含裂隙类软岩单轴抗压强度的机器学习预测

打开文本图片集

关键词：单轴抗压强度；机器学习；软岩；裂隙；弱胶结

中图分类号： TU458+.3 文献标志码：A 文章编号：1003-5168（2025）21-0033-08

DOI： 10.19968/j.cnki.hnkj.1003-5168.2025.21.007

Abstract： [Purposes] To quickly and accurately determine the uniaxial compressive strength （UCS）of weakly cemented rocks with fractures by applying machine learning technology.[Methods] In this study， fivefracture geometry parameters （fracture inclination， width，length，number，and fracture spacing）were selected as input variables， with UCS as the prediction target.A numerical model was established using the PFC method，and a dataset containing 538 data groups was constructed. Using the data from this dataset for training， prediction models for UCS were developed by applying Decision Regresson Tree （DRT）， Multi-layer Perceptron （MLP）， Support Vector Regression （SVR）， and Random Forest （RF）.[Findings] The results show that the SVR model achieves high accuracy，with a root mean square error （RMSE）of 0.315 8，a mean absolute error （MAE）of 0.1825，and a coefficient of determination （R²）of O.9683.Notably，correlation analysis reveals that fracture length and fracture angle are the two most significant parameters influencing UCS prediction.[Conclusions] This study provides a reliable reference for the rapid evaluation of the uniaxial compressive strength of weakly cemented fractured rock.

Keywords： uniaxial compressive strength; machine learning; soft rock; fracture; weak cementation

0 引言

岩石作为一种天然的非均质材料，内部含有不同尺寸和形式的缺陷，如裂隙、节理、空洞和断层等。（剩余9236字）