| 引用本文: | 高锋阳,裴淑萍,李龙,查鹏堂,王钦娟.基于改进生成对抗网络的可控可解释性风光场景生成方法[J].电力自动化设备,2026,46(5):118-126 |
| Gao Fengyang,Pei Shuping,Li Long,Zha Pengtang,Wang Qinjuan.Controllable and interpretable wind-solar scenario generation method based on improved generative adversarial network[J].Electric Power Automation Equipment,2026,46(5):118-126 |
|
| 摘要: |
| 针对风光可再生能源接入电网所带来的不确定性和间歇性，提出一种基于条件Wasserstein生成对抗网络-梯度惩罚(cWGAN-GP)的可控可解释的风光出力场景生成方法。cWGAN-GP通过引入条件向量调控风光出力场景的生成过程，使生成数据能够满足特定统计特征要求，结合梯度惩罚提升训练稳定性，确保生成场景质量，并准确捕捉风光出力的时空相关性；利用斯皮尔曼相关系数分析高维气象特征与风光出力之间的关系，筛选关键气象因子作为控制变量，以增强模型的可控性和对气象条件的适应能力。在IEEE 33节点系统上进行仿真验证，结果表明，所提方法能有效生成符合统计规律的风光出力场景，并较好地保持时空相关性。 |
| 关键词: 不确定性 场景生成 可解释性 cWGAN-GP 时空特征 |
| DOI：10.16081/j.epae.202508010 |
| 分类号:TM73 |
| 基金项目:甘肃省重点研发计划项目(23YFFA0059) |
|
| Controllable and interpretable wind-solar scenario generation method based on improved generative adversarial network |
|
Gao Fengyang1, Pei Shuping1, Li Long1, Zha Pengtang1, Wang Qinjuan2
|
|
1.School of Automation and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China;2.Economic and Technological Research Institute of State Grid Gansu Electric Power Company(New Energy Power System Research Institute),Lanzhou 730050, China
|
| Abstract: |
| Aiming at the uncertainty and intermittency brought by the integration of wind and solar renewable energy into power grid, a controllable and interpretable wind-solar output scenario generation method based on the conditional Wasserstein generative adversarial network with gradient penalty(cWGAN-GP) is proposed. The generation process of wind-solar output scenarios is regulated by introducing conditional vectors in cWGAN-GP, which ensures that the generated data meets specific statistical characteristics requirement. The gradient penalty is incorporated to enhance the training stability, which ensures the quality of generated scenarios while accurately capturing the spatiotemporal correlation of wind-solar output. The Spearman correlation coefficient is employed to analyze the relationship between high dimensional meteorological features and wind-solar output, and key meteorological factors are selected as the control variables to improve the model controllability and its adaptability to meteorological conditions. The IEEE 33-bus system is used for simulation and verification, and the results show that the proposed method can effectively generate wind-solar output scenarios conforming to statistical rules, and maintain spatiotemporal correlation well. |
| Key words: uncertainty scenario generation interpretability cWGAN-GP spatiotemporal characteristic |
