岩体质量分类的PCA-RF模型及应用

为了更合理地确定岩体质量类别，将主成分分析（PCA）与随机森林（RF）算法相结合，提出一种岩体质量分类的PCA-RF模型。选取能够充分反映岩体质量类别的5项指标进行分析，运用主成分分析法对各指标进行相关性处理，依据方差累计贡献率得出3个主成分，从而消除指标间的相关性，减少模型输入。然后采用随机森林模型对岩体质量进行分类，选用现场20组数据作为训练样本、10组数据作为测试样本，利用交叉验证的方法估计泛化误差。结果表明，该方法分类结果与实际结果较吻合，平均准确率达96.7%，同时得出岩体质量所处类别的概率分布，进一步反映岩体质量的复杂度，为工程建设提供更详细的参考依据。

PCA-RF Model for the Classification of Rock Mass Quality and Its Application

In order to determine the classification of rock mass quality more reasonably，PCA-RF classification model of rock mass quality was proposed which combined with principal component analysis and random forest algorithm.Five classification indexes were chosen which can fully reflect the rock mass quality category.The correlation analysis of indexes was calculated by principal component analysis，and three principal components were abtained by accumulated variance devoted rate，which can eliminate the correlation between each index and reduce the inputs of model.Then，the classification of rock mass quality was determined by random forest model.Twenty sets field data were chosen as training samples，and ten sets field data were chosen as testing samples.The generalization errors were estimated by cross-validation method.The results show that the classification results satisfyingly agree with the actual results at the average accuracy of 96.7%，and the probability distribution of classifications that can reflect the complexity of rock mass quality was calculated simultaneously，which can provide more detailed reference for engineering construction.