基于集成学习算法的土石混合体斜坡稳定性预测模型研究

Research on Slope Stability Prediction Model of Soil-Rock Mixture Based on Integrated Learning Algorithm

由于土石混合体具有显著的非均质性,导致土石混合体斜坡的稳定性难以预测。为此,基于49个土石混合体斜坡实例样本,选择含石率、基覆面倾角、坡高和坡角4个数据特征作为输入参数,将斜坡稳定性系数作为预测对象,采用Boosting、Bagging、Stacking 3种集成学习算法将各个基学习器的预测结果合并后输入线性回归模型,构建了斜坡稳定性预测模型,并且对比分析了3种算法模型在优化前和优化后的预测结果。结果表明:在3种算法模型中,Boosting算法模型的预测精度相对最高;在通过果蝇优化算法优化后,3种算法模型的预测精度都得到显著提升,而Boosting算法模型仍具有最高的预测精度,FOA-Boosting的R2值接近1。

That the significant heterogeneity of soil-rock mixture make the stability of soil-rock mixture slope is difficult to predict.Therefore,based on 49 slope samples of soil-rock mixture,four data characteristics of rock content,base overburden inclination,slope height and slope angle were selected as input parameters.The slope stability coefficient was prediction object.The prediction results of each base learner were combined by three integrated learning algorithms of Boosting,Bagging and Stacking and input into the linear regression model to construct a slope stability prediction model.The prediction results of the three algorithm models before and after optimization were compared and analyzed.The results show that the prediction accuracy of the Boosting algorithm model is relatively the highest;After the fruit fly optimization algorithm,the prediction accuracy of the three algorithm models have been significantly improved,while the Boosting algorithm model still has the highest prediction accuracy,and the R2 value of FOA-Boosting is close to 1.