基于高斯过程回归的岩体结构面粗糙度系数预测模型  

A prediction model of the joint roughness coefficient based on Gaussian process regression

作　　者：郑可馨 吴益平[1] 李江 苗发盛 柯超 ZHENG Kexin;WU Yiping;LI Jiang;MIAO Fasheng;KE Chao(Faculty of Engineering,China Universityof Geosciences(Wuhan),Wuhan 430074,China;Information Center,Department of Natural Resources of Hubei Province,Wuhan 430071,China)

机构地区：[1]中国地质大学(武汉)工程学院,武汉430074 [2]湖北省自然资源厅信息中心,武汉430071

出　　处：《地质科技通报》2024年第4期252-261,共10页Bulletin of Geological Science and Technology

基　　金：湖北省自然科学基金项目(2023AFB580);贵州省省级科技计划项目(黔科合支撑[2023]一般127);国家自然科学基金项目(42377161,41977244)。

摘　　要：岩体结构面粗糙度系数(JRC)的估算是岩体力学性质评价的重要环节,由于单一统计参数法难以全面表征岩体结构面的复杂粗糙形貌,单一统计参数法建立的JRC计算模型精度较低。选取表征结构面粗糙形态的8种统计参数,结合主成分分析法(PCA)和高斯过程回归(GPR)算法,构建基于多参数融合的JRC预测模型。以公开的112条岩体结构面剖面线数据集(其中95条作为训练样本,17条为验证样本)为例进行分析研究,最后将预测所得JRC与实测值对比并分析预测效果。结果表明:由高斯过程回归构建的JRC预测模型决定系数(R^(2))高达0.972,均方根误差(MSE)为0.517,反映出高斯过程回归方法在小样本条件下构建多统计参数与JRC值隐式关系的适用性,为今后人工智能在JRC指标预测方面实现合理预测提供了思路。[Objective]Estimating the joint roughness coefficient(JRC)is essential for evaluating the mechanical properties of a rock mass.Due to the limitation of a single statistical parameter for characterizing morphology,JRC values estimation by a single statistical parameter may produce a sufficiently unreliable result.[Methods]To address the existing challenges in determining JRC values,a model based on Gaussian process regression(GPR)combined with principal component analysis(PCA)was proposed for the quantitative evaluation of JRC.Notably,eight parameters were selected as indicators for the comprehensive expression of the rock joint roughness.To analyse the model′s performance,a publicly available dataset of 112 rock joint profiles was used as an example,of which 95 were chosen as training samples and 17 were chosen as validation samples.The reliability of the model was verified by comparing the predicted results with the measured JRC values.[Results]The results show that the derived GPR model demonstrates promising performance(R^(2)=0.972,MSE=0.517)for estimation of JRC values,indicating the high applicability of the model in constructing implicit relationships between multiple statistical parameters and JRC values even under small sample conditions.[Conclusion]In general,the GPR model may provide a new way of estimating JRC values with artificial intelligence.

关 键 词：岩体结构面 粗糙度 高斯过程回归 统计参数 预测 

分 类 号：P642[天文地球—工程地质学]