Short-Term Wind Power Forecasting Supporting Time-Scale Adaptive Update and Fusion of Multivariable Spatio-Temporal Information

View at publisher Summarise with VIRAT
Authors: Xiaodong Li, Chang Liu, Shijie Guan, Lin Ye, Yingying Zheng
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Short-term power forecasting for single wind farms mainly focuses on analyzing the temporal relationships in historical power generation data, often neglecting the spatial relationships between multiple meteorological variables in wind power generation systems. Moreover, the spatial coupling relationships between meteorological variables, such as wind speed and air pressure, exhibit dynamic changes across different time scales, which are challenging for machine learning models to capture. To address these issues, this paper proposes a novel short-term wind power forecasting method from a graph-based perspective, which supports time-scale adaptive updates and integrates multi-variable spatiotemporal information. First, multi-scale dynamic identification is performed using Fourier transform. Then, a multi-scale adaptive graph convolutional network is used to model wind farms and meteorological variables as graph structures, thereby capturing the spatial relationships of variables across multiple scales. Simultaneously, Mamba attention, combined with Transformers, is introduced to explore temporal dependencies, resulting in improved wind power output forecasting. Finally, an NRBO-STL trend error compensation module is designed to correct the forecasting results. Experiments demonstrate that the proposed method effectively leverages the spatio-temporal coupling characteristics between wind power output and meteorological data, achieving higher forecasting accuracy.

Fields of Study & Subjects

Energy Load and Power Forecasting Solar Radiation and Photovoltaics Traffic Prediction and Management Techniques

Publication details

Journal / source
IEEE Transactions on Industry Applications
Pages
1–14
Publication date
1 January 2026
Language
EN