基于天牛须优化算法的相关向量机边坡稳定性分析

Slope Stability Analysis on Relevance Vector Machine Based on Beetle Antennae Search Optimization Algorithm

为了快速、准确地预测边坡稳定性,及时控制边坡危害,提出了一种基于天牛须(beetle antennae search,BAS)优化算法的相关向量机(relevance vector machine,RVM)边坡稳定性分析模型。基于RVM模型,建立了边坡影响因素与稳定性的非线性映射关系;采用BAS算法对RVM模型参数进行优化,提出了基于BAS算法的RVM边坡稳定性分析优化模型;并将该模型应用于京新高速公路的边坡稳定性分析。分析结果表明:与实际值相比,基于BAS-RVM模型的最大绝对值相对误差为3.90%;在相同学习样本下,与RVM模型、支持向量机(support vector machine,SVM)模型和径向基函数(radical basis function,RBF)模型的预测值相比,BAS-RVM模型预测结果的可信度和拟合度更好、精度更高,其平均绝对值误差(mean absolute error,EMA)、均方根误差(root mean square error,ERMS)、相对均方误差(relative root mean square error,ERRMS)远低于其他3种模型。

In order to predict slope stability quickly and accurately as well as control slope damage in time,a slope stability analysis model of relevance vector machine(RVM)based on beetle antennae search(BAS)optimization algorithm was proposed.Based on RVM model,the nonlinear mapping relationship between slope influence factors and stability was established.The parameters of RVM model were optimized by BAS algorithm,and an optimization model of RVM slope stability analysis based on BAS algorithm was proposed.The proposed model was applied to the slope stability analysis of Beijing-Urumqi expressway.The analysis results show that compared with the actual value,the maximum absolute relative error based on the BAS-RVM model is 3.90%.Under the same learning sample,compared with the RVM model,support vector machine(SVM)model and radical basis function(RBF)model,the BAS-RVM model has better reliability and fit,and higher accuracy in predicting results,whose mean absolute error(EMA),root mean square error(ERMS)and relative root mean square error(ERRMS)are much lower than those of other three kinds of models.