西北大年夜学第七届先辈控制实际与应用 “至善前沿服装论坛t.vhao.net”暨中国主动化学会青年任务委员会走进高校系列研究会

发布者:王翔宇发布时间:2019-08-02浏览次数:14



西北大年夜学第七届先辈控制实际与应用
“至善前沿服装论坛t.vhao.net”
中国主动化学会青年任务委员会
走进高校系列研究会
2019 年 8 月 3 日
西北大年夜学四牌坊校区


服装论坛t.vhao.net赞助:高校双一流扶植经费
国度天然迷信基金委
中心高校根本科研营业基金
江苏省高校优势学科扶植工程专项资金
服装论坛t.vhao.net接洽:张亚(yazhang@seu.edu.cn)
陈杨杨(yychen@seu.edu.cn)
主办单位:西北大年夜学主动化学院

复杂体系丈量与控制教导部重点实验室





地点:西北大年夜学四牌坊校区榴园宾馆中大年夜厅


2019 8 3 日周六下午

掌管人:张亚 西北大年夜学


Distributed Consensus Optimization from Feedback-Forward Control  
Perspective
Lihua Xie

Nanyang Technological University, Singapore 14:00pm-14:50pm


OPTIMAL PERSISTENT MONITORING USING MULTIPLE AGENTS

王燕舞传授 华中科技大年夜学 14:50pm-15:20pm


无人体系的视觉定位,跟踪及协同最优控制

孟子阳副传授 清华大年夜学 15:20pm-15:50pm


茶歇 15:50pm-16:00pm


2019 8 3 日周六下午

掌管人:陈杨杨 西北大年夜学

 

基于事宜触发的完全分布式协同控制

李忠奎研究员 北京大年夜学 16:00pm-16:30pm


资本受限下的分布式估计与优化及在电力体系中的应用

朱善迎副传授 上海交通大年夜学 16:30pm-17:00pm


状况受限下的高阶滑模控制

丁世宏传授 江苏大年夜学 17:00pm-17:30pm


Distributed average consensus of continuous-time multi-agent systems under G
expectation

刘帅传授 山东大年夜学 17:30pm-18:00pm




Distributed Consensus Optimization from Feedback-Forward  

Control Perspective
Lihua Xie
School of Electrical and Electronic Engineering
Nanyang Technological University, Singapore
Abstract
Distributed consensus optimization aims to seek the common minimum of the sum of  
private costs in a network through only local information sharing. The problem has  
attracted a lot of interests in recent years due to its wide applications. In this talk, by  
treating the Laplacian operator under a weight-balanced and strongly connected  
network as a pseudo projector, we tackle the distributed optimization problem from a  
feedback-feedforward control viewpoint. The proposed control framework allows us  
to design distributed optimization algorithms and analyze their convergence using  
existing control techniques, which simplifies greatly the algorithm analysis. Two  
classes of controllers, namely proportional-integral (PI) and proportional-feedforward  
controllers, are developed. We also analyze and compare the convergence rate of  
proportional-integral and proportional-feedforward (PF) controllers for a class of  
quadratic costs, and demonstrate that the PF controller achieves faster convergence  
with a proper choice of control gains. The discrete-time implementation of the  
algorithms will also be studied.  
About the Speaker
Lihua Xie is a professor with the School of Electrical and  
Electronic Engineering, Nanyang Technological University  
(NTU), Singapore, and the Director, Delta-NTU Corporate Lab  
for Cyber-Physical Systems. His current research interests  
include networked control, multi-agent systems, sensor  
networks, compressive sensing, localization, and unmanned  
systems. He is an Editor-in-Chief of Unmanned Systems and0
Associate Editor, IEEE Transactions on Network Control  
Systems. He has served as Editor for IET Book Series in Control  
and Associate Editor for Automatica, IEEE Transactions on  
Automatic Control, IEEE Transactions on Control Systems Technology, IEEE  
Transactions on Circuit and Systems-II, etc. He was an IEEE Distinguished Lecturer  
(2011-2014), appointed member (2011) and elected member (Jan. 2016- Dec. 2018)  
of Board of Governors of IEEE Control System Society. Dr Xie is a Fellow of IEEE and  

Fellow of IFAC.


