| 引用本文: | 高怡杰,蔡德福,俞耀文,戴敬威,姚伟,陈金富.面向电力系统安全约束经济调度的数字孪生方法[J].电力自动化设备,2025,45(7):197-203 |
| GAO Yijie,CAI Defu,YU Yaowen,DAI Jingwei,YAO Wei,CHEN Jinfu.Digital twin method for security-constrained economic dispatch of power system[J].Electric Power Automation Equipment,2025,45(7):197-203 |
|
| 摘要: |
| 随着可再生能源发电的大规模并网，其高波动性使得电力系统的运行更加接近设备极限。为了预判安全风险，研究面向安全约束经济调度的数字孪生方法。基于包含物理实体、虚拟实体、连接、孪生数据和服务的5维框架，建立由净负荷预测和安全约束经济调度组成的数字孪生模型，用于模拟电力系统的未来运行状态；描述校正型安全约束经济调度的数学模型与预想故障筛选算法；采用分组时间序列Transformer为长时间模拟提供准确的净负荷预测，并结合跟随移动领导者算法在线更新预测模型以适应净负荷分布随时间的变化。算例结果验证了所提方法的安全性和准确性。 |
| 关键词: 数字孪生 安全约束经济调度 深度学习 净负荷预测 在线学习 |
| DOI：10.16081/j.epae.202502005 |
| 分类号: |
| 基金项目:国家电网有限公司科技项目(5100?202199558A-0-5-ZN) |
|
| Digital twin method for security-constrained economic dispatch of power system |
|
GAO Yijie1, CAI Defu2, YU Yaowen1, DAI Jingwei1, YAO Wei3, CHEN Jinfu3
|
|
1.School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan 430074, China;2.Electric Power Research Institute of State Grid Hubei Electric Power Co.,Ltd.,Wuhan 430077, China;3.School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan 430074, China
|
| Abstract: |
| Along with the large-scale grid integration of renewable energy generation, its high volatility makes the power system operation closer to the equipment limit. In order to predict the security risk, a digital twin method for security-constrained economic dispatch is studied. Based on a five-dimension framework including physical entity, virtual entity, connection, twin data and service, a digital twin model composed of net-load prediction and security-constrained economic dispatch is established to simulate the future operation status of power system. The mathematical model of corrective security-constrained economic dispatch and the expected failure filtering algorithm are described. The patch time series of Transformer is used to provide accurate net-load prediction for long-term simulation. Combined with the algorithm of follow the moving leader, the prediction model is updated online to adapt to the change of net-load distribution over time. The example results verify the security and accuracy of the proposed method. |
| Key words: digital twin security-constrained economic dispatch deep learning net-load prediction online learning |
