signal.Combined with AE characteristics parameters and wavelet energy spectrum coefficient,the rock AE signals and the signal sources of interference were recognized by artificialneural network. The main research achievements are as follows:The fractal dimension of AE sequences reflects the different rock failure processes.The phase space m and scale coefficient K values can be calculated, and the evolution ofthe fractal dimension of AE during failure can be summarized as the “fluctuation�'continuous descending” mode, and the gradual descent of fractal dimension can beregarded as the precursor information to predict the collapse of the rock mass.Distribution characteristics of wavelet energy spectrum coefficient of both the rockAE signals and the interference signal of have been got, using the Fourier transform andthe wavelet transformation.The BP neural network has been designed and trained to high precisely identifymultiple typical AE signal by selecting the AE characteristic parameters and the waveletenergy spectrum coefficient as the characteristic vectors.