线性回归与神经网络组合模型实现变形预测

Deformation prediction based on combination of linear regression and neural network

传统变形监测单一的变形分析模型,对于复杂施工状况的监测数据,变形预测精度不高,不能充分满足工程施工变形预测的要求。文中研究建立了一种线性回归分析模型和非线性的BP神经网络模型相结合的组合预测模型,并将其应用于太湖隧道围堰变形监测数据分析,计算给出了采用经典权和IOWGA算子两种不同的定权方式建立组合预测模型的效果,对预测精度进行了计算比较。结果表明,采用IOWGA算子建立组合模型的预测精度优于单一模型的预测精度,也优于经典权的组合模型的预测精度。

The accuracy of deformation prediction of traditional deformation monitoring single deformation analysis model for the monitoring data of complex construction condition is not high,and it cannot sufficiently meet the requirements of engineering construction deformation prediction.In this paper,a combination forecasting model of linear regression analysis model and nonlinear BP neural network model was studied and established,and it was applied to the deformation monitoring data analyzing for the cofferdam of Taihu tunnel.The results of using traditional weighting and IOWGA operator to establish the combined forecasting model were given.Their forecasting precision was calculated and compared.The results show that the combination model of IOWGA had a better precision than single model.It was also over the traditional weighting combination model.