全球能源互联网2024，Vol.7Issue(3)：271-282,12.DOI:10.19705/j.cnki.issn2096-5125.2024.03.004
基于典型源荷耦合模式的中长期时序场景生成
Medium-and Long-term Time-series Scenarios Generation Based on Typical Source-load Coupling Modes
摘要
Abstract
The degree of source-load coupling in high percentage renewable energy power systems is greatly increased under the influence of meteorology.Scenarios generation methods based on generative models are well-established tools for analyzing source-load uncertainty,however,the quality of directly generated medium-and long-term source-load coupling time series scenarios is poor in practical applications,which is often limited by the problem of small matching samples of source-load data.In this paper,a medium-to long-term time series scenarios generation method based on typical source-load coupling modes is proposed.First,the concept of source-load respective meteorological modes is defined.Then,the source and loads in similar typical meteorological modes in the historical data are aggregated into defined typical source-load coupling modes by meteorological distances.Next,a generative adversarial network is employed to generate typical wind and photovoltaic scenarios,and the generated scenarios are matched with the typical source-load coupling modes.Finally,the typical source-load coupling scenarios are obtained based on the matched correspondence.The simulation results for the Japanese region as an example show that the method proposed in the paper can improve the scenarios'interpretability in terms of meteorology,and at the same time reduce the impact of the small number of matching samples of source-load data on the generation of medium-and long-term source-load coupled scenarios.关键词
典型源荷耦合模式/气象距离/场景生成/中长期时序场景Key words
typical source-load coupling mode/meteorological distance/scenarios generation/medium-and long-term time-series scenarios分类
信息技术与安全科学引用本文复制引用
黄津钜,赵鹏飞,周航,徐新智,高艺,孙英云..基于典型源荷耦合模式的中长期时序场景生成[J].全球能源互联网,2024,7(3):271-282,12.基金项目
国家重点研发计划(2023YFB2405900).National Key Research and Development Program(2023YFB2405900). （2023YFB2405900）