OPTIMAL PERSISTENT MONITORING USING MULTIPLE  

AGENTS
王燕舞传授
华中科技大年夜学
Abstract
Persistent monitoring arises in applications such as city patrolling, ecological  
surveillance, traffic monitoring, smart-grid security and ocean sampling. The  
dynamically changing environments in these applications require the agents to  
perpetually move in certain mission space. The challenge in this type of problems is to  
design the agent trajectories under physical motion constraints in order to optimize an  
overall performance metric. In this talk, we will discuss the one-dimensional optimal  
persistent monitoring problem using multiple agents. The goal is to control the  
movements of agents to minimize a performance metric associated with the  
environment (targets) over a finite time horizon. We will firstly present our recent result  
of monitoring targets with different weights, and then present a result of fulfilling the  
task using second-order agents. In contrast to earlier results limited to first-order  
dynamics for agents, we control their accelerations rather than velocities, thus leading  
to a better approximation of agent behavior in practice and to smoother trajectories.  
Bounds on both velocities and accelerations are also taken into consideration. Gradient
based algorithms are proposed to obtain sub-optimal solutions. Simulation examples  
will also be included to demonstrate the effectiveness of our results.
About the Speaker
王燕舞,华中科技大年夜学特聘岗传授、博士生导师。研究兴趣包含复杂搜集的
群集行动分析与控制、多无人体系调和控制、智能电网的优化调剂等。研究成果
曾获湖北省第六批优良博士学位论文奖、教导部天然迷信奖一等奖(第 3 完成
人),湖北省天然迷信奖一等奖(第 1 完成人)、湖北省出色青年基金支撑。2008

年当选教导部新世纪优良人才网job.vhao.net支撑筹划。


无人体系的视觉定位,跟踪及协同最优控制

孟子阳副传授
清华大年夜学
Abstract
申报起首对集群体系的研究背景停止简单的简介,然后详细报告请示近期的几个
研究成果,重要包含,基于时间戳同步的视觉惯性里程计,基于超微型无人机的
视觉目标跟踪,静态体系的分布式最优协同控制办法,和多无人体系的集合聚
集控制。
About the Speaker
孟子阳,清华大年夜学副传授,博士生导师,国度“青年千人”筹划当选者,洪
堡学者基金取得者。重要从事群体的智能控制与优化,微纳航天器,和无人系
统的智能导航的研究任务。在国际重要刊物和学术会议上发表论文 90 余篇,其
SIAM 汇刊,IEEE 汇刊,IFAC 汇刊等论文 30 余篇。掌管国度天然迷信基金
重点项目,面上项目,设备预研教导部结合基金等多个项目。取得 2018 年度北

京市迷信技巧一等奖。


基于事宜触发的完全分布式协同控制

李忠奎传授
北京大年夜学
Abstract
基于事宜触发的协同控制成绩近期遭到了很多存眷。其研究重点是设计合适
分布式控制战略和事宜触发条件,后者决定自平面甚么时候采样并与邻居通信。现有
绝大年夜多半文献存在的成绩是控制律或/和事宜触发函数依附于搜集拓扑的谱信息
等全局信息,即其实不是真正分布式。本申报将报告请示我们在基于事宜触发的完全分
布式协同控制律设计方面的一些停顿。
About the Speaker
李忠奎,现为北京大年夜学工学院力学与工程迷信系研究员,博士生导师。2005
年于国防科技大年夜学航天与材料工程学院获学士学位,2010 年于北京大年夜学工学院
获博士学位。重要从事分布式协同控制、搜集化控制体系、自立无人体系分析与
控制等方面的研究任务。撰写英文专著 1 部,发表 SCI 论文 50 多篇。当选 2017
年度教导部青年长江学者,2018 年科睿唯安然球高被引迷信家,和 2018 年度
北京市科技新星筹划。曾获 2015 年国度天然迷信二等奖(排名第三),第四届杨
嘉墀科技奖二等奖,2012 年全国优良博士学位论文奖等荣誉与嘉奖。第一作者
论文获 2013 IET Control Theory & Applications Premium Award 2009-2011

