长沙理工大学学报（自然科学版）2025，Vol.22Issue(2)：110-119,10.DOI:10.19951/j.cnki.1672-9331.20250107001
基于远区气象站和BO-LSTM模型的大跨桥梁温度预测
Prediction of long-span bridge temperature based on remote meteorological stations and BO-LSTM model
摘要
Abstract
[Purposes]Meteorological data have the advantages of strong stability,high reliability,and long-term availability,so predicting the temperature field of bridges based on the forecast meteorological temperature holds significant engineering importance.[Methods]A long short-term memory(LSTM)model based on Bayesian optimization(BO)was established.By conducting deep training on the BO-LSTM model using meteorological temperature data and measured bridge temperatures,the bridge temperature history can be accurately predicted by inputting forecast meteorological temperatures.[Findings]Based on the health monitoring system of Hedong Bridge in Guangdong Province and the data from a remote meteorological station,the coefficient of determination between the bridge temperature history predicted by the proposed method and the measured bridge temperature history reaches 0.961,whereas that between the meteorological temperature and the measured temperature is only 0.874.Therefore,the established BO-LSTM prediction model enhances the correlation between meteorological temperature and bridge temperature through data mining and learning.[Conclusions]This paper achieves a relatively accurate prediction of the temperature history of long-span bridges based on meteorological temperature,providing an accurate and reliable intelligent prediction method to address issues such as data loss and drift caused by sensor malfunctions,power outages,and other factors.关键词
桥梁智能运维/温度预测/气象数据/长短期记忆网络模型/贝叶斯优化Key words
intelligent operation and maintenance of bridge/temperature prediction/meteorological data/long short-term memory model/Bayesian optimization分类
交通运输引用本文复制引用
彭日旭,黄友钦,饶瑞..基于远区气象站和BO-LSTM模型的大跨桥梁温度预测[J].长沙理工大学学报（自然科学版）,2025,22(2):110-119,10.基金项目
国家自然科学基金(51925802) （51925802）
广州市教育局项目(2024312337) Project(51925802)supported by the National Natural Science Foundation of China （2024312337）
Project(2024312337)supported by the Education Bureau of Guangzhou Municipality （2024312337）
