多图层模型下电网调度空间异构数据特征融合  

Spatial heterogeneous data feature fusion for power grid dispatching under multilayer model

作　　者：陈斌 陈郑平 李泽科 郭久煜 范海威 李军良 CHEN Bin;CHEN Zhengping;LI Zeke;GUO Jiuyu;FAN Haiwei;LI Junliang(State Grid Fujian Electric Power Co.,Ltd.,Fuzhou 350000,China;Beijing Kedong Power Control System Co.,Ltd.,Beijing 100000,China)

机构地区：[1]国网福建省电力有限公司,福建福州350000 [2]北京科东电力控制系统有限责任公司,北京100000

出　　处：《电子设计工程》2025年第5期162-165,171,共5页Electronic Design Engineering

基　　金：国网福建省电力有限公司2020年科技项目(52130419002F)。

摘　　要：电网调度空间的数据量通常较大,并且在不同层次和时间尺度上都会产生大量的数据,对其融合计算复杂度较高。为此,引入多图层模型,提出电网调度空间异构数据特征融合方法。构建空间异构数据特征多图层融合模型和模糊隶属函数,计算相空间重构平均值,获取多元目标数据融合方差。采用修正函数修正状态,获取数据特征融合跟踪结果。使用k最近邻(k-Nearest Neighbors,kNN)算法求解节点相似度和空间距离。构建数据特征共享模型,实现调度空间异构数据特征的有效融合。由实验结果可知,该方法对于单节点和多节点的异构数据特征均具有良好的融合效果。The data volume of power grid scheduling space is usually huge,and a large amount of data is generated at different levels and time scales,and the computational complexity of fusion is high.Therefore,a multi-graph layer model is introduced and a feature fusion method of heterogeneous data in power grid scheduling space is proposed.The multi-layer fusion model and fuzzy membership function of spatial heterogeneous data features are constructed,and the mean value of phase space reconstruction is calculated to obtain the fusion variance of multiple target data.The modified function is used to correct the state,and the tracking result of data feature fusion is obtained.The k-Nearest Neighbors(kNN)algorithm is used to calculate node similarity and spatial distance.A data feature sharing model is constructed to realize the effective fusion of heterogeneous data features in scheduling space.The experimental results show that the proposed method has a good fusion effect for both single node and multi-node heterogeneous data features.

关 键 词：多图层模型 电网调度空间 异构数据 特征融合 

分 类 号：TN108.1[电子电信—物理电子学]