【报告摘要】In many application domains, including biochemical systems, energy systems, and integrated circuits, extracting governing equations from data is a fundamental challenge that underpins effective control and optimization. Compared with classical system identification, network identification is considerably more demanding: it requires collecting data from large-scale interconnected nodes and constructing dynamical models while accounting for intricate interdependencies and correlations across the network. In this presentation, I will briefly introduce several model identification problems arising in large-scale networked systems. In particular, I will focus on the problem of network identifiability, showing how graph-theoretical tools can serve as a key enabler for characterizing when and how network models can be identified. Finally, I will present a method for allocating external excitation signals in a way that enables accurate identification of complex networked system models.

【报告人简介】成晓东博士，安徽人，现任荷兰瓦赫宁根大学应用数学组助理教授，IEEE高级会员。他于2011年和2014年分别在西北工业大学获得学士和硕士学位，于2018年以最高荣誉（cum laude）获得荷兰格罗宁根大学博士学位。2019至2022年间，他先后在荷兰埃因霍温理工大学和英国剑桥大学工程系担任博士后研究员。他的主要研究方向包括系统辨识、模型降阶和数据驱动方法，以及控制理论在农业系统，水和能源中的应用。他曾获得2019年IFAC旗舰期刊《Automatica》的优秀论文奖以及2020年《IEEE Transactions on Control Systems Technology》杰出论文奖。