摘要
电网线路参数辨识具有十分重要的意义,而电网结构日益复杂和数据污染等因素对参数辨识造成不良影响,传统的参数辨识方法已不能充分满足现有电力工作的需要。针对这一问题,文中提出了一种将图卷积运算和Transformer结合的多任务自注意力图卷积(SAGCN)模型,从不同的角度深度挖掘电网线路的空间特征。切比雪夫图卷积对空间特征信息进行预处理,Graph Former实现全局信息交互与邻接关系重构。通过对比传统算法和噪声干扰实验表明:所提出的模型不仅提高了多目标、多支路参数识别的精度,同时具有良好的鲁棒性。
The identification of power grid line parameters is of great significance.However,factors such as the increasingly complex grid structure and data pollution have negative effects on parameter identification,and traditional parameter identification methods cannot fully meet the current needs.In response to this problem,this paper proposes a multi⁃task Self⁃attention Graph Convolution(SAGCN)model that combines graph convolution operations and Transformers to deeply mine the spatial characteristics of power grid lines from different perspectives.Chebyshev graph convolution preprocesses spatial feature information,and Graph Former realizes global information interaction and adjacency reconstruction.By comparing traditional algorithms and noise interference experiments,the results show that the proposed model not only improves the accuracy of multi⁃target and multi⁃branch parameter recognition,but also has good robustness.
作者
汪梦龙
宋公飞
WANG Meng-long;SONG Gong-fei(School of Automation,Nanjing University of Information Science and Technology,Nanjing 210044,China;Jiangsu Collaborative Innovation Center of Atmospheric Environment and Equipment Technology,Nanjing University of Information Science and Technology,Nanjing 210044,China)
出处
《信息技术》
2024年第8期31-37,共7页
Information Technology
基金
国家自然科学基金项目(61973170,61973168)。
