摘要
含分布式光伏的主动配电网具有较高的新能源渗透率和动态负荷特征,传统状态估计方法在应对复杂动态行为和数据质量问题时存在局限,影响配电网调度优化与安全性。提出一种基于图神经网络(GNN)与Informer模型融合的含光伏主动配电网状态估计方法。借助GNN构建量测模型,有效捕捉配电网拓扑结构的空间依赖特性;利用Informer的长时间序列预测能力,构建状态预测模型,精准地刻画配电网的动态物理特性。所提模型不仅能够处理量测数据中的缺失和噪声问题,而且能够增强状态估计的鲁棒性和物理一致性。基于IEEE 33节点系统及某实际配电网的仿真结果表明,所提方法在状态估计的准确性、抗干扰能力、物理一致性和计算效率方面均有显著提升。
The active distribution network with distributed photovoltaic has high renewable energy penetration rate and dynamic load characteristic,traditional state estimation methods have limitation in dealing with the problems of complex dynamic behavior and data quality,which affects the dispatch optimization and security of distribution network.A state estimation method for active distribution network with photovoltaic is proposed based on the combination of graph neural network(GNN)and Informer model.The GNN is used to construct the measurement model,which effectively captures the spatial dependency characteristic of topological structure of distribution network.The long-term sequence prediction ability of Informer is used to construct the state prediction model,which precisely describes the dynamic physical property of distribution network.The proposed model not only can handle the missing and noisy data problem in the measurement data,but also can enhance the robustness and physical consistency of state estimation.The simulative results based on the IEEE 33-bus system and a real distribution network show that the aspects of the proposed method in state estimation accuracy,anti-interference capability,physical consistency,and computational efficiency are significantly improved.
作者
郝蛟
王冬
邱剑
郭创新
HAO Jiao;WANG Dong;QIU Jian;GUO Chuangxin(Shenzhen Power Supply Co.,Ltd.,Shenzhen 518002,China;College of Electrical Engineering,Zhejiang University,Hangzhou 310007,China)
出处
《电力自动化设备》
北大核心
2025年第8期12-19,共8页
Electric Power Automation Equipment
基金
中国南方电网科技项目(090000KK52222149)
国家自然科学基金资助项目(U22B2098)。
关键词
主动配电网
分布式光伏
状态估计
分布式新能源
图神经网络
INFORMER
active distribution network
distributed photovoltaic
state estimation
distributed renewable energy
graph neural network
Informer
