基于马尔科夫模型的分布式电网CPS网络攻击动态检测

Dynamic Detection of CPS Network Attacks in Distributed Power Grid Based on Markov Model

传统信息物理系统(cyber physical systems, CPS)网络攻击检测方法存在隐性攻击因子漏检的情况。为此引进马尔科夫模型,设计一种针对分布式电网CPS网络攻击的动态检测方法。考虑到分布式电力资源属于隐性资源,因此要在马尔科夫模型的应用下,对隐性参数进行提取,并利用模型的统计功能,对攻击因子进行总结与归纳。同时,绘制CPS网络攻击因子概念模型图,以此为参照,描述CPS网络攻击下分布式电网运行,并结合攻击的可检测性,定位不同CPS网络攻击因子的入口。此外,通过对比实验证明,设计的检测方法可以准确地检测不同类型CPS网络攻击因子,平均检测准确率可高达96.0%。

The traditional network attack detection methods of cyber physical systems(CPS) have the problem of missing hidden attack factors.Therefore, Markov model is introduced to design a dynamic detection method for distributed power grid CPS network attack.Considering that the distributed power resources belong to recessive resources, it is necessary to extract the recessive parameters under the application of Markov model, and summarize the attack factors by using the statistical function of the model.At the same time, the conceptual model diagram of CPS network attack factor is drawn to describe the operation of distributed power grid under CPS network attack.Combined with the detectability of attack, the entrance of different CPS network attack factors is located.In addition, comparative experiments show that the designed detection method can accurately detect different types of CPS network attack factors, and the average detection accuracy can be as high as 96.0%.