LONG-TERM WIND POWER AND GLOBAL WARMING PREDICTION USING MARS, ANN, CART, LR, AND RF

2024-06-01
Yılmaz, Yavuz
Nalçacı, Gamze
Kańczurzewska, Marta
Weber, Gerhard Wilhelm
The modeling of electricity generation plays a crucial role in investment and long-term planning in power systems, primarily due to the significant volatility associated with wind and solar energy sources. Nevertheless, forecasting wind speeds for wind turbines based on weather conditions over an extended period is difficult and not feasible. This study provides long-term projections for wind power generation derived from a 2 MW wind turbine for the upcoming year and subsequent years utilizing the Multivariate Adaptive Regression Splines (MARS), Artificial Neural Networks (ANN), Classification And Regression Tree (CART), Linear Regression (LR) and Random Forest (RF) techniques. The research is carried out in two distinct phases. During Phase 1 all considered predictive methods are compared. The research demonstrates that the MARS algorithm is a robust and efficient predictor for wind-based power generation, exhibiting strong competitiveness in its performance. During Phase 2, the MARS algorithm is employed to forecast the future 30-year wind power generation capacity lifespan hourly for nine cities in Texas, USA. It is projected that El Paso and Dallas will witness a mean rise of 8.6% in wind power capacity over three decades, while the remaining seven cities are anticipated to have an average decline of 7.7%. Hence, it is imperative to do a comprehensive and extended evaluation employing the MARS technique compared to ANN, CART, LR and RF before installing a wind turbine. This analysis would serve as a crucial resource for investors, engineers, and researchers involved in decision-making processes on wind turbine projects.
Journal of Industrial and Management Optimization
Citation Formats
Y. Yılmaz, G. Nalçacı, M. Kańczurzewska, and G. W. Weber, “LONG-TERM WIND POWER AND GLOBAL WARMING PREDICTION USING MARS, ANN, CART, LR, AND RF,” Journal of Industrial and Management Optimization, vol. 20, no. 6, pp. 2193–2216, 2024, Accessed: 00, 2024. [Online]. Available: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85189461699&origin=inward.