Abstract: The change in b-value of rock acoustic emission is an important precursors of rock fracture. However, b-value is affected by artificial settings such as sample size, start amplitude and amplitude interval. It is of great significance to quantify the error of b-value. In this paper, the maximum likelihood method and the least squares method are employed to calculate the b-value of the rock fracture acoustic emission amplitude catalogues composed of random numbers with different sample sizes, start amplitudes and amplitude intervals. The results distributed within the error are recorded as valid data, and the error of b-value is analyzed according to the proportion of valid data. The result is verified using the acoustic emission amplitude catalog of indoor triaxial compression test data. For maximum likelihood method, the valid data proportion reduces as the sample size increases, while augments as the start amplitude increases. For least squares method, as sample size increases, valid data proportion first increases and then declines. The most valid data is obtained when the sample size is around 2000. Given the same amplitude interval, as the start amplitude increases, the corresponding optimal amplitude interval reduces from 22 dB to 18 dB. This study provides a conducive reference for the reasonable calculation of b-values of rock fracture acoustic emission.

Key words: b-value of acoustic emission, error analysis, least squares method, maximum likelihood method, valid data proportion, Monte-Carlo method