基于灰色关联度的线性组合模型在基坑坡顶位移预测中的应用  

On the Application of the Gray-Correlation-Based Linear Combined Model to Predicting the Horizontal Displacement of the Crest of Foundation Pits

作　　者：唐江明 宋大忠 隋元松[3] 梁进军[1] 

机构地区：[1]解放军理工大学野战工程学院 [2]解放军78170部队 [3]解放军理工大学训练部

出　　处：《国防交通工程与技术》2014年第6期78-80,71,共4页Traffic Engineering and Technology for National Defence

摘　　要：基坑坡顶水平位移预测对基坑安全施工具有重要意义。以昆山市建滔广场花桥一期工程为例,结合BP神经网络模型和时间序列ARMA模型各自特性,建立了基于灰色关联度的线性组合模型预测其坡顶水平位移。实例分析表明组合模型可以在保证较好的预测精度的前提下,提高了预测的整体性,具有工程实用价值。The prediction of the horizontal displacement of the crest of the foundation pit is of great importance for the safe construction of foundation pits.With the first-stage project of the Jiantao Plaza at Huaqiao of Kunshan City as a practical example, and the respective characteristics of the BP neural network model and the ARAM time-series model made use of in a comprehensive way,a grey-correlation-based linear combined model is established in the paper to predict the horizontal displacement of the crest of the foundation pit.The analysis of practical examples proves that the combined model may ensure better accuracy of the prediction with the integrity of the prediction improved.It may be safely concluded that the combined model is of certain practical value in engineering.

关 键 词：坡顶位移 神经网络 时间序列 灰色关联度 

分 类 号：TU196.1[建筑科学—建筑理论] TU753