To predict the flow of ships arriving at anchorage., this paper proposes an advanced prediction framework named as EEMD-edRVFL-RF., which integrates the ensemble empirical mode decomposition (EEMD) method with ensemble deep random vector functional link network using random forest (edRVFL-RF) model. The subsequences decomposed from EEMD and original data are input into the edRVFL-RF model for prediction. In the EEMD-edRVFL-RF., a moving window strategy-based EEMD method is firstly utilized to decompose the original ship traffic flow sequence, which helps to reduce the data complexity and avoid data leakage. Then., K-fold cross-validation and grid search algorithm are introduced to determine the hyperparameters of the deep random vector functional connection network layer by layer. Afterwards, a RF model is inserted to define optimal weight of each output layer prediction in the network structure. Finally, the validity of the proposed framework is verified by using ship flow data of anchor at Three Gorges-Gezhou Dams Hub. The results demonstrate that the accuracy of the proposed framework is better than other classical ship flow prediction models. The prediction results can provide data support for Three Gorges-Gezhou Dams navigation authority to manage ship navigation.
Ship Traffic Flow Prediction Using EEMD-edRVFL-RF
07.08.2024
1280400 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
A modified unbiased GM(1,1) prediction model based EEMD
IEEE | 2016
|Train wheel size prediction based on EEMD-LSTM model
British Library Conference Proceedings | 2022
|Regional ship traffic flow prediction method and system
Europäisches Patentamt | 2023
|Short-Term Passenger Flow Prediction for Urban Rail Based on Improved EEMD-Ensemble Learning
Springer Verlag | 2024
|