• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:刘鹏飞,李 瑶,李捍东.基于LSTM-CGAN 的风电场景生成方法[J].软件工程,2025,28(2):17-20.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
基于LSTM-CGAN 的风电场景生成方法
刘鹏飞1,李 瑶2,李捍东1
(1.贵州大学电气工程学院,贵州 贵阳 550025;
2.国网四川省电力公司天府新区供电公司,四川 成都 610213)
1722758039@qq.com; 342707210@qq.com; 470394668@qq.com
摘 要: 针对传统风电场景生成方法未充分利用风电功率的预测误差,以及未合理考虑风电序列时间相关性的问题,提出了一种基于LSTM-CGAN(LongShort-Term MemoryConditionalGenerativeAdversarialNetwork)的风电场景生成方法。该方法在条件生成对抗网络模型的训练过程中引入了符合风电预测误差分布的随机噪声,同时使用深度长短期记忆网络搭建条件生成对抗网络的生成器和判别器。算例结果表明,所提方法生成的场景集对风电真实场景的覆盖率能够保持在98%以上,刻画风电的不确定性也不会过于保守,能够更好地学习到风电序列的时间相关性。
关键词: 风电;误差拟合;长短期记忆网络;条件生成对抗网络
中图分类号: TP399    文献标识码: A
基金项目: 国家自然科学基金(52167007)
Wind Power Scenarios Generation Method Based on LSTM-CGAN
LIU Pengfei1, LI Yao2, LI Handong1
(1.The Electrical Engineering College, Guizhou University, Guiyang 550025, China;
2.State Grid Sichuan Electric Power Company Tianfu New Area Power Supply Company, Chengdu 610213, China)
1722758039@qq.com; 342707210@qq.com; 470394668@qq.com
Abstract: Addressing the issues that traditional wind power scenario generation methods do not fully utilize the prediction errors of wind power and do not adequately consider the temporal correlation of wind power sequences, this paper proposes a wind power scenario generation method based on Long Short-Term Memory Conditional Generative Adversarial Network (LSTM-CGAN). This method introduces random noise that conforms to the distribution of wind power prediction errors during the training process of the conditional generative adversarial network model, while employing deep Long Short-Term Memory networks to construct the generator and discriminator of the conditional generative adversarial network. The results of the case studies indicate that the scenario set generated by the proposed method maintains a coverage rate of over 98% for real wind power scenarios, and it effectively captures the uncertainty of wind power without being overly conservative, allowing for better learning of the temporal correlations in wind power sequences.
Keywords: wind power; error fitting; Long Short-Term Memory network; Conditional Generative Adversarial Network (CGAN)


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