统迷信最好论文奖。


资本受限下的分布式估计与优化及在电力体系中的应用

朱善迎副传授
上海交通大年夜学
Abstract
随着信息技巧的生长,大年夜量具有感知和履行才能的设备和终端融入复杂物
理过程,经过过程信息获得、传输、处理和控制等相互感化,构成了融合控制和信
息通信的多维静态体系—搜个人系,其优化运转触及信息层面、传输层面和控
制层面等多方面成绩。本申报将聚焦信息层面的分布式感知、高效融合与资本
优化调剂等成绩,简介资本受限下的分布式估计和优化等方面的研究停顿。具
体内容触及异构搜集的分布式估计、感知-传输的结合设计与资本优化,并扼要
简介其在电力体系的状况估计与经济调剂等方面的应用。
About the Speaker
朱善迎,上海交通大年夜学主动化系副传授。2013 年卒业于上海交通大年夜学主动
化系获工学博士学位。2013 年至 2015 年,在新加坡南洋理工大年夜学和伯克利
新加坡教导同盟(BEARS)展开博士后研究任务。2015 年参加上海交通大年夜学主动
化系。重要研究范畴为分布式估计和优化、工业搜个人系的感知-传输结合设
计、微动力网的能量管理和交易机制设计等。掌管国度天然迷信基金、国度重
点研发筹小船课题等 6 项。在 IEEE 汇刊、AutomaticaApplied Energy
CDCICASSPGLOBECOM 等国际期刊和会议发表论文 60 余篇,协作出版
英文专著一部。取得国度天然迷信二等奖及省部级一等奖等 3 项。现为 IEEE
IES 工业信息学技巧委员会委员,中国主动化学会青工委委员,TCCT 多自平面
控制学组委员。担负多个国际会议的 Local Arrangement/Track/Session Chair

委员。


状况受限下的高阶滑模控制

丁世宏传授
江苏大年夜学
Abstract
高阶滑模可以或许处理传统滑模控制中的相对阶限制和抖振成绩,是以在实际和
工程界惹起了广泛存眷。现有的高阶滑模控制实际普通只能处理搅扰由常数限制
情况下的控制设计成绩,多为部分控制成果,而扰动由函数限制情况下的高阶滑
模全局控制设计成绩在实际体系控制中却更加广泛。针对该成绩,申报人将简介
搅扰由函数限制条件下的高阶滑模控制办法研究的最新停顿。起首,简介高阶滑
模控制成绩的来源及生长;其次,指出以后高阶滑模控制所存在的成绩;进而,
给出搅扰由函数限制条件下高阶滑模研究的最新停顿,和状况和控制输入受限
情况下的高阶滑模控制成绩。
About the Speaker
丁世宏,江苏大年夜学电气信息工程学院传授,博士生导师,江苏省杰青,江苏
省主动化学会理事。2004 9 月本科卒业于安徽师范大年夜学,2007 2 月和 2010
6 月卒业于西北大年夜学主动化学院,分别获硕士和博士学位。于 2008 9 -
2009 9 月、2011 1 -2011 12 月,2008 7 -8 月,分别在在美国德克
萨斯大年夜学圣安东尼奥分校、澳大年夜利亚西悉尼大年夜学和韩国岭南大年夜学从事协作研究。
AutomaticaIEEE Transactions on Automatic ControlIEEE Transactions on  
Industrial ElectronicsSystems & Control Letters 等控制范畴主流刊物以第一或通
讯作者发表 SCI 检索论文 30 余篇,掌管国度天然迷信基金 2 项,江苏省出色青
年基金 1 项,其他省部级项目多项,取得教导部天然迷信二等奖和机械工业科技
进步二等奖各 1 项。现为国际期刊《International Journal of Intelligent Computing  
and Cybernetics》编委,研究偏向为高阶滑模控制实际、功率变换器、电动汽车

主动安然体系控制等。


Distributed average consensus of continuous-time multi-agent  

systems under G-expectation
刘帅 传授
山东大年夜学
Abstract
Consensus is a fundamental problem in distributed computing and multi-agent  
systems basing each agent interacts with its neighbor agents. Distributed consensus  
control is to design a distributed control protocol for each node such that all of nodes’  
states reach a common decision or agreement. Most existing works were focused on  
discrete-time, continuous-time systems with undirected, directed and time-varying  
digraphs; Some delayed, drop-out situations or additive noises and multiplicative  
noises were considered. However, the nature of measure is non-additivity when  
measurement errors are unavoidable. In this talk, based on the local information, I  
present a novel distributed consensus under G-expectation to explain this  
phenomenon. I will show that different consensus under G-expectation with proposed  
protocol. Convergence rates in both mean square sense and quasi-surly sense are also  
discussed.
About the Speaker
山东大年夜学齐鲁学者特聘传授,博士生导师。分别于 2004 年,2007 年在山东
大年夜学取得工学学士和硕士学位,2012 年取得新加坡南洋理工大年夜学博士学位。2012
年至 2017 年时代,在新加坡-伯克利 (University of California at Berkeley) 结合研
究中间担负高等研究员任务。2017 3 月至今,山东大年夜学控制迷信与工程学院,
山东大年夜学出色青年学者,2017 年当选国度“****”青年项目。