Analysis of the AE signal parameters during granite compression process based on the K-means clustering method

Abstract

The average frequency (AF), rise-time divided by amplitude (RA), and peak frequency of acoustic emission (AE) generated during the granite uniaxial and triaxial compression failure process are taken as the main factors of the type of sample crack. Further, AE signals can be characterized based on the K-means clustering method. Results show that there is a large difference between the AE signals generated by uniaxial and triaxial compression. The former has more diversity in the category, e.g., better clustering effect and uniaxial or triaxial compression. AE signal parameters can be divided into four categories based on the K-means clustering method. Acoustic signals A, B, C, D in the two compression modes are substantially similar and uniaxial compression mainly generates tensile crack before the peak stress, and shear crack appears after the peak stress. During the triaxial compression process, tensile and shear cracks are both generated before the peak stress, but shear crack is no longer present after the peak stress.