| 引用本文: | 张科鑫,窦晓波,李炜祺,胡永鸣,俞婧雯,戴睿鹏.基于量测数据补全的有源配电网电压优化技术[J].电力自动化设备,2023,43(11):67-74 |
| ZHANG Kexin,DOU Xiaobo,LI Weiqi,HU Yongming,YU Jingwen,DAI Ruipeng.Voltage optimization technology of active distribution network based on measurement data completion[J].Electric Power Automation Equipment,2023,43(11):67-74 |
|
| 摘要: |
| 有源配电网由于分布式新能源影响，电压存在越限的风险，但配电网实时量测仅可部分观测，优化问题无法求解。针对这个问题，提出了一种基于量测数据补全的有源配电网电压优化技术。在仅能获得部分节点实时量测的状态下，采用增强生成对抗网络补全算法，得到完整的配电网量测数据。根据补全的实时数据，计及补全误差，设计电压误差修正模型，修正优化电压目标，在电压越限时对配电网电压进行优化，提高电压质量。通过IEEE 33节点算例验证了所提方法相对于生成对抗网络在部分实时观测的情况下能够高精度补全缺失量测数据，降低电压波动，提高配电网运行的稳定性。 |
| 关键词: 增强生成对抗网络 非实时观测 数据补全 补全误差修正 电压优化 |
| DOI：10.16081/j.epae.202301010 |
| 分类号:TM761 |
| 基金项目:国家电网公司总部科技项目(5400?202199278A?0?0?00) |
|
| Voltage optimization technology of active distribution network based on measurement data completion |
|
ZHANG Kexin, DOU Xiaobo, LI Weiqi, HU Yongming, YU Jingwen, DAI Ruipeng
|
|
School of Electrical Engineering, Southeast University, Nanjing 210096, China
|
| Abstract: |
| Due to the influence of distributed new energy, the active distribution network has the risk of exceeding the limit of voltage, but the real-time measurement of the distribution network can only be partially observed, and the optimization problem cannot be solved. Aiming at this problem, a voltage optimization technology of active distribution network based on measurement data completion is proposed. In the state where only real-time measurements of some nodes can be obtained, the augmented generative adversarial network completion algorithm is used to obtain the complete distribution network measurement data. According to the real-time data of the completion, taking the completion error into account, the voltage error correction model is designed, the optimized voltage target is revised, and the voltage of the distribution network is optimized when the voltage exceeds the limit, so as to improve the voltage quality. The IEEE 33-bus example verifies that the proposed method can complete the missing measurement data with high precision compared with the generative adversarial network in the case of partial real-time observation, reduce the voltage fluctuation, and improve the stability of the distribution network operation. |
| Key words: boost-generative adversarial network non-real-time observation data completion completion error correction voltage optimization |
