| 引用本文: | 符杨,李仪,葛晓琳,凡婉秋,魏书荣,刘璐洁.基于月度电量数据自适应伪增强的多海上风电场联合竞价博弈模型[J].电力自动化设备,2025,45(1):1-8 |
| FU Yang,LI Yi,GE Xiaolin,FAN Wanqiu,WEI Shurong,LIU Lujie.Joint bidding game model for multiple offshore wind farms based on monthly electricity data adaptive pseudo augmentation[J].Electric Power Automation Equipment,2025,45(1):1-8 |
|
| 本文已被：浏览 6150次 下载 1152次 |
码上扫一扫！ |
|
|
| 基于月度电量数据自适应伪增强的多海上风电场联合竞价博弈模型 |
|
符杨1,2, 李仪1, 葛晓琳1,2, 凡婉秋3, 魏书荣1,2, 刘璐洁1,2
|
|
1.上海电力大学 教育部海上风电技术工程研究中心，上海 200090;2.上海电力大学 海上风电研究院，上海 200090;3.上海慧程生物医疗科技有限公司，上海 201403
|
|
| 摘要: |
| 为降低随机性导致海上风电参与中长期电力市场竞争时面临的电量风险，提出一种基于月度电量数据自适应伪增强的多海上风电场中长期联合竞价博弈模型。考虑多海上风电场月度电量空间相关性对联合竞价策略的影响，利用带有自适应伪增强机制的生成对抗网络模拟多海上风电场月度电量预测误差场景，以增强小样本的多样性并减少对抗生成网络的过拟合现象；针对计及空间相关性与合作成本导致联盟收益函数不满足超可加性的问题，引入最优联盟结构的概念以求得总收益最大的联盟划分方式，并利用并行处理与联盟有效拆分的概念对求解算法进行改进，提高了执行效率；在分配联盟收益时，考虑到Shapley值法不能体现成员的真实贡献与重要程度，引入图合作博弈中的A-T解对分配方案进行改进，避免了不满足超可加性联盟的收益参与分配的问题。算例仿真结果验证了所提模型与方法的优越性以及收益分配方案的合理性。 |
| 关键词: 海上风电 电力市场 生成对抗网络 合作博弈 并行动态规划 A-T解 |
| DOI：10.16081/j.epae.202408020 |
| 分类号:TM614 |
| 基金项目:国家自然科学基金资助项目(52077130)；上海市青年科技启明星计划项目(21QA1403500)；上海绿色能源并网工程技术研究中心项目(13DZ2251900) |
|
| Joint bidding game model for multiple offshore wind farms based on monthly electricity data adaptive pseudo augmentation |
|
FU Yang1,2, LI Yi1, GE Xiaolin1,2, FAN Wanqiu3, WEI Shurong1,2, LIU Lujie1,2
|
|
1.Engineering Research Center of Offshore Wind Technology Ministry of Education, Shanghai University of Electric Power, Shanghai 200090, China;2.Offshore Wind Power Research Institute, Shanghai University of Electric Power, Shanghai 200090, China;3.Shanghai h-visions Biomedical Technology Co.,Ltd.,Shanghai 201403, China
|
| Abstract: |
| In order to mitigate the electricity risk faced by offshore wind power when participating in the medium and long-term electricity market due to its stochastic nature, a joint bidding game model for multiple offshore wind farms based on monthly electricity data adaptive pseudo augmentation is proposed. Considering the impact of spatial correlation of monthly electricity for multiple offshore wind farms on joint bidding strategy, the generative adversarial network with adaptive pseudo augmentation mechanism is used to simulate the monthly electricity prediction error scenario of multiple offshore wind farms, which enhances the diversity of small sample and reduce the overfitting phenomenon of generative adversarial network. Aiming at the problem that the coalition revenue function does not meet the superadditivity caused by the consideration of spatial correlation and cooperation cost, the concept of optimal coalition structure is introduced to obtain the coalition division mode with the maximum total revenue, and the concepts of parallel processing and effective optimal coalition structure are used to improve the solving algorithm, which improves the execution efficiency. Considering that the Shapley value method cannot reflect the true contribution and importance of members, the A-T solution in graph cooperative game is introduced to improve the allocation scheme when allocating coalition benefit, which avoids the problem that the coalition benefit not satisfying the superadditivity participates in the allocation. The superiority of the proposed model and method and the rationality of the benefit distribution scheme are verified by the example simulative results. |
| Key words: offshore wind power electricity market generative adversarial network cooperative game parallel dynamic programming A-T solution |
