| 引用本文: | 孙锴,张大海,李亚平,严嘉豪.考虑时空不确定性的风电出力场景生成方法[J].电力自动化设备,2024,44(7):101-107 |
| SUN Kai,ZHANG Dahai,LI Yaping,YAN Jiahao.Scenario generation method of wind power output considering spatiotemporal uncertainty[J].Electric Power Automation Equipment,2024,44(7):101-107 |
|
| 摘要: |
| 为准确描述风电出力的不确定性及时空相关性，提出一种考虑时空不确定性的风电出力场景生成方法。将生成对抗网络作为风电出力的场景生成模型，将卷积神经网络作为模型生成器与判别器以实现时间特征的提取，采用特征工程方式实现不同风电场间出力空间相关性的量化；通过格拉姆角场方式进行特征变换，并合理设置网络结构及参数进行网络训练，得到生成器输入与输出场景间的映射关系。采用实测数据对所提方法的有效性进行对比验证，实验结果表明所提方法具有较强的风电出力不确定性表示能力。 |
| 关键词: 场景生成 时空特性 特征工程 不确定性 生成对抗网络 |
| DOI：10.16081/j.epae.202312016 |
| 分类号:TM73 |
| 基金项目:国家重点研发计划项目(2022YFB2403400) |
|
| Scenario generation method of wind power output considering spatiotemporal uncertainty |
|
SUN Kai1, ZHANG Dahai1, LI Yaping2, YAN Jiahao2
|
|
1.School of Electrical Engineering, Beijing Jiaotong University, Beijing 100044, China;2.China Electric Power Research Institute(Nanjing),Nanjing 210003, China
|
| Abstract: |
| In order to accurately describe the uncertainty and spatiotemporal correlation characteristic of wind power output, a scenario generation method of wind power output considering the spatiotemporal uncertainty is proposed. The generative adversarial network is taken as the scenario generation model of wind power output, the convolution neural network is taken as the model generator and discriminator to extract the time feature, and the quantization of spatiotemporal correlation characteristic between the outputs of different wind farms is realized by the mode of feature engineering. The feature transformation is carried out through the mode of Gramian angular field, the network structure and parameters are reasonably set for network training, and the mapping relationship between input and output scenarios of generator is obtained. The measured data is adopted for comparison and verification of the proposed method, and the experimental results show that the proposed method has strong ability to express the uncertainty of wind power output. |
| Key words: scenario generation spatiotemporal characteristic feature engineering uncertainty generative adversarial network |
