摘要
针对大坝变形监测序列的多尺度、年周期性和非平稳趋势性特征,采用最小二乘与广义回归神经网络的组合模型法进行预报,同时考虑到大坝变形与水位因子的相关性,在模型中引入水位数据进行大坝变形的预报。通过实验表明,加入水位数据的组合预报模型相比单独采用大坝变形监测数据的模型,预报精度有显著的提高。
Aiming to the multi-scale,annual periodicity and non-stationary trend of dam deformation monitoring data,we used the LS+GRNN model to predict the dam deformation.Meanwhile,considering that the dam deformation monitoring data were tightly correlated with the water-level data,we introduced the water-level data into the LS+GRNN model to further improve the accuracy of prediction.By comparing the prediction accuracy of dam deformation monitoring data by LS+GRNN model with water-level data and LS+GRNN model,we concluded that the LS+GRNN model with water-level data could improve the prediction accuracy.
出处
《地理空间信息》
2020年第6期103-105,I0003,共4页
Geospatial Information
基金
国家自然科学基金资助项目(U1531128)。