基于异常根源分析的工业CPS攻击故障辨识方法

Identification method of industrial CPS attack and fault based on anomaly root cause analysis

针对工业信息物理系统(Cyber-Physical System,CPS)面对的网络攻击在现场系统中引起的故障与物理系统偶发故障现象上难以区分的问题。基于贝叶斯概率图,结合路径推理算法,文中提出一种攻击故障辨识方法,以炼化分馏系统作为实验对象,验证了在不同攻击故障场景下的有效辨识能力,与当前数据驱动方法相比不依赖已有数据集,可识别未知攻击故障。

For the cyber-physical system(Cyber-Physical System,CPS),it is difficult to distinguish between the faults caused by the on-site system and the incidental faults of the physical system.Based on the Bayesian probability map and the path inference algorithm,this paper proposes an attack and fault identification method.The refining and chemical fractionation system is used as the experimental object to verify the effective identification ability under different attack fault scenarios.Compared with the current data-driven method,it does not rely on existing data sets and can identify unknown attack and faults.