A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Dam deformation prediction method and system based on characteristic decomposition and deep learning
The invention discloses a dam deformation prediction method based on characteristic decomposition and deep learning, and relates to the technical field of dam deformation prediction, and the method comprises the steps: decomposing dam deformation sequences of a source domain and a target domain into subsequences; respectively evaluating the subsequences, and extracting feature components of the subsequences to obtain feature components of a source domain and a target domain; performing prediction model training by using the source domain feature component matched with the target domain feature component; freezing a shallow network of the prediction model, inputting target domain feature component data to perform retraining and parameter fine tuning on the frozen prediction model, and obtaining a target domain feature component prediction model; and a real-time deformation prediction result is obtained by using the target domain feature component prediction model. And predicting the insufficiently observed concrete dam by referring to deformation data of other similar dams. Potential rules of characteristic components are effectively explored, prediction precision and generalization ability are improved under the condition of insufficient observation, and therefore accurate dam deformation prediction is made.
本发明公开了一种基于特征分解和深度学习的大坝变形预测方法，涉及大坝变形预测技术领域，包括：将源域和目标域的大坝变形序列均分解为子序列；分别对子序列进行评估，并提取子序列的特征分量，得到源域和目标域特征分量；利用与目标域特征分量相匹配的源域特征分量进行预测模型训练；冻结预测模型的浅层网络，输入目标域特征分量数据对冻结后的预测模型进行再训练微调参数，得到目标域特征分量预测模型；并利用目标域特征分量预测模型得到实时变形预测结果。通过借鉴其他类似大坝的变形数据，对观测不足的混凝土大坝进行预测。有效地探索特征成分的潜在规律，在观测不足的情况下提高预测精度和泛化能力，从而做出准确的大坝变形预测。
Dam deformation prediction method and system based on characteristic decomposition and deep learning
The invention discloses a dam deformation prediction method based on characteristic decomposition and deep learning, and relates to the technical field of dam deformation prediction, and the method comprises the steps: decomposing dam deformation sequences of a source domain and a target domain into subsequences; respectively evaluating the subsequences, and extracting feature components of the subsequences to obtain feature components of a source domain and a target domain; performing prediction model training by using the source domain feature component matched with the target domain feature component; freezing a shallow network of the prediction model, inputting target domain feature component data to perform retraining and parameter fine tuning on the frozen prediction model, and obtaining a target domain feature component prediction model; and a real-time deformation prediction result is obtained by using the target domain feature component prediction model. And predicting the insufficiently observed concrete dam by referring to deformation data of other similar dams. Potential rules of characteristic components are effectively explored, prediction precision and generalization ability are improved under the condition of insufficient observation, and therefore accurate dam deformation prediction is made.
本发明公开了一种基于特征分解和深度学习的大坝变形预测方法，涉及大坝变形预测技术领域，包括：将源域和目标域的大坝变形序列均分解为子序列；分别对子序列进行评估，并提取子序列的特征分量，得到源域和目标域特征分量；利用与目标域特征分量相匹配的源域特征分量进行预测模型训练；冻结预测模型的浅层网络，输入目标域特征分量数据对冻结后的预测模型进行再训练微调参数，得到目标域特征分量预测模型；并利用目标域特征分量预测模型得到实时变形预测结果。通过借鉴其他类似大坝的变形数据，对观测不足的混凝土大坝进行预测。有效地探索特征成分的潜在规律，在观测不足的情况下提高预测精度和泛化能力，从而做出准确的大坝变形预测。
Dam deformation prediction method and system based on characteristic decomposition and deep learning
一种基于特征分解和深度学习的大坝变形预测方法及系统
CHEN XUDONG (author) / HU SHAOWEI (author) / CHEN ZEHUA (author) / LI LIUYANG (author) / SUN WENHAO (author) / GUO JINJUN (author) / QIN XIANGNAN (author)
2024-04-26
Patent
Electronic Resource
Chinese
A River Water Quality Prediction Method Based on Dual Signal Decomposition and Deep Learning
| DOAJ | 2024
Research Progress of Machine Learning in Deep Foundation Pit Deformation Prediction
| DOAJ | 2025
| DOAJ | 2025