电力系统状态估计欺诈性数据防御方法

发布时间：2018-11-28 08:03

【摘要】：电力系统状态估计欺诈性数据是黑客在量测数据中恶意注入的以篡改状态估计输出结果为目的的隐蔽性不良数据。鉴于欺诈性数据能够有效躲避传统不良数据检测和辨识,通过构建特定的欺诈性数据,黑客能够蓄意操纵状态估计输出结果,进而导致错误调度指令的产生,威胁电力系统的安全可靠运行。因此,在确保未来智能电网更好服务于国民经济发展的层面上,研究实际电力系统中存在的数据安全漏洞,并制定相应的防御措施具备迫切的现实意义。根据已有文献所提防御方法的防御效果,将欺诈性数据防御方法分为检测、辨识和遏制3类。检测法仅能够判定欺诈性数据是否存在,但无法锁定欺诈性数据的具体位置;辨识法进一步实现了对欺诈性数据所在位置的锁定。由于遏制法着眼于从根本上瓦解黑客构建欺诈性数据的可能性,因此,研究中将遏制法视为保障电力系统状态估计结果可靠性最具成效的防御途径。量测信息物理保护法、网络参数动态调节法、拓扑结构动态调节法以及电力信息融合防御法是现阶段提出的4种欺诈性数据遏制法,但在应用过程中难免面临设备投资较高、全局输电能力受限、计算复杂度较高等因素的制约。为克服上述防御算法的不足,本文提出以完善不良数据检测和辨识算法固有数据安全漏洞为核心的欺诈性数据遏制法,论文工作及主要研究成果如下:1、系统性综述了电力系统状态估计欺诈性数据防御法的研究成果。在总结现有欺诈性数据防御法不足的基础上,指出完善不良数据检测和辨识算法固有的数据安全漏洞是防御欺诈性数据的核心。2、提出基于状态向量挖掘的欺诈性数据防御法。采用隐马尔科夫模型建模历史运行状态数据库,将黑客难以操纵的发电机输出功率视为观察状态序列,并通过维特比算法解码系统状态序列。3、提出基于状态变量修正的欺诈性数据防御法。提出状态一致性检验用以锁定疑似受到欺诈性数据影响的状态变量,并通过所构建的状态变量修正向量打破欺诈性数据的隐蔽性,使其无法躲避不良数据检测和辨识算法,从而实现对欺诈性数据的辨识和剔除。4、采用IEEE-14节点和IEEE-118节点标准测试系统仿真实验验证所提出欺诈性数据防御法遏制欺诈性数据的有效性。
[Abstract]:Power system state estimation fraud data is the hidden bad data that hackers inject maliciously into the measurement data to tamper with the output of state estimation. Since fraudulent data can effectively avoid traditional bad data detection and identification, by constructing specific fraudulent data, hackers can deliberately manipulate the output of state estimation, which leads to the generation of false scheduling instructions. It threatens the safe and reliable operation of power system. Therefore, in order to ensure the future smart grid to better serve the development of the national economy, it is of urgent practical significance to study the data security vulnerabilities in the actual power system and to formulate corresponding defense measures. According to the defensive effect of the methods mentioned in the literature, the methods of fraudulent data defense are divided into three categories: detection, identification and containment. The detection method can only determine whether the fraudulent data exists, but it can not lock the location of the fraudulent data. Because containment method focuses on the possibility of disintegrating hackers to construct fraudulent data, containment method is regarded as the most effective defense way to ensure the reliability of power system state estimation results. Physical protection of measurement information, dynamic adjustment of network parameters, dynamic adjustment of topological structure and protection of power information fusion are four kinds of fraudulent data containment methods proposed at present, but it is inevitable to face high investment in equipment in the process of application. The global transmission capacity is limited and the computational complexity is high. In order to overcome the shortcomings of the above defense algorithms, this paper proposes a method of fraudulent data containment based on the improvement of bad data detection and identification of inherent data security vulnerabilities. The main results of this paper are as follows: 1. The research results of power system state estimation fraudulent data defense method are systematically reviewed. On the basis of summing up the deficiency of the existing fraudulent data defense methods, it is pointed out that improving the data security holes inherent in the bad data detection and identification algorithm is the core of the defense against fraudulent data. A state vector mining based data defense method is proposed. Using hidden Markov model to model the historical running state database, the generator output power which is difficult to manipulate by hackers is regarded as the observation state sequence, and the system state sequence is decoded by Viterbi algorithm. In this paper, a method of data fraud defense based on state variable correction is proposed. A state consistency test is proposed to lock the state variables that are suspected to be affected by fraudulent data, and to break the concealment of the fraudulent data by modifying the vector of the state variables, which makes it unable to avoid the bad data detection and identification algorithm. Finally, the identification and elimination of fraudulent data are realized. 4. The effectiveness of the proposed method is verified by using IEEE-14 node and IEEE-118 node standard test system.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM732

【参考文献】