基于特征选取与SVR优化的软土路基弹性模量预测模型  

Prediction Model of Elastic Modulus of Soft Soil Subgrade Based on Feature Selection and SVR Optimization

作　　者：肖乾 袁颖 XIAO Qian;YUAN Ying(Anhui Technical College of Industry and Economy,Hefei 230051,China;Hebei GEO University,Shijiazhuang 050031,China)

机构地区：[1]安徽工业经济职业技术学院,安徽合肥230051 [2]河北地质大学城市地质与工程学院,河北石家庄050031

出　　处：《河北地质大学学报》2025年第1期69-77,共9页Journal of Hebei Geo University

基　　金：安徽省高等学校自然科学项目(2021ZK02);安徽省高等学校哲学社会科学项目(2022AH052659)。

摘　　要：文章提出基于岩土参数选取与支持向量回归(SVR)模型参数优化的软土路基弹性模量预测方法,以简化求解弹性模量的过程。设计一种大步长粗搜、小步长精搜的改进型粒子群算法(PSO),用来优化支持向量回归弹性模量预测模型的参数,寻找全局最优解。研究表明:根据确定系数、均方误差、绝对误差可知,相较于其他模型,改进的PSO-SVR模型更容易跳出局部最优解,具有较高的准确性、稳定性、鲁棒性。Based on the prediction method of the parameters and the support vector regression(SVR)model,the elastic modulus.A modified particle swarm algorithm(PSO)is designed for the long and long intensive search to optimize the parameters of the elastic modulus prediction model of support vector regression to find the global optimal solution.The study shows that according to the correlation coefficient,mean squared error and absolute error,compared with other models,PSO-SVR model is easier to jump out of the local optimal solution,and has high accuracy,stability and robustness.

关 键 词：特征选取 PSO SVR 弹性模量 

分 类 号：P642.11[天文地球—工程地质学]