基于量子粒子群优化的支持向量机模型反演土力学参数  

Inversion of Soil Mechanical Parameters Using Support Vector Machine Model Optimized by Quantum Particle Swarm

作　　者：刘月华 朱庆闯 毕乃晨 LIU Yuehua;ZHU Qingchuang;BI Naichen(China Construction Eighth Engineering Division Corp.,Ltd.,Jinan 250000,Shandong,China)

机构地区：[1]中建八局第二建设有限公司,山东济南250000

出　　处：《建筑施工》2025年第3期361-365,共5页Building Construction

摘　　要：岩土的力学参数常通过室内或原位试验测得,因受尺寸效应和人为因素影响,有时试验值与真实值之间偏差会很大。为探索更多、更准确的岩土力学参数的确定方法,利用量子粒子群算法优化支持向量机后对岩土物理力学参数进行反演。首先采用正交和均匀试验对需要反演的参数进行设计,然后结合有限差分软件FLAC3D得到学习样本和测试样本,通过量子粒子群优化(QPSO)向量机模型(SVM)建立反演参数与位移间复杂的非线性映射关系。把地铁站点基坑在不同开挖工况下的地表产生的水平及竖向位移的计算值与实测值进行对比分析,以验证该分析理论及方法的合理性。由于不同参数变化能使目标函数值产生不相同的变化结果,故分别采用竖向和水平位移对泥炭质土的压缩模量E_s和泊松比μ进行反演,研究结果表明取竖向位移对E_s进行反演,水平位移对μ进行反演,其结果更合理。此分析方法为岩土参数反演方法提供了一种新的思路。Geotechnical mechanical parameters are typically obtained through laboratory or in-situ tests.However,due to size effects and human factors,significant deviations often occur between test values and actual values.To achieve more accurate determination of geotechnical parameters,the quantum particle swarm optimization(QPSO)algorithm is used to optimize a support vector machine(SVM)model for parameter inversion.Orthogonal and uniform experimental designs are employed for parameter setup,and learning and testing samples are generated using the finite difference software FLAC3D.The QPSO-SVM model establishes a complex nonlinear mapping between inverted parameters and displacements.Horizontal and vertical displacements calculated under various excavation conditions at a subway station site are compared with measured values to validate the proposed method.Variations in parameters lead to distinct changes in the objective function;thus,vertical displacement is used to invert the compression modulus(ES),while horizontal displacement is applied for Poisson's ratio(μ)of peat soil.Results indicate that these approaches yield more reasonable outcomes.This method provides a novel approach for geotechnical parameter inversion.

关 键 词：参数反演 支持向量机(SVM) 泥炭质土 量子粒子群优化(QPSO) 

分 类 号：TU452[建筑科学—岩土工程]