【摘要】 针对基坑支护结构位移与土层参数具有小样本及非线性的特征，提出一种以模拟退火（SA）算法与粒子群（PSO）算法混合优化最小二乘支持向量机（LSSVM）的位移反分析方法。一是通过均匀试验构造学习与测试样本，运用SA-PSO混合算法对最小二乘支持向量机进行参数寻优，寻找模型最优参数组合，并建立最小二乘支持向量机非线性回归模型；二是构建预测位移与实测位移间的目标函数，运用SA-PSO混合算法迭代寻优基坑土层参数值。应用于昆明某实际基坑工程中，反演结果表明此方法具有一定的可行性。更多还原

【Abstract】 For the characteristics of small samples and nonlinearity of the displacement and soil parameters of the foundation support structure,a displacement inverse analysis method of least squares support vector machine(LSSVM)is proposed with the hybrid simulated annealing(SA)algorithm and particle swarm(PSO)algorithm. The first is to construct learning and testing samples through uniform experiments,apply the SA-PSO hybrid algorithm to the least squares support vector machine for parameter optimization,find the optimal combination of model parameters,and establish the least squares support vector machine nonlinear regression model; the second is to construct the objective function between predicted and measured displacements,and apply the SA-PSO hybrid algorithm to iteratively optimize the parameter values of the foundation pit soil layer. Applied to a practical foundation pit project in Kunming,the inversion results show that this method has certain feasibility.更多还原