智能电网中虚假数据注入攻击检测方法研究

发布时间：2018-05-31 21:09

本文选题：智能电网 + 状态估计　；参考：《华北电力大学(北京)》2017年硕士论文

【摘要】：虚假数据注入攻击(False Data Injection Attacks,FDIAs)是针对智能电网中电力系统状态估计的新型攻击手段,是一种典型的数据完整性攻击方式。FDIAs通过篡改电网状态估计值,成功绕过传统不良数据检测机制,使控制中心做出错误决策,引发严重的物理电网故障。研究高效可行的FDIAs检测方法,对构建安全、稳定运行的智能电网信息物理系统具有非常重要的意义。针对FDIAs,本文重点研究其检测方法。通过分析FDIAs原理及国内外研究现状,对现有检测方法从集中式检测和分布式检测两个角度进行对比分析。现有检测方法大多忽略了FDIAs对电网物理特性的影响以及两者之间的关系;当检测到FDIAs时,很少有检测方法提出恢复系统测量量使系统在较短时间内恢复正常运行状态;现有分布式检测方法通常为基于不同地理区域将电网划分为不同子网系统,耗费较高的经济成本。为解决上述问题,本文从集中式和分布式两个角度,分别提出基于节点电压稳定性指标的检测方法和基于矩阵分割的零空间映射两级检测方法。基于节点电压稳定性指标的检测方法中,引入节点电压稳定性指标(Node Voltage Stability Index,NVSI),分析FDIAs对NVSI值的影响。根据系统中节点NVSI值,运用改进的聚类算法对节点进行聚类,辨识节点脆弱性等级;针对脆弱性等级高的节点,提出状态预测检测法实现FDIAs检测。若FDIAs存在,则利用测量量预测值更新测量量,使系统恢复正常运行状态。基于矩阵分割的零空间映射两级检测方法中,分析雅克比矩阵H各行间的线性相关性,设计基于矩阵相似度的矩阵分割算法,对待检测电网系统进行划分;在每个子网系统中,提出零空间映射检测方法实现FDIAs检测;当FDIAs存在时,则利用零空间逆映射方法进行测量量恢复。针对所提两种检测方法,本文运用Matpower电力仿真包,在IEEE 14-bus,IEEE30-bus和IEEE 118-bus标准测试系统中进行仿真实验,验证方法可行性与有效性。实验结果表明,基于节点电压稳定性指标的检测方法在规模较小的系统中计算NVSI值所需时间较短,开销较小,能够快速检测到FDIAs;当系统规模较大时,基于矩阵分割的零空间映射检测法具有较高的检测率,能够有效检测FDIAs。
[Abstract]:False Data Injection attack (false Data Injection AttacksFDIAs) is a new attack against power system state estimation in smart grid. It is a typical data integrity attack. FDIAs tamper with power grid state estimation. By successfully bypassing the traditional bad data detection mechanism, the control center makes the wrong decision and causes the serious physical power grid failure. The study of efficient and feasible FDIAs detection method is of great significance to the construction of a safe and stable intelligent grid information physical system. For FDI Ass, this paper focuses on its detection methods. By analyzing the principle of FDIAs and the current research situation at home and abroad, the existing detection methods are compared and analyzed from two aspects: centralized detection and distributed detection. Most of the existing detection methods ignore the influence of FDIAs on the physical characteristics of the power network and the relationship between the two. When the FDIAs is detected, few detection methods propose to restore the measurement of the system so that the system can return to normal operation in a relatively short time. The existing distributed detection methods are usually based on different geographical regions to divide the power network into different subnet systems, which cost a lot of money. In order to solve the above problems, this paper presents two detection methods based on node voltage stability index and zero-space mapping method based on matrix partitioning from the view of centralization and distribution, respectively. In the detection method based on node voltage stability index, the node voltage stability index (Node Voltage Stability Index) is introduced to analyze the influence of FDIAs on NVSI value. According to the NVSI value of the nodes in the system, the improved clustering algorithm is used to cluster the nodes to identify the vulnerability level of the nodes. For the nodes with high vulnerability level, a state prediction detection method is proposed to realize the FDIAs detection. If FDIAs exists, the measurement is updated by using the predictive value of the measurement to restore the normal operation state of the system. In the zero-space mapping two-level detection method based on matrix segmentation, the linear correlation between lines of Jacobian matrix H is analyzed, and a matrix segmentation algorithm based on matrix similarity is designed to divide the detection grid system. The null space mapping detection method is proposed to realize FDIAs detection, and when FDIAs exists, the null space inverse mapping method is used to recover the measurements. Aiming at the two detection methods mentioned above, this paper uses Matpower electric power simulation package to carry out simulation experiments in IEEE 14-bus IEEE30-bus and IEEE 118-bus standard test system, and verifies the feasibility and validity of the method. The experimental results show that the detection method based on the nodal voltage stability index takes less time and less overhead to calculate the NVSI value in a small system, and it can quickly detect the FDI As.When the system scale is large, The null space mapping detection method based on matrix segmentation has high detection rate and can effectively detect FDIAs.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM76

【参考文献】