摘要
为准确反映新能源出力的时序特征,特别是极端天气对系统的影响,提出一种考虑极端天气的先验知识引导风光短期出力场景生成方法研究,首先,基于极端度量指标进行数据识别,构建正常天气和极端天气数据集,然后,针对2种数据集利用混合密度网络(MDN)获取分布信息,最后,基于先验分布信息得到噪声序列,将其与历史实测数据输入并应用生成对抗网络(GAN)模型,获得风光出力场景。基于浙江某地区风电和光伏场站数据对所提方法的有效性进行验证,分析结果表明,所生成的风、光出力场景分别符合其真实时空相关关系,同时可以捕捉到极端天气下功率曲线随机特性,能够反映某地区风光实际出力时的波动性,为电力系统调度方案制定提供参考。
To accurately reflect the temporal characteristics of renewable energy output,especially the impact of extreme weather on the system,this paper proposes a short-term wind/photovoltaic output scenario generation method guided by prior knowledge considering extreme weather.Firstly,data identification is performed based on extreme metrics to construct datasets for normal and extreme weather conditions.Then,a mixed density network(MDN) is used to obtain distribution information for both datasets.Finally,the noise sequence is obtained based on prior distribution information,which is combined with historical measured data to train the generative adversarial network(GAN) model to generate wind and photovoltaic output scenarios.The effectiveness of the proposed method is verified using data from wind and photovoltaic stations in a region of Zhejiang Province.The analysis results show that the generated wind and photovoltaic output scenarios conform to their real spatiotemporal correlations and can capture the stochastic characteristics of power curves under the extreme weather,reflecting the actual output fluctuations of wind and photovoltaic power in the region.This provides a reference for the formulation of power system dispatch plans.
作者
陈浩
张文朝
黄志光
张怡静
时艳强
李剑锋
赵永宁
CHEN Hao;ZHANG Wenchao;HUANG Zhiguang;ZHANG Yijing;SHI Yanqiang;LI Jianfeng;ZHAO Yongning(State Grid Corporation of China East China Division,Shanghai 200120,China;Beijing Kedong Electric Power Control System Co.,Ltd.,Beijing 100192,China;College of Information and Electrical Engineering,China Agricultural University,Beijing 100083,China)
出处
《智慧电力》
北大核心
2025年第3期44-52,共9页
Smart Power
基金
国家自然科学基金资助项目(52207144)。
关键词
场景生成
先验知识
极端天气
风光出力
混合密度-对抗生成网络
scenario generation
prior knowledge
extreme weather
wind and photovoltaic power output
MDN-GAN model
作者简介
陈浩(1984),男,上海人,工程硕士,高级工程师,从事电力系统安全稳定研究。E-mail:chenhao@ec.sgcc.com.cn;通信作者:李剑锋(1996),男,重庆人,硕士研究生,从事电力系统研究。E-mail:lijianfe1102@163.com。
