Integrating Drivetrain Thermal Inertia Into Wind Power Forecasting: A Multi‐Model Benchmarking Approach
International Journal of Energy Research · 2026-01 · 2026 (1)
DOI: 10.1155/er/1399433 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Energy Research, online erschienen am 01.11.2026. Verfasst von Hesam Mohammad Asghari und 1 weiteren Autor:innen. Die Arbeit nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
As renewable energy usage scales up globally, wind energy development in arid regions increasingly faces the challenges of compound aerothermal events, where high wind potential coincides with extreme ambient thermal regimes. Here, we establish the predictive capability necessary to analyze operational response to these events through a case study of the Mil‐Nader wind farm in Sistan, Iran. We first quantify thermal derating during the 120‐day wind season using supervisory control and data acquisition (SCADA) data from twenty‐two 2.5‐MW turbines. Our analysis reveals critical features of thermal derating, including a thermal lag of 40–50 min ( τ lag = 40–50 min, with a peak cross‐correlation of R max = 0.84) in the response of turbine performance to ambient heat due to drivetrain thermal inertia, as well as a critical temperature threshold of 36°C above which the loss of electrical power relative to optimal aerodynamic potential exceeds 40%. Distinct from conventional static methods, the proposed sequential long short‐term memory (LSTM) architecture explicitly captures the prolonged time‐lag effect associated with the cumulative thermal mass of the gearbox oil. We benchmark the predictive fidelity of our model against alternative recurrent deep learning (DL) models, specifically the gated recurrent unit (GRU), gradient boosting algorithms via extreme gradient boosting (XGBoost), traditional multilayer perceptron (MLP) neural networks, and persistence baselines. The proposed LSTM framework achieves superior predictive accuracy ( R 2 = 0.94, root mean square error [RMSE] = 62.5 kW, and a capacity‐normalized prediction fidelity of 92.8%), statistically outperforming GRU (91.4%), XGBoost (86.5%), and MLP (82.4%). This provides essential load‐management and predictive maintenance (PdM) capability, enabling utility operators to take proactive corrective actions to prevent equipment damage and mitigate power losses in desert environments.
Zitieren
APA
Asghari, H. M., & Ghaffari, A. (2026). Integrating Drivetrain Thermal Inertia Into Wind Power Forecasting: A Multi‐Model Benchmarking Approach. International Journal of Energy Research, 2026(1). https://doi.org/10.1155/er/1399433
Vancouver
Asghari HM, Ghaffari A. Integrating Drivetrain Thermal Inertia Into Wind Power Forecasting: A Multi‐Model Benchmarking Approach. International Journal of Energy Research. 2026;2026(1). doi:10.1155/er/1399433
BibTeX
@article{Asghari2026,
author = {Asghari, Hesam Mohammad and Ghaffari, Ali},
title = {Integrating Drivetrain Thermal Inertia Into Wind Power Forecasting: A Multi‐Model Benchmarking Approach},
journal = {International Journal of Energy Research},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/er/1399433}
}
EUVIMED