Tyan-WP: A Wind Power Foundation Model for Ultra-Short-Term Probabilistic Forecasting

2026-06-07

AI Summary

This study addresses the limited generalization of existing wind power forecasting models under data-scarce conditions, which hinders rapid grid integration of new wind farms. To overcome this challenge, the authors propose the first foundation model for wind power prediction, leveraging large-scale pretraining to jointly encode static site metadata—such as geographic coordinates, terrain features, and ecoregions—with dynamic interactions between power output and meteorological covariates. The architecture introduces a static site embedding module and a power-aware meteorological fusion mechanism, enabling, for the first time, zero-shot probabilistic very-short-term forecasting across geographically diverse locations. Experimental results demonstrate substantial improvements over state-of-the-art methods, with a 19.9% reduction in mean absolute error (MAE) and a 22.2% reduction in continuous ranked probability score (CRPS), while also exhibiting strong out-of-distribution generalization on unseen sites in the UK.

Abstract

Global wind power capacity, especially in China, is booming, with new farms spanning diverse terrains and climates. The industry urgently needs accurate wind power foundation models to shorten commissioning and accelerate grid connection. This is because site-specific time series models (TSMs) are not well suited to data-scarce scenarios and generalize poorly, while generic large time series models (LTSMs) are mostly limited to univariate inputs and cannot fully exploit static site attributes or the dependencies between power and meteorological covariates, leading to insufficient accuracy. To fill this gap, we propose \textbf{Tyan-WP}, the first wind power foundation model for ultra-short-term probabilistic forecasting. Pretrained on a large-scale wind power dataset covering more than 126,000 U.S. sites over seven years, Tyan-WP further improves zero-shot forecasting through two domain-specific module designs: static site embedding using coordinate, terrain, and ecoregion metadata, and a power-aware meteorological fusion (PAMF) module that models interactions between historical power and meteorological covariates. Under a unified evaluation protocol, Tyan-WP surpasses eight site-specific supervised TSMs on 10 in-domain sites and outperforms eleven generic LTSMs on 127 in-domain sites, reducing MAE by 19.9%, RMSE by 16.6%, CRPS by 22.2%, and AQL by 21.7%, while raising R^2 by 16.7%. It further demonstrates strong cross-geography generalization on six real U.K. sites. These results show that the wind power foundation model can achieve accurate zero-shot forecasting without target-site training, providing a practical pathway for rapid turbine onboarding and probabilistic risk management at new wind farms.

Problem

• wind power forecasting
• foundation model
• probabilistic forecasting
• time series models
• meteorological covariates

Innovation

• wind power foundation model
• ultra-short-term probabilistic forecasting
• static site embedding
• power-aware meteorological fusion
• zero-shot forecasting