Time Series Analysis for Offshore Wind Energy Forecasting: Hybrid Model SARIMA and Genetic Algorithms

Authors

  • Fabrício de A. Santos Institute of Electrical Energy, Federal University of Maranhão, MA
  • Nerval de J. S. Junior Institute of Electrical Energy, Federal University of Maranhão, MA
  • Hellen. D. P. de Souza Institute of Electrical Energy, Federal University of Maranhão, MA
  • Shigeaki L. de Lima Institute of Electrical Energy, Federal University of Maranhão, MA

Keywords:

Time series, Forecast, SARIMA, Genetic Algorithms, Energy

Abstract

The conduct of studies on regional wind potential is essential to identify new projects and improve measurements, aiming for the efficient installation of turbines. The challenge of predicting wind behavior for energy extraction is of great interest, given the importance of accurate forecasts for the success of the sector. In this context, this article proposes a hybrid forecasting model, SARIMA and Genetic Algorithms, aiming to improve the accuracy of wind predictions using a wind measurement database, converting them into power generation scenarios. The SARIMA (5, 1, 5)x(6, 0, 6) model estimated by the algorithm was the best solution, demonstrating better fitness in tests. Its performance in MSE and MAE was 0.864 and 0.616, respectively.

Downloads

Published

2024-10-18

Issue

Vol. 4 No. 1 (2024): CBA2024

Section

Articles