A Unified Graph Formulation for Spatio-Temporal Wind Forecasting
With the rapid adoption of wind energy globally, there is a need for accurate short-term forecasting systems to improve the reliability and integration of such energy resources on a large scale. While most spatio-temporal forecasting systems comprise distinct components to learn spatial and temporal...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-10-01
|
| Series: | Energies |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1996-1073/16/20/7179 |
| _version_ | 1797573969132388352 |
|---|---|
| author | Lars Ødegaard Bentsen Narada Dilp Warakagoda Roy Stenbro Paal Engelstad |
| author_facet | Lars Ødegaard Bentsen Narada Dilp Warakagoda Roy Stenbro Paal Engelstad |
| author_sort | Lars Ødegaard Bentsen |
| collection | DOAJ |
| description | With the rapid adoption of wind energy globally, there is a need for accurate short-term forecasting systems to improve the reliability and integration of such energy resources on a large scale. While most spatio-temporal forecasting systems comprise distinct components to learn spatial and temporal dependencies separately, this paper argues for an approach to learning spatio-temporal information jointly. Many time series forecasting systems also require aligned input information and do not naturally facilitate irregular data. Research is therefore required to investigate methodologies for forecasting in the presence of missing or corrupt measurements. To help combat some of these challenges, this paper studied a unified graph formulation. With the unified formulation, a graph neural network (GNN) was used to extract spatial and temporal dependencies simultaneously, in a single update, while also naturally facilitating missing data. To evaluate the proposed unified approach, the study considered hour-ahead wind speed forecasting in the North Sea under different amounts of missing data. The framework was compared against traditional spatio-temporal architectures that used GNNs together with temporal long short-term memory (LSTM) and Transformer or Autoformer networks, along with the imputation of missing values. The proposed framework outperformed the traditional architectures, with absolute errors of around 0.73–0.90 m per second, when subject to 0–80% of missing input data. The unified graph approach was also better at predicting large changes in wind speed, with an additional 10-percentage-point improvement over the second-best model. Overall, this paper investigated a novel methodology for spatio-temporal wind speed forecasting and showed how the proposed unified graph formulation achieved competitive results compared to more traditional GNN-based architectures. |
| first_indexed | 2024-03-10T21:16:40Z |
| format | Article |
| id | doaj.art-6e0df39ea7cd44999e26a5cba18c1322 |
| institution | Directory Open Access Journal |
| issn | 1996-1073 |
| language | English |
| last_indexed | 2024-03-10T21:16:40Z |
| publishDate | 2023-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Energies |
| spelling | doaj.art-6e0df39ea7cd44999e26a5cba18c13222023-11-19T16:23:18ZengMDPI AGEnergies1996-10732023-10-011620717910.3390/en16207179A Unified Graph Formulation for Spatio-Temporal Wind ForecastingLars Ødegaard Bentsen0Narada Dilp Warakagoda1Roy Stenbro2Paal Engelstad3Department of Technology Systems, University of Oslo, P.O. Box 70, 2027 Kjeller, NorwayDepartment of Technology Systems, University of Oslo, P.O. Box 70, 2027 Kjeller, NorwayInstitute for Energy Technology, P.O. Box 40, 2027 Kjeller, NorwayDepartment of Technology Systems, University of Oslo, P.O. Box 70, 2027 Kjeller, NorwayWith the rapid adoption of wind energy globally, there is a need for accurate short-term forecasting systems to improve the reliability and integration of such energy resources on a large scale. While most spatio-temporal forecasting systems comprise distinct components to learn spatial and temporal dependencies separately, this paper argues for an approach to learning spatio-temporal information jointly. Many time series forecasting systems also require aligned input information and do not naturally facilitate irregular data. Research is therefore required to investigate methodologies for forecasting in the presence of missing or corrupt measurements. To help combat some of these challenges, this paper studied a unified graph formulation. With the unified formulation, a graph neural network (GNN) was used to extract spatial and temporal dependencies simultaneously, in a single update, while also naturally facilitating missing data. To evaluate the proposed unified approach, the study considered hour-ahead wind speed forecasting in the North Sea under different amounts of missing data. The framework was compared against traditional spatio-temporal architectures that used GNNs together with temporal long short-term memory (LSTM) and Transformer or Autoformer networks, along with the imputation of missing values. The proposed framework outperformed the traditional architectures, with absolute errors of around 0.73–0.90 m per second, when subject to 0–80% of missing input data. The unified graph approach was also better at predicting large changes in wind speed, with an additional 10-percentage-point improvement over the second-best model. Overall, this paper investigated a novel methodology for spatio-temporal wind speed forecasting and showed how the proposed unified graph formulation achieved competitive results compared to more traditional GNN-based architectures.https://www.mdpi.com/1996-1073/16/20/7179spatio-temporal forecastinggraph neural networksmissing datairregular time serieswind energy |
| spellingShingle | Lars Ødegaard Bentsen Narada Dilp Warakagoda Roy Stenbro Paal Engelstad A Unified Graph Formulation for Spatio-Temporal Wind Forecasting Energies spatio-temporal forecasting graph neural networks missing data irregular time series wind energy |
| title | A Unified Graph Formulation for Spatio-Temporal Wind Forecasting |
| title_full | A Unified Graph Formulation for Spatio-Temporal Wind Forecasting |
| title_fullStr | A Unified Graph Formulation for Spatio-Temporal Wind Forecasting |
| title_full_unstemmed | A Unified Graph Formulation for Spatio-Temporal Wind Forecasting |
| title_short | A Unified Graph Formulation for Spatio-Temporal Wind Forecasting |
| title_sort | unified graph formulation for spatio temporal wind forecasting |
| topic | spatio-temporal forecasting graph neural networks missing data irregular time series wind energy |
| url | https://www.mdpi.com/1996-1073/16/20/7179 |
| work_keys_str_mv | AT larsødegaardbentsen aunifiedgraphformulationforspatiotemporalwindforecasting AT naradadilpwarakagoda aunifiedgraphformulationforspatiotemporalwindforecasting AT roystenbro aunifiedgraphformulationforspatiotemporalwindforecasting AT paalengelstad aunifiedgraphformulationforspatiotemporalwindforecasting AT larsødegaardbentsen unifiedgraphformulationforspatiotemporalwindforecasting AT naradadilpwarakagoda unifiedgraphformulationforspatiotemporalwindforecasting AT roystenbro unifiedgraphformulationforspatiotemporalwindforecasting AT paalengelstad unifiedgraphformulationforspatiotemporalwindforecasting |