摘要
风电高渗透接入电网导致电力电量平衡难度增加,为充分挖掘风电出力时序特征,生成特定风电场景以辅助电网提前进行分析计算,提出基于长短期记忆-辅助分类器生成对抗网络(LSTM-ACGAN)的特定风电场景生成方法。提出考虑电力电量平衡的优化问题框架对原始风电场景进行分类;在ACGAN的基础上提出了生成器引入长短期记忆层的LSTM-ACGAN模型结构,以提高模型对时序特征的学习能力;使用分类后的风电场景对其训练,以实现对特定类型风电场景的高效生成。在新英格兰10机39节点系统进行算例分析,所提模型的场景生成整体准确率相比传统ACGAN提升近10%;将生成风电场景用于鲁棒调度,能够显著提升机组组合结果的鲁棒性。
The high penetration of wind power into the power grid leads to the increasing difficulty of balancing electric power and energy.To fully exploit the sequential characteristics of wind power output,specific wind power scenarios are generated to assist the power grid in conducting analysis and calculations in advance.A method for generating specific wind power scenarios based on the long short-term memory auxiliary classifier generative adversarial network(LSTM-ACGAN)is proposed.The original wind power scenarios are classified within the proposed optimization problem framework considering electric power and energy balance.The LSTM-ACGAN model structure with an LSTM layer introduced in the generator is proposed based on the ACGAN to improve the learning ability of the model for sequential characteristic.The classi⁃fied wind power scenarios are trained to achieve the high efficiency generation for specific types of wind power scenarios.A case study is conducted in the New England 10-machine 39-bus system,where the overall accuracy of the proposed model is nearly 10%higher than the traditional ACGAN.Using the generated wind power scenarios for robust scheduling can significantly improve the robustness of the unit commitment results.
作者
葛彦硕
周艳真
兰健
郭庆来
GE Yanshuo;ZHOU Yanzhen;LAN Jian;GUO Qinglai(Weiyang College,Tsinghua University,Beijing 100084,China;Department of Electrical Engineering,Tsinghua University,Beijing 100084,China)
出处
《电力自动化设备》
北大核心
2025年第11期10-16,24,共8页
Electric Power Automation Equipment
基金
国家自然科学基金资助项目(U22B2097,52321004)。
关键词
辅助分类器生成对抗网络
长短期记忆网络
电力电量平衡
风电场景生成
生成式人工智能
auxiliary classifier generative adversarial network
long short-term memory network
electric power and energy balance
wind power scenarios generation
generative artificial intelligence
