Intelligent optimized deep learning hybrid models of neuro wavelet, Fourier Series and Recurrent Kalman Filter for tidal currents constitutions forecasting. (15th December 2020)

• Record Type: Journal Article
• Authors: Aly, Hamed H.H.
• Is Part Of: Ocean engineering. Volume 218(2020)
• Publication Date: 2020-12-15
• Subjects: Tidal currents forecasting -- Renewable energy -- Deep learning -- Hybrid models; Ocean engineering -- Periodicals; Ocean engineering; Periodicals; 620.4162
• DOI: 10.1016/j.oceaneng.2020.108254
• Languages: English
• ISSNs: 0029-8018

Abstract

Abstract: Harmonic tidal currents constitutions forecasting is the first step to integrate tidal power into the main grid. It is important to improve the accuracy of tidal current forecasting models to improve its integration. This paper is proposing accurate hybrid deep learning models for harmonic tidal currents constitutions forecasting to improve the overall system accuracy. These hybrid models involve various combinations of Wavelet and Artificial Neural Networks (WNN and ANN), Fourier Series based on least squares (FS) and Recurrent Kalman Filter (RKF). Four main hybrid models are proposed and tested. From these models the best model with the highest performance is the hybrid of WNN, ANN and FS. The order of the techniques used is very important. Different combinations with different orders are tested. Six models with the best combinations are tested with different techniques order. The novelty of this work is the hybrid of optimized intelligent techniques for deep learning. The proposed work is validated using different data set which is the tidal currents direction and wave height and prove their effectiveness. All proposed models are accurate, but the best model is a hybrid of WNN, ANN and FS. Highlights: Nine different hybrid models are proposed for harmonic tidal currents constitutions forecasting. The hybrid approaches are combinations of WNN, ANN, FS and RKF. The most accurate hybrid model is the model of ANN, WNN and FS in order. Two different datasets are used (tidal direction and wave height) for work validation.