Energy industries and governments consider ocean wave power a promising renewable energy source for reaching the net-zero plan by 2050 and restricting the rise in global temperatures. It expects the potential global ocean wave power production to be around 337 GW annually. Although wave energy forecasting critically enables economic dispatch, optimal power system management, and the integration of wave energy into power grids, the forecasting process is complicated by the stochastic, intermittent, and non-stationary nature of waves. Thus, this paper proposes a novel hybrid forecasting model comprising an adaptive decomposition-based method (Nelder-Mead variational mode decomposition) and a convolutional neural network featuring bi-directional long short-term memory. Furthermore, we propose a fast and effective optimiser to adjust the hybrid model's hyper-parameters and evaluate the decomposition technique's role in increasing the accuracy of wave energy flux predictions considering a forecasting period of 6 h. With regard to assessing the proposed model's effectiveness, we use a real wave dataset from a buoy positioned off Favignana Island in the Mediterranean Sea and compare the proposed model with six well-known forecasting methods and five hybrid deep-learning models. According to our findings, the proposed model significantly outperforms existing approaches over extended time periods and compared with the bi-directional long short-term memory, the developed adaptive decomposition method, and new hyper-parameters tuner improve the prediction accuracy at 45% and 13.6%, respectively.

Wave power forecasting using an effective decomposition-based convolutional Bi-directional model with equilibrium Nelder-Mead optimiser / Neshat, M.; Majidi Nezhad, M.; Sergiienko, N. Y.; Mirjalili, S.; Piras, G.; Astiaso Garcia, D.. - In: ENERGY. - ISSN 0360-5442. - 256:(2022), pp. 1-16. [10.1016/j.energy.2022.124623]

Wave power forecasting using an effective decomposition-based convolutional Bi-directional model with equilibrium Nelder-Mead optimiser

Majidi Nezhad M.;Piras G.;Astiaso Garcia D.
2022

Abstract

Energy industries and governments consider ocean wave power a promising renewable energy source for reaching the net-zero plan by 2050 and restricting the rise in global temperatures. It expects the potential global ocean wave power production to be around 337 GW annually. Although wave energy forecasting critically enables economic dispatch, optimal power system management, and the integration of wave energy into power grids, the forecasting process is complicated by the stochastic, intermittent, and non-stationary nature of waves. Thus, this paper proposes a novel hybrid forecasting model comprising an adaptive decomposition-based method (Nelder-Mead variational mode decomposition) and a convolutional neural network featuring bi-directional long short-term memory. Furthermore, we propose a fast and effective optimiser to adjust the hybrid model's hyper-parameters and evaluate the decomposition technique's role in increasing the accuracy of wave energy flux predictions considering a forecasting period of 6 h. With regard to assessing the proposed model's effectiveness, we use a real wave dataset from a buoy positioned off Favignana Island in the Mediterranean Sea and compare the proposed model with six well-known forecasting methods and five hybrid deep-learning models. According to our findings, the proposed model significantly outperforms existing approaches over extended time periods and compared with the bi-directional long short-term memory, the developed adaptive decomposition method, and new hyper-parameters tuner improve the prediction accuracy at 45% and 13.6%, respectively.
File allegati a questo prodotto
File Dimensione Formato  
Neshat_Wave_2022.pdf

solo gestori archivio

Tipologia: Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 3.43 MB
Formato Adobe PDF
3.43 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1657096
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? 1
social impact