Dan Li, Yue Hu, Baohua Yang, Zeren Fang, Yunyan Liang, Shuai He
{"title":"基于 TimesNet-GRU 架构的新型风能预测迁移学习策略","authors":"Dan Li, Yue Hu, Baohua Yang, Zeren Fang, Yunyan Liang, Shuai He","doi":"10.1063/5.0200518","DOIUrl":null,"url":null,"abstract":"Currently, data-driven deep learning models are widely applied in the field of wind power prediction. However, when historical data are insufficient, deep learning models struggle to exhibit satisfactory predictive performance. In order to overcome the issue of limited training data for new wind farms, this study proposes a novel transfer learning strategy to address the challenge of less-sample learning in short-term wind power prediction. The research is conducted in two stages. In the pre-training stage, the TimesNet-GRU prediction model is established using data from a source wind farm. Parallel TimesNet modules are employed to extract multi-period features from various input feature sequences, followed by the extraction of long- and short-term features from the time series through gate recurrent unit (GRU). In the transfer learning stage, an effective transfer strategy is designed to freeze and retrain certain parameters of the TimesNet-GRU, thereby constructing a prediction model for the target wind farm. To validate the effectiveness of this approach, the results from testing with actual data from five wind farms in northwest China demonstrate that the proposed method exhibits significant advantages over models without transfer learning as explored in this study.","PeriodicalId":16953,"journal":{"name":"Journal of Renewable and Sustainable Energy","volume":null,"pages":null},"PeriodicalIF":1.9000,"publicationDate":"2024-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A novel transfer learning strategy for wind power prediction based on TimesNet-GRU architecture\",\"authors\":\"Dan Li, Yue Hu, Baohua Yang, Zeren Fang, Yunyan Liang, Shuai He\",\"doi\":\"10.1063/5.0200518\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Currently, data-driven deep learning models are widely applied in the field of wind power prediction. However, when historical data are insufficient, deep learning models struggle to exhibit satisfactory predictive performance. In order to overcome the issue of limited training data for new wind farms, this study proposes a novel transfer learning strategy to address the challenge of less-sample learning in short-term wind power prediction. The research is conducted in two stages. In the pre-training stage, the TimesNet-GRU prediction model is established using data from a source wind farm. Parallel TimesNet modules are employed to extract multi-period features from various input feature sequences, followed by the extraction of long- and short-term features from the time series through gate recurrent unit (GRU). In the transfer learning stage, an effective transfer strategy is designed to freeze and retrain certain parameters of the TimesNet-GRU, thereby constructing a prediction model for the target wind farm. To validate the effectiveness of this approach, the results from testing with actual data from five wind farms in northwest China demonstrate that the proposed method exhibits significant advantages over models without transfer learning as explored in this study.\",\"PeriodicalId\":16953,\"journal\":{\"name\":\"Journal of Renewable and Sustainable Energy\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.9000,\"publicationDate\":\"2024-05-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Renewable and Sustainable Energy\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://doi.org/10.1063/5.0200518\",\"RegionNum\":4,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Renewable and Sustainable Energy","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1063/5.0200518","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
A novel transfer learning strategy for wind power prediction based on TimesNet-GRU architecture
Currently, data-driven deep learning models are widely applied in the field of wind power prediction. However, when historical data are insufficient, deep learning models struggle to exhibit satisfactory predictive performance. In order to overcome the issue of limited training data for new wind farms, this study proposes a novel transfer learning strategy to address the challenge of less-sample learning in short-term wind power prediction. The research is conducted in two stages. In the pre-training stage, the TimesNet-GRU prediction model is established using data from a source wind farm. Parallel TimesNet modules are employed to extract multi-period features from various input feature sequences, followed by the extraction of long- and short-term features from the time series through gate recurrent unit (GRU). In the transfer learning stage, an effective transfer strategy is designed to freeze and retrain certain parameters of the TimesNet-GRU, thereby constructing a prediction model for the target wind farm. To validate the effectiveness of this approach, the results from testing with actual data from five wind farms in northwest China demonstrate that the proposed method exhibits significant advantages over models without transfer learning as explored in this study.
期刊介绍:
The Journal of Renewable and Sustainable Energy (JRSE) is an interdisciplinary, peer-reviewed journal covering all areas of renewable and sustainable energy relevant to the physical science and engineering communities. The interdisciplinary approach of the publication ensures that the editors draw from researchers worldwide in a diverse range of fields.
Topics covered include:
Renewable energy economics and policy
Renewable energy resource assessment
Solar energy: photovoltaics, solar thermal energy, solar energy for fuels
Wind energy: wind farms, rotors and blades, on- and offshore wind conditions, aerodynamics, fluid dynamics
Bioenergy: biofuels, biomass conversion, artificial photosynthesis
Distributed energy generation: rooftop PV, distributed fuel cells, distributed wind, micro-hydrogen power generation
Power distribution & systems modeling: power electronics and controls, smart grid
Energy efficient buildings: smart windows, PV, wind, power management
Energy conversion: flexoelectric, piezoelectric, thermoelectric, other technologies
Energy storage: batteries, supercapacitors, hydrogen storage, other fuels
Fuel cells: proton exchange membrane cells, solid oxide cells, hybrid fuel cells, other
Marine and hydroelectric energy: dams, tides, waves, other
Transportation: alternative vehicle technologies, plug-in technologies, other
Geothermal energy