摘要
黄土高填方场地的工后沉降预测结果,是后续地面工程规划布局及建设时机确定的重要依据。针对黄土高填方场地初期获得的工后沉降数据历时较短,采用传统单项模型方法预测效果较差的不足,提出了基于主成分分析(PCA)的工后沉降组合预测方法。该方法的基本思路是:首先对各单项模型的预测结果进行主成分分析求出主成分,接着采用最优模型选择准则(AIC)确定用于建模的主成分数量,其次建立沉降实测值(因变量)与所选取主成分(自变量)之间的多元回归预测模型,然后对模型预测值与实测值进行了比较,评价组合模型的预测效果,最后采用所建立的组合模型向后多步预测。实例检验结果表明,组合模型预测精度明显优于各单项模型,对于预测黄土高填方场地的工后沉降具有参考价值。
Accurate prediction of post-construction settlement of loess deep filled ground is of great reference to the planning layout and construction timing of subsequent ground projects.However,due to the short duration of the postconstruction settlement data obtained at the beginning of the project,the prediction effect of using a single model is poor,and the combined prediction method of post-construction settlement based on the main component analysis(PCA)is proposed.The basic idea of this method is as follows.Firstly,the main component analysis of the prediction results of each individual model is carried out to find out the main component,then the optimal model selection criterion(AIC)is used to determine the number of the main components used for modeling,followed by the establishment of a multi-regression prediction model between the sequencing measured values(caused variables)and the selected main components(arguments),and then the model prediction values and measured values are compared,the prediction effect of the combined model is evaluated,and finally the combined model is used to predict the future settlement.The example test results show that the prediction accuracy of the combined model is obviously better than that of each individual model,and it has a high reference value for predicting the settlement after the loess deep filled ground.
作者
于永堂
郑建国
黄鑫
YU Yongtang;ZHENG Jianguo;HUANG Xin(Shaanxi Key Laboratory of Engineering Behavior and Foundation Treatment for Special Soil,China Jikan Institute of Engineering Investigation and Design Co.,Ltd.,Xi'an,Shaanxi 710043,China;College of Civil Engineering,Xi'an University of Architecture and Technology,Xi'an,Shaanxi 710055,China)
出处
《水利与建筑工程学报》
2021年第3期117-123,共7页
Journal of Water Resources and Architectural Engineering
基金
国家自然科学基金项目(41790442)
陕西省“三秦学者”创新团队支持计划资助(2013KCT-13)
陕西省技术创新引导专项(基金)计划项目(2020CGHJ-002)。
关键词
黄土高填方
工后沉降
组合预测
预测模型
principal component analysis
loess deep filled ground
combination prediction
prediction model
