一、报告题目
Distributed Kalman Filters over Multi-Agent Systems
二、报告人
何兴康
三、报告时间
6月29日（星期五）下午3：00
四、报告地点
软件园校区微电子学院105
五、报告人简介
Xingkang He received the B.S. degree in School of Mathematics from Hefei University of Technology in 2013, and the Ph.D. degree in Academy of Mathematics and Systems Science, Chinese Academy of Sciences at Beijing in 2018.
Dr. He received the National Scholarship for Doctor in 2017 and the 2018 Excellent Graduate of Beijing. He received the 2018 Best Paper Award of Data Driven Control and Learning Systems Conference. His research interests include filtering theory, distributed Kalman filter, event-based state estimation and multi-agent systems.
六、报告摘要
This talk will introduce some of his recent research results on distributed filtering (or distributed state estimation) over multi-agent networks. In particular, the basic distributed filtering problem will be first discussed and addressed in the time-driven communication scheme of agents. To tackle the unknown correlation between estimates of agents, a scalable and fully distributed recursive filter, named consistent distributed Kalman filter (DKF), is provided to achieve dynamic estimation for the states of potentially unstable systems. Under very mild conditions, including the global observability of the system and the connectivity of the multi-agent network, the stability of the proposed filter is analyzed. Then, to avoid the redundancy communications between agents, a novel event-triggered communication scheme is studied. Based on the event-triggered scheme, an event-triggered DKF is put forward and further analyzed theoretically in terms of the consistency and stability of estimation. Finally, for a class of systems with state equality constraints (SEC), a fusion-projection based DKF is investigated. Based on the new-defined extended global observability condition, which is milder than existing conditions, the stability of the filter is studied. Besides, it is shown that the global SEC are satisfied at each agent in two limitation cases.
