Dataset and code for the manuscript "Cluster analysis of rock block shapes: A novel quantitative method for characterising rock mass structures"

作者：Zheqi Tian, Jun Zheng, Ruichen Zhang, Rafael Jimenez, Dongpo Wang, Jiongchao Wang · 发表于：Figshare · 年份：2026 · DOI：10.6084/m9.figshare.31828123 · 研究领域：Algorithm、Geology、Computer science、Data mining、Artificial intelligence、Geometry

This dataset supports the manuscript titled “Cluster analysis of rock block shapes: A novel quantitative method for characterising rock mass structures.” The study develops a quantitative geometric characterisation framework for rock block systems at the system scale , with the aim of providing a more representative description of complex block geometries for site-specific engineering analyses. The proposed method introduces a Shape Similarity Index (SSI) to quantify the geometric similarity between polyhedral rock blocks based on their maximum volumetric overlap ratio. Pairwise SSI values are then used to construct a shape similarity matrix, which serves as the basis for spectral clustering of the rock block system. For each resulting block group, the dominant size category is identified to ensure engineering relevance, and the representative shape , representative size , and representative rock block are subsequently determined.The dataset includes three main components. The first contains the MATLAB source code implementing the proposed clustering-based characterisation method, together with the supporting function files required for geometric calculation, point-in-polyhedron testing, scaling, and representative block determination. The second contains the case-study datasets for an open-pit mine in the United States, including both the structural plane system (DFN) data and the derived rock block system data , such as structural-plane centre points, orie...