摘要
由于复杂工程地质条件和环境因素的综合影响,边坡变形呈现复杂非线性演变特征。针对位移时间序列未能完全考虑环境因素对边坡变形的影响,故将影响边坡变形的有效降雨量加入监测位移时序,组成多因素位移时间序列。引入粒子群算法(PSO)对支持向量机(SVM)的模型参数寻优,结合滚动预测方法,建立了适合边坡变形预测的多因素位移时间序列PSO-SVM模型。以华光潭一级厂房后边坡表面观测位移为例进行预测分析,研究表明,新模型预测结果科学可靠,有效弥补了传统PSO-SVM后期预测泛化能力的不足,提高了模型的预测精度。新模型在边坡位移时序预测中具有一定的工程应用价值。
Due to the comprehensive effect of complex engineering geological conditions and environmental factors,the slope deformation presents complicated nonlinear evolution characteristics. Aim at the displacement time series failing to fully consider the effect of environmental factors on slope deformation,so the effective rainfall which influences slope deformation is involved to monitor displacement time series,the multi-factor displacement time series are constituted. Using the optimization algorithm to support vector machine（ SVM） by leading to particle swarm optimization（ PSO）,combined with rolling forecast algorithm,the multi-factor displacement time series PSO-SVM model is establlished for slope displacement forecast. Taking the slope surface deformation forecast near Huaguangtan hydropower station first-class powerhouse for example,the results show that the forecast by new model is safe and reliable,and it is feasible to make up the generalization ability of the traditional PSO-SVM,and it improves the predictive accuracy. The new model has certain practial value for predicting slope displacement time series.
出处
《勘察科学技术》
2015年第1期1-5,共5页
Site Investigation Science and Technology
关键词
支持向量机
位移时间序列
粒子群算法
边坡变形
多因素
support vector machine（SVM）
displacement time series
particle swarm optimization（PSO）
slope deformation
multi-factor
