基于鲁棒性能的信息物理融合系统乘性攻击检测

Robust Performance Based Multiplicative Attack Detection for Cyber-Physical Systems

中文关键词  信息物理融合系统  攻击检测  稳定裕度  数据驱动

英文关键词  Cyber-physical systems  attack detection  stability margin  data-driven

中文摘要

      针对信息物理融合系统乘性攻击检测的难题，本文提出两种基于鲁棒性能的乘性攻击检测与数据驱动实现策略. 首先, 利用互质分解和间隙度量理论, 建立信息物理融合系统和乘性攻击模型. 其次，利用稳定裕度性能指标, 评估乘性攻击对闭环系统稳定性能的影响，给出了稳定裕度的下界. 再次，提出基于稳定裕度和基于残差鲁棒性能的乘性攻击检测策略，并设计相应的检测阈值和检测逻辑. 进一步，利用子空间辨识方法，在线辨识稳定裕度和残差鲁棒性能，提出基于鲁棒性能的乘性攻击检测的数据驱动实现方法. 最后, 利用飞行器系统的仿真, 验证了所提出的基于鲁棒性能的乘性攻击检测方法的有效性.

英文摘要

      To address the problem of multiplicative attack detection in closed-loop cyber-physical systems, this paper proposes two methods of robust performance based multiplicative attack detection and their data-driven realization. First, apply the coprime factorization and gap metric to establish the models of cyber-physical systems and multiplicative attacks. Then, the stability margin is used to evaluate the effects of multiplicative attacks on the stability performance of closed-loop systems, and the low bound of stability margin is presented. Furthermore, the stability margin based and residual robust performance based attack detection schemes are proposed, and the corresponding detection thresholds and detection logics are designed. Moreover, with the aid of subspace identification, the data-driven realizations of the robust performance based attack detection are presented, via online identifying the stability margin and residual robust performance. Finally, the simulations on flight vehicle system are applied to verify the effectiveness of the robust performance based multiplicative attack detection methods.