异构网络通信数据融合方法研究  

Research on Heterogeneous Network Communication Data Fusion Methods

作　　者：谭如超 何群 王华 周欣 肖辉 TAN Ruchao;HE Qun;WANG Hua;ZHOU Xin;XIAO Hui(Information&Telecommunication Branch,State Grid Jiangxi Electric Power Co.,Ltd.,Nanchang 330096,China)

机构地区：[1]国网江西省电力有限公司信息通信分公司,江西南昌330096

出　　处：《自动化仪表》2023年第7期50-54,60,共6页Process Automation Instrumentation

基　　金：国网江西省电力有限公司科技计划基金资助项目(52183520007U)。

摘　　要：为实现异构网络的高效通信、填补缺失通信数据,提出了一种基于典型相关分析的异构网络通信数据融合方法。首先,采集异构网络通信数据,对数据进行清洗、规约以及标准化处理。然后,通过典型相关分析方法搜索预处理后的两组异构网络通信数据投影方向,将最大相关性投影结果作为通信数据特征。最后,建立判别准则函数以获取最终特征向量,依据特征向量融合异构网络通信数据。试验结果表明:该方法融合异构网络通信数据时的乒乓效应出现次数少且变化幅度小,数据融合可靠性和正确程度高。该方法应用后,可降低网络通信阻塞率、有效补齐缺失数据。该研究为实现异构网络通信数据的有效融合提供借鉴。To achieve efficient communication in heterogeneous networks and fill in the missing communication data,a heterogeneous network communication data fusion method based on canonical correlation analysis is proposed.Firstly,the heterogeneous network communication data is collected,and the data is cleaned,statute and normalized.Then,the projection direction of the two sets of heterogeneous network communication data after pre-processing by the canonical correlation analysis method is searched and the projection result with the maximum correlation is used as the communication data features.Finally,a discriminant criterion function is established to obtain the final feature vector,and the heterogeneous network communication data are fused based on the feature vector.The experimental results show that the method has less Ping-pong effect and less variation when fusing heterogeneous network communication data,and the data fusion is high degree of reliable and correct.The application of the method can reduce the network communication blockage rate and effectively fill in the missing data.This research provides reference for realizing the effective fusion of heterogeneous network communication data.

关 键 词：典型相关分析 异构网络 通信数据融合 增量式挖掘 数据规约 特征向量 数据清洗 通信阻塞 

分 类 号：TH-3[机械工程]