Abstract: The morphology of gravel particles is a critical factor influencing the mechanical behavior of conglomerates. The existing shape parameter systems are characterized by an excessive number of indices and overlapping classification criteria，which limit the applicability of classification methods. To address these issues，this study proposes a novel gravel particle shape classification method that integrates multi-pa⁃ rameter dimensionality reduction with machine learning. Initially，three-dimensional reconstruction data of gravel particles are obtained through CT scanning technology，and shape parameters including form，round⁃ ness，and sphericity are calculated using Python programming. Subsequently，Pearson correlation coeffi⁃ cient heatmaps are employed to quantify parameter correlations，and representative shape parameters are se⁃ lected through hierarchical clustering combined with principal component analysis. Finally，the K-Means clustering method is implemented，the optimal number of clusters is determined by the elbow method and silhouette coefficient，resulting in the classification of gravel particle shapes into three or four categories， and the classification method is simplified. The proposed classification method provides valuable insights for analyzing the mechanical properties of conglomerates and optimizing resource development efficiency. Keywords gravel particle morphology，principal component analysis，K-Means clustering，machine learning，classification methods