网络攻击下产品与供应链协同演进系统数据驱动变更控制设计  

Data-driven change control design for product and supply chain synchronous evolution systems under cyber-attacks

作　　者：李庆奎[1] 高雪峰 彭晨[2] 张蕴隆 易军凯 Qingkui LI;Xuefeng GAO;Chen PENG;Yunlong ZHANG;Junkai YI(School of Automation,Beijing Information Science and Technology University,Beijing 100192,China;School of Mechanical Engineering and Automation,Shanghai University,Shanghai 200072,China)

机构地区：[1]北京信息科技大学自动化学院,北京100192 [2]上海大学机电工程与自动化学院,上海200072

出　　处：《中国科学：信息科学》2023年第2期325-343,共19页Scientia Sinica(Informationis)

基　　金：国家重点研发计划(批准号:2020YFB1708200)资助项目。

摘　　要：产品与供应链协同演进系统(product and supply chain synchronous evolution system,PSCSE)是一类复杂的分布式信息物理系统(cyber-physical system,CPS),含有大量的未建模动态与不确定性.剧烈的需求波动及突发的网络事件,使系统结构参数极易发生变化.产品设计变更是维持PSCSES稳定、满足用户需求及保证经济效益的重要手段.本文研究基于分布式CPS的PSCSES在受到DoS攻击下的应急变更控制问题.首先,针对PSCSES在网络攻击下数据包丢失问题,利用每条子链的历史数据设计基于循环神经网络(recurrent neural network,RNN)的预测器以弥补因网络攻击造成的库存数据丢失;其次,利用博弈论思想将H∞一致性控制问题转化为多人零和图博弈问题,提出一种应急变更补偿机制;进而基于Q-learning的策略迭代技术设计了模型未知的控制器求解算法,实现了系统的库存状态H∞一致性.最后,仿真实验验证了所提方法的有效性.A product and supply chain synchronous evolution system(PSCSES)is a kind of complex distributed cyber-physical system(CPS),which contains several unmodeled dynamics and uncertainties,and the system structure and parameters are vulnerable to changes due to drastic demand fluctuations and unexpected network events.Thus,a product design change is an important way to maintain the stability of the synchronous evolution system to meet customers’demands and ensure economic efficiency.In this paper,the emergency change control problem of a distributed CPS-based PSCSES under DoS attacks is studied.First,an emergency change compensation mechanism in which a predictor based on recurrent neural networks is proposed to compensate for the lost inventory data caused by cyber-attacks.Second,the H∞consensus control problem is transformed into a multiplayer zero-sum graph game using game theory,and a Q-learning-based H∞consensus control protocol is designed,such that sufficient conditions that guarantee the consensus of the system can be derived.The bullwhip effects caused by the customers’demand and design change are analyzed thereby.Third,using the strategy iteration algorithm,the online solution is resolved for the controller design under the unknown model.Finally,the effectiveness of the proposed method is verified through the simulation of a collaborative production supply chain.

关 键 词：网络攻击 数据驱动 供应链 变更控制设计 多智能体 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术] F274[自动化与计算机技术—计算机科学与技术]