2026· Computers, Materials & Continua· pp. 1-10· 0 citations· 29 references
Abstract
: Short-term wind power forecasting up to 72 h ahead is essential for reliable power system operation and efficient electricity market participation under high renewable penetration. However, operational numerical weather prediction (NWP) products are substantially coarser than the spatial and temporal scales governing turbine-level inflow variability. In contrast, turbine-level refinement of coarse NWP for multi-step forecasting remains largely unexplored. To address this issue, this study proposes a multi-scale short-term wind power forecasting framework that combines graph-based NWP super-resolution with transformer-based temporal forecasting. First, a graph learning module refines coarse NWP data into turbine-level wind speed representations by incorporating turbine layout topology and wind-direction-aware spatial interactions. Second, a Temporal Fusion Transformer (TFT) performs parallel multi-horizon forecasting across turbines using the refined wind speeds, historical power measurements, and turbine-specific static covariates. Farm-level forecasts are subsequently obtained through hierarchical aggregation. Case studies on two real-world wind farms with contrasting terrain characteristics demonstrate that the proposed framework consistently outperforms representative baselines, delivering a 34.1% decrease in normalized mean absolute error (NMAE) and a 28.6% gain in normalized continuous ranked probability score (NCRPS). The results indicate that turbine-level refinement of coarse NWP inputs before temporal forecasting significantly improves short-term forecasting performance, providing an effective and scalable solution for multi-scale wind power forecasting.
Accurate short-term wind power forecasting is critical for grid stability and electricity trading. While transformer-based deep learning is state of the art, industry practice relies mostly on local models trained for individual assets. This study investigates whether a global temporal fusion transformer (TFT), trained...
Single-height wind speed is often insufficient to represent vertical
wind-field variations in ultra-short-term wind power forecasting, while fluctuations in supervisory
control and data acquisition (SCADA) data quality can degrade forecasting performance.
A forecasting method that combines multi-height wind...
Nan-Tian Huang, Zhong-Yuan Niu, G. Cai et al.· Recent Advances in Computer...· 0 citations
Reliable short-term inflow forecasting is critical for optimizing hydroelectric power plant (HPP) operations, where water availability is directly influenced by dynamic and often extreme weather patterns. Inaccurate inflow estimations can result in either uncontrolled water spills or reservoir levels dropping below ope...
W. Prakoso, Ari Santoso, Hendra Kusuma· JAREE (Journal on Advanced R...· 0 citations
Accurate hub-height wind-speed forecasting is a prerequisite for the safe and economical integration of wind power into the electricity grid, because forecasting errors propagate directly into imbalance penalties, reserve sizing, and turbine wear, and short-term predictions (one to twenty-four hours ahead) are of direc...
Jian-Guo Yan, Zhe Li, Chuang-Ye Li et al.· Journal of Marine Science an...· 0 citations
Accurate ultra-short-term regional load forecasting provides critical support for power grid dispatch and electricity market transactions. Modeling the spatio-temporal correlations among load nodes within a region is an effective method for improving the accuracy of regional load forecasting. However, existing methods...
To address the nonlinear amplification of numerical weather prediction (NWP) errors and the difficult trade-off between coverage and sharpness in short-term offshore wind power forecasting, this paper proposes PRWind, a physics-guided framework for short-term probabilistic forecasting. Built upon a Transformer encoder,...
Xiu-Yong Zhao, Hai-Chuan Long, Kai-Ze Liu et al.· Atmosphere· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.