A novel Weighted Average Weakening Buffer Operator based Fractional order accumulation Seasonal Grouping Grey Model for predicting the hydropower generation

• Published in: Energy (Oxford) 2023-08, Vol.277, p.127568, Article 127568
• Main Authors: Li, Zekai, Hu, Xi, Guo, Huan, Xiong, Xin
• Format: Article
• Language: English
• Subjects: Hydropower generation; Predictive accuracy; Weighted average weakening buffer operator based fractional order accumulation seasonal grouping grey model (WAWBO-FSGGM(1, 1))
• ISSN: 0360-5442
• DOI: 10.1016/j.energy.2023.127568

This paper proposes a novel Weighted Average Weakening Buffer Operator based Fractional order accumulation Seasonal Grouping Grey Model (WAWBO-FSGGM(1,1)) for accurate prediction on the hydropower generation. Firstly, to group the raw data with notable seasonality, this paper divides them into four groups to reorganize their order and establish a SGGM(1,1) model by utilizing the seasonal data grouping approach. Secondly, to eliminate the external interference, the WAWBO is applied for fitting the raw data and improving the predictive performance. Thirdly, to minimize the perturbation bound of the model, the fractional accumulated generating as a special data transformation is utilized by the fractional derivative parameters based on the series of the r-Fractional Accumulated Generating Operation (r-FAGO). Hence, a WAWBO-FSGGM(1,1) model is established to accurately predict the hydropower generation. Fourthly, to seek for a solution of the parameters that minimizes the predictive errors, the Whale Optimization Algorithm (WOA) is utilized for the WAWBO-FSGGM(1,1) model to further improve the predictive accuracy with fast-convergence. Finally, the empirical results demonstrate that the MAPE values of the WAWBO-FSGGM(1,1) model in five selected Chinese provinces and the U.S.A. are respectively 1.4182%, 1.7081%, 0.7971%, 2.3752%, 0.9062%, and 1.0056%, which are all the smallest among all the predicted results. •A WAWBO-FSGGM(1,1) model is proposed to predict the hydropower generation.•The model uses the seasonal data grouping and the WAWBO to improve the predictive accuracy.•The r-FAGO is used by the fractional derivative parameters for accurate prediction.•The WOA is proposed to further improve the predictive accuracy with fast-convergence.