项目名称: 复杂动态网络系统的辨识方法研究
项目编号: No.61304138
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 刘艳君
作者单位: 江南大学
项目金额: 23万元
中文摘要: 复杂网络由许多个节点和节点间的连线组成。节点间的连线构成网络的拓扑结构,节点或者连线都可表示网络中的动态子系统。复杂网络的拓扑结构与网络中子系统的动态特性辨识是复杂网络研究中的重要内容。本项目拟针对节点表示可测端点信号,连线表示节点间动态特性的这类复杂网络,从系统辨识的角度,基于获得的节点量测信号,研究复杂网络拓扑结构辨识以及子系统动态特性的建模方法。主要内容包括:(1)当复杂网络中动态子系统用FIR或ARX模型描述时,基于Group Lasso和Group LAR等群组变量选择方法,挖掘复杂网络中的关联变量,实现网络拓扑结构的辨识;(2)研究基于压缩感知技术,在有限组测量数据下辨识复杂网络拓扑结构以及建立子系统动态特性模型的方法;(3)在网络结构已知的条件下,基于现有的闭环辨识方法,探讨子系统动态特性参数辨识的方法;(4)基于递阶辨识原理,研究网络结构下能够有效减少计算量的辨识方法。
中文关键词: 复杂网络;多变量系统;时滞估计;参数估计;压缩感知重构
英文摘要: A complex network consists of nodes and links, where the nodes or links may represent the dynamic subsystems of the network. The problem of finding the topology structure as well as identifying the dynamics of subsystems plays an important role in the research of complex networks. This project aims to develop identification methods to recover the topology structure and to determine the subsystems dynamics of a class of complex networks, in which the nodes represent signals and the links represent the transfer functions between signals, based on the signal measurements. The main work includes: (1) study the topology identification methods by mining the feature variables, based on some group variable selecting techniques, such as Group Lasso and Group LAR, when the subsystems of the network are modeled by FIR/ARX models; (2) develop topology and subsystems identification methods with finite data set, using the compressive sensing techniques; (3)extend the existing close-loop identification methods to identify the dynamic networks under known topology structures; (4)develop computationally efficient identification algorithms for complex networks based on the hierarchical identification principle.
英文关键词: complex network;multivariable system;time-delay estimation;parameter estimation;compressed sensing recovery