To extract more in-depth information of acoustic emission (AE) signal-cloud in rock failure under triaxial compression, the spatial correlation of scattering AE events in a granite sample is effectively described by the cube-cluster model. First, the complete connection of the fracture network is regarded as a critical state. Then, according to the Hoshen-Kopelman (HK) algorithm, the real-time estimation of fracture connection is effectively made and a dichotomy between cube size and pore fraction is suggested to solve such a challenge of the one-to-one match between complete connection and cluster size. After, the 3D cube clusters are decomposed into orthogonal layer clusters, which are then transformed into the ellipsoid models. Correspondingly, the anisotropy evolution of fracture network could be visualized by three orthogonal ellipsoids and quantitatively described by aspect ratio. Besides, the other three quantities of centroid axis length, porosity, and fracture angle are analyzed to evaluate the evolution of cube cluster. The result shows the sample dilatancy is strongly correlated to four quantities of aspect ratio, centroid axis length, and porosity as well as fracture angle. Besides, the cube cluster model shows a potential possibility to predict the evolution of fracture angle. So, the cube cluster model provides an in-depth view of spatial correlation to describe the AE signal-cloud.

为了在三轴压缩下的岩石破坏中提取声发射 (AE) 信号云的更深入信息，立方体簇模型有效地描述了花岗岩样品中散射 AE 事件的空间相关性。首先，将裂缝网络的完全连接视为临界状态。然后，根据 Hoshen-Kopelman (HK) 算法，有效地进行了裂缝连接的实时估计，并提出了立方体尺寸和孔隙率之间的二分法，以解决完整的裂缝之间一对一匹配的挑战。连接和集群大小。之后，将 3D 立方体簇分解为正交层簇，然后将其转换为椭球模型。相应地，裂缝网络的各向异性演化可以通过三个正交椭球进行可视化，并通过纵横比进行定量描述。此外，还分析了质心轴长、孔隙度和断裂角这三个量来评价立方体簇的演化。结果表明，样品的剪胀性与纵横比、质心轴长、孔隙率和断裂角四个量密切相关。此外，立方体聚类模型显示了预测断裂角度演化的潜在可能性。因此，立方体聚类模型提供了空间相关性的深入视图来描述 AE 信号云。结果表明，样品的剪胀性与纵横比、质心轴长、孔隙率和断裂角四个量密切相关。此外，立方体聚类模型显示了预测断裂角度演化的潜在可能性。因此，立方体聚类模型提供了空间相关性的深入视图来描述 AE 信号云。结果表明，样品的剪胀性与纵横比、质心轴长度、孔隙率和断裂角四个量密切相关。此外，立方体聚类模型显示了预测断裂角度演化的潜在可能性。因此，立方体聚类模型提供了空间相关性的深入视图来描述 AE 信号云。