摘要
湿陷系数是描述黄土物理力学性质的重要指标,利用现有资料建立黄土湿陷性预测模型具有重要的工程应用价值。通过粒子群算法(PSO)优化支持向量机(SVM),建立基于PSO-SVM的黄土湿陷性预测模型,SVM用于描述湿陷系数与黄土物理力学指标间的非线性关系,PSO对SVM参数进行全局寻优,避免人为选择参数的盲目性,从而提高模型的预测精度。结果表明PSO-SVM方法在精度和适用性方面由于传统的人工神经网络方法,建立的黄土湿陷性预测模型可以满足工程应用需求。
The collapsibility coefficient is an important index to describe the physical and mechanical properties of loess.Particle swarm optimization(PSO)was used to optimize support vector machine(SVM),and a prediction model of loess collapsibility was established based on PSO–SVM.SVM was used to describe the nonlinear relationship between the collapsibility coefficient and loess physical and mechanical indexes.PSO was used to optimize the parameters of SVM globally to avoid the blindness of artificial parameter selection,so as to improve the prediction accuracy of the model.The results show that because of the traditional artificial neural network method,the prediction model of loess collapsibility established by PSO–SVM method can meet the requirements of engineering application in terms of accuracy and applicability.
作者
熊汝全
刘凌军
刘国辉
陆海燕
王阳
Xiong Ru-quan;Liu Ling-jun;Liu Guo-hui;Lu Hai-yan;Wang Yang
出处
《建筑技术开发》
2023年第1期61-63,共3页
Building Technology Development
关键词
湿陷系数
粒子群算法
支持向量机
预测模型
coefficient of collapsibility
particle swarm optimization
support vector machine
predictive model
