Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Short-Term Forecasting for Bohai Sea Ice Management Based on Machine Learning Theory
In the Bohai, sea ice drift may lead to severe vibration and fatigue failure of structures. Therefore, accurate and rapid sea ice forecasting is critical to ensure the safety of offshore structures. In this study, a short-term forecast model of regional sea ice is presented by combining field monitoring, environmental forecasting, and Elman neural network methods. Data for the winters of 2013–2014, 2017–2018, and 2018–2019 were adopted to train the sea ice model. Forecasts of sea ice parameters, such as the ice thickness, speed, and direction, were provided for offshore flexible structures. The results show that the forecast accuracy of the Elman network model decreases with an increase in forecasting period, while the actual ice thickness exhibits a positive correlation with the forecast accuracy. In comparison with wavelet neural network methods, the Elman network model attains a significantly lower demand on the time length of training data. Furthermore, the Elman network model achieves a higher accuracy in regional sea ice forecasting than does the hybrid Lagrangian–Eulerian model.
Short-Term Forecasting for Bohai Sea Ice Management Based on Machine Learning Theory
In the Bohai, sea ice drift may lead to severe vibration and fatigue failure of structures. Therefore, accurate and rapid sea ice forecasting is critical to ensure the safety of offshore structures. In this study, a short-term forecast model of regional sea ice is presented by combining field monitoring, environmental forecasting, and Elman neural network methods. Data for the winters of 2013–2014, 2017–2018, and 2018–2019 were adopted to train the sea ice model. Forecasts of sea ice parameters, such as the ice thickness, speed, and direction, were provided for offshore flexible structures. The results show that the forecast accuracy of the Elman network model decreases with an increase in forecasting period, while the actual ice thickness exhibits a positive correlation with the forecast accuracy. In comparison with wavelet neural network methods, the Elman network model attains a significantly lower demand on the time length of training data. Furthermore, the Elman network model achieves a higher accuracy in regional sea ice forecasting than does the hybrid Lagrangian–Eulerian model.
Short-Term Forecasting for Bohai Sea Ice Management Based on Machine Learning Theory
Yu, Songsong (Autor:in) / Zhang, Dayong (Autor:in) / Li, Siyin (Autor:in) / Lv, Qixin (Autor:in) / Yue, Qianjin (Autor:in) / Li, Gang (Autor:in)
27.08.2021
Aufsatz (Zeitschrift)
Elektronische Ressource
Unbekannt
Modeling and Forecasting of Bohai Sea Ice
| British Library Online Contents | 2000
A self-learning short-term traffic forecasting system
| Online Contents | 2012
TECHNICAL PAPERS - Modeling and Forecasting of Bohai Sea Ice
| Online Contents | 2000
Short-term water demand forecasting using hybrid supervised and unsupervised machine learning model
| Springer Verlag | 2020