【作者】 孙凯；

【导师】 曲正伟；

【作者基本信息】 燕山大学 ， 电气工程， 2023， 硕士

【摘要】 随着信息化时代到来,电网中应用大量智能采集和通信设备进行数据采集和实时控制,信息系统与电力物理系统高度耦合为电力信息物理系统。然而信息系统的高度开放性给电力物理系统的安全性带来了的威胁,各种网络攻击事件频发并造成了严重的损失。本文针对电力信息物理系统的虚假数据攻击检测和定位开展研究,对于保障电网安全稳定运行具有较强的理论和现实意义。首先,基于耦合性的特征对电力信息物理系统进行建模分析,并从图论的角度对系统拓扑进行简化。对状态估计和虚假数据攻击原理进行建模分析。为了区分虚假数据的类型,将由于设备老化等原因引发的不实数据定义为无意虚假数据注入,将攻击者设计的可以绕过传统坏数据检测单元的恶意虚假数据注入攻击定义为有意虚假数据注入攻击,对基于完整拓扑和不完整拓扑信息所设计的有意虚假数据注入攻击进行建模分析。其次,为了将不同种类的数据攻击类型进行分类识别,首先分析了传统的目标函数检测法和估计分类法的局限性,在此基础上提出了基于欧氏距离的自适应虚假数据检测方法,提高系统攻击识别能力,为系统遭受攻击后作出对不同类型攻击的优先级处理提供理论基础。IEEE 14 bus系统仿真实验验证了所提算法的有效性。最后,考虑到临近节点间的信任度会由于攻击导致级联变动,结合蒙特卡洛定位算法,提出一种改进蒙特卡洛攻击节点定位算法。节点遭受到虚假数据攻击时,其信任度的突变将会影响临近节点的信任度,通过设置锚节点来感知这些变化,并利用该算法对疑似受攻击节点的采样和滤波,完成对受攻击节点的定位。IEEE 118 bus系统仿真实验验证了所提算法的实用性。更多还原

【Abstract】 With the advent of the information age,a large number of intelligent acquisition and communication equipment are applied in the power grid for data acquisition and real-time control,and the information system and the power physics system are highly coupled into the power cyber-physical system.However,the high openness of the information system poses a threat to the security of the power physical system,and various cyber attacks occur frequently and cause serious losses.This paper focuses on the detection and localization of false data attacks in power cyber-physical systems,which has strong theoretical and practical significance for ensuring the safe and stable operation of power grids.Firstly,based on the characteristics of coupling,the power cyber-physical system is modeled and analyzed,and the system topology is simplified from the perspective of graph theory.Modeling and analysis of state estimation and false data attack principles.In order to distinguish the types of fake data,the false data caused by sporadic and aging equipment are defined as unintentional false data injection,the malicious false data injection attack designed by the attacker that can bypass the traditional bad data detection unit is defined as the intentional false data injection attack,and the intentional false data injection attack designed based on complete topology and incomplete topology information is modeled and analyzed.Secondly,in order to classify and identify different types of data attacks,the limitations of the traditional objective function detection method and estimation classification method are analyzed,and on this basis,an adaptive false data detection method based on Euclidean distance is proposed to improve the system attack identification ability and provide a theoretical basis for the priority processing of different types of attacks after the system is attacked.IEEE 14 bus system simulation experiments verify the effectiveness of the proposed algorithm.Finally,considering that the trust between neighboring nodes will cause cascade changes due to attacks,combined with the Monte Carlo positioning algorithm,an improved Monte Carlo attack positioning algorithm is proposed.When a node suffers from a false data attack,the sudden change in its trust degree will affect the trust of neighboring nodes,and these changes are perceived by setting anchor nodes,and the algorithm is used to sample and filter suspected attacked nodes to complete the location of the attacked nodes.IEEE 118 bus system simulation experiments verify the practicability of the proposed algorithm.更多还原