基于多元线性回归的渡槽变形预测模型研究

Analysis of the Prediction of MLR-Based Monitoring Model for Aqueduct Deformation

提出了一种基于多元线性回归的渡槽沉降变形健康监测模型,介绍了应用MLR建立变形预测模型的步骤,并与多元逐步回归(MSR)进行对比.针对预测样本数目的不同取值,研究其在实际工程中的预测能力.通过对浙江天台红旗渡槽的监测数据研究,验证了基于MLR的渡槽变形预测模型的预测精度满足工程要求,预测能力略优于MSR,且短期预测能力优于长期预测能力.因此,合理选择预报天数对于渡槽变形预测尤为重要.

In this paper,a health monitoring model of aqueduct deformation based on multiple linear regressions is proposed.The steps of using MLR to establish deformation prediction model are introduced and compared with multiple stepwise regression(MSR).According to the different values of the number of prediction samples,the prediction ability in practical engineering is studied.Through the study of monitoring data of Hongqi aqueduct in Tiantai,Zhejiang Province,it is verified that the prediction accuracy of MLRbased prediction model for aqueduct deformation meets the engineering requirements,and the prediction ability is slightly better than MSR,and the short-term prediction ability is better than the long-term prediction ability.Therefore,it is particularly important to choose the reasonable forecast time for the prediction of aqueduct deformation.