中文摘要：

为寻求岩石临界破坏判据和前兆特征,在粗砂岩单轴压缩声发射（AE）试验的基础上,研究了岩石破坏过程中AE信号频段占比随应力变化特征,重点分析高、低两个特征频段占比随应力变化规律,同时对两个特征频段中不同应力水平下AE幅值关联维数进行计算与分析,并建立了基于频段占比与应力间关系的多频段AE信号主频识别判据模型。研究表明：AE信号频段占比的分布特征能较好地诠释岩石破坏所经历的主要过程;岩石破坏过程中,较低频段AE信号（31.25～46.875 k Hz）占比先减小后增大,较高频段AE信号（140.625～156.25 k Hz）占比先增大后减小。在临界破坏状态下,高、低两个特征频段占比分别出现最大值和最小值,且二者中AE幅值关联维数都下降到最低。通过对特征频段占比与应力之间的耦合分析,利用特征频段占比、AE幅值关联维数的变化可更准确地对岩石临界破坏前兆进行判别和预测。

英文摘要：

In order to obtain the criteria and precursor characteristics of critical rock fracture, laboratory experiments on the characteristics of acoustic emission（AE） of gritstone specimens under uniaxial compression are carried out. The relationships between proportions of AE frequency bands and stresses in the process of rock failure are analyzed. The variation characteristics of proportions of two characteristic frequency bands with stresses are analyzed especially. The relevant fractal dimensions of AE amplitude at different stress levels are calculated and analyzed in the two frequency bands. A multi-band AE dominant frequency recognition criterion model is established based on the quantitative relationships between proportions of frequency bands and stresses. The results indicate that the distribution characteristics of the proportions of frequency bands can reflect the main failure stages of rock. In the process failure of rock, the proportions of AE lower frequency band signals（31.25～46.875 k Hz） decrease firstly and then increase, and the proportions of AE higher frequency band signals（140.625～156.25 k Hz） increase firstly and then decrease. The minimum value and the maximum value appear in the critical state of rupture in the lower and higher frequency bands. And the minimum values of the relevant fractal dimensions of AE amplitude appear in the critical state of rupture in the two frequency bands. Based on the coupling analysis of proportions of characteristic frequency bands and stresses, the accuracy for estimating and predicting the critical state of rock can be improved by using the variations of the proportions of characteristic frequency bands and the relevant fractal dimensions of AE amplitude.