加权有向图社区发现的子系统划分  

Weighted directed graph based community detection for subsystem decomposition

作　　者：杨晓峰 谢巍[1] 张浪文[1] YANG Xiao-feng;XIE Wei;ZHANG Lang-wen(College of Automation Science and Technology,South China University of Technology,Guangzhou Guangdong 510640,China;Guangzhou Institute of Standardization,Guangzhou Guangdong 510110,China)

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510640 [2]广州市标准化研究院,广东广州510110

出　　处：《控制理论与应用》2020年第9期1923-1930,共8页Control Theory & Applications

基　　金：国家自然科学基金项目(61803161,61973125);江门市创新科研团队引进基金项目(2017TD03)资助。

摘　　要：提出一种基于加权有向图的社区发现子系统划分方法,并应用于分布式状态估计设计.针对一类复杂非线性系统,构建考虑连接边强度的加权有向图,引入社区发现算法将复杂非线性系统划分成多个子系统.同时考虑子系统之间连接边的数量和有向图顶点之间的连接强度,使得划分得到的子系统内部关联较强,而子系统之间的耦合强度较弱.针对划分得到的子系统,设计基于信息交互的分布式滚动时域估计算法,并与已有的子系统划分方法对比,在相同的状态估计设定下,所提出的子系统划分方法能够有效提高状态估计的性能.This paper presents a subsystem decomposition method based on community structure detection with weighted directed graph.The configured subsystems are used for the design of distributed state estimation.For a complex nonlinear system,a weighted digraph considering the strength of the connected edge is constructed,and a community structure detection is introduced to divide a complex nonlinear system into several subsystems.The proposed subsystem decomposition method takes both the number of connecting edges between subsystems and the connection strength of the vertex of the digraph into account.To this end,the configured subsystems have stronger connections within a subsystem and weaker couplings among the subsystems.For the decomposed subsystem,a distributed moving horizon estimation method based on information interaction is designed.Compared with the existing methods,the proposed method can effectively improve the performance of state estimation for a same distributed state estimation.

关 键 词：子系统划分 滚动时域估计 社区发现算法 加权有向图 

分 类 号：O157.5[理学—数学]