2018年5月4日学术报告

编辑:吴秦时间:2018-04-28点击数:

题目: Identification of dynamical network systems

报告人:俞成浦 教授

地点:江南大学物联网工程学院C317

时间:2018年5月4日星期五下午3:00

邀请人:丁锋 教授


内容简介:

Along with the popular applications of network systems in all sorts of research areas, the network model identification becomes a hot research topic in the system identification community. The identification problems of small-scale and large-scale network systems are discussed separately.

(i) The small-scale network identification (or gray-box identification) boils down to solving a non-convex optimization problem. The gradient-descent typed identification methods are usually sensitive to the selection of initial values. A low-rank optimization approach is presented which can yield good initial estimates, resulting in more chances to reach the global optimal solution to the non-convex optimization problem.

(ii) To deal with the identification of a large-scale network system, the network is partitioned into many subnetworks and each subnetwork is to be identified using its own input and output data. A spatio-temporal subspace identification method is proposed for the subnetwork identification, and sufficient conditions are provided for the consistent identification.


报告人简介:

俞成浦  北京理工大学教授,博士生导师。2014/05在新加坡南洋理工大学获得博士学位;2014/06-2017/08在荷兰代尔夫特工业大学做博士后;2017/09入职北京理工大学复杂系统智能控制与决策国家重点实验室;2018/02入选第十四批国家“千人计划”青年项目。

主要研究方向为系统辨识,信号处理,分布式优化。已在控制和信号处理领域顶级期刊IEEE Transactions on Automatic Control, Automatica和IEEE Transactions on Signal Processing上发表14篇一作论文。


技术支持:信息化建设与管理中心

地址:江苏省无锡市蠡湖大道1800号

邮编:214122

联系电话:0510-85910633,85910631

服务邮箱:iot@jiangnan.edu.cn