Details

Autor(en) / Beteiligte

• Zhang, Zujun
• Tian, Zhongchu
• Wu, Zhengqian
• Xu, Binlin

Titel

Development and analysis of a BP-LSTM-Kriging temperature field prediction model for the arch ring section of the reinforced concrete arch bridge

Ist Teil von

• Structures (Oxford), 2024-06, Vol.64, p.106564, Article 106564

Ort / Verlag

Elsevier Ltd

Erscheinungsjahr

2024

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• In this study, a temperature field prediction model based on backpropagation (BP), long short-term memory (LSTM), and kriging interpolation was proposed to predict the temperature field of the entire section of the arch ring of a reinforced concrete arch bridge. A section of the arch foot of the Shuiluo River Bridge, a reinforced concrete arch bridge constructed using the cable-stayed cantilever cast-in-situ method, was considered as the research example. First, the initial sample database was established based on the monitoring results of temperature field measurement points and meteorological data of the arch rib section, and a finite element model was established according to the basic theory of solar radiation to expand the sample database. Second, the combined BP-LSTM network model was established to construct the nonlinear mapping relationship between “structural characteristics, atmospheric environment characteristics, and time characteristics — temperature” to realize the prediction accuracy of the temperature of the section’s measuring and extended sample points. Subsequently, Kriging interpolation was introduced for the unbiased estimation of predicted temperatures at the output measurement points of the combined BP-LSTM network model to increase the temperature prediction data at any position of no measurement points in the section and to realize high-precision fitting of the temperature field at any position in the entire section. Finally, the prediction results were evaluated using the mean absolute percentage error (MAPE) and root mean square error (RMSE) in conjunction with the results of on-site temperature field monitoring. The results showed that the combined BP-LSTM network model has strong reconstruction and generalization abilities for temperature prediction at the measurement points. The proposed model can realize high-precision prediction of the temperature field of the monitoring section of the arch rings. The MAPE in the evaluation of the prediction results was 0.25021, and the RMSE was 0.15323. Hence, we verified that the proposed temperature field prediction model could comprehensively and accurately predict the temperature field distribution at any location in the entire section according to the monitoring results of the limited temperature measurement points in the arch ring section.