电力系统恶意攻击检测关键技术研究

发布时间：2018-03-28 17:03

本文选题：电力系统　切入点：隔离装置　出处：《上海电力学院》2017年硕士论文

【摘要】：随着计算机技术和通信技术在电力系统中的广泛应用,以表盘式为特征的传统量测系统逐渐升级为数字化、自动化和高度信息化的现代量测系统。依赖操作人员经验的检测和监测被高效的自动化数据采集和检测系统替代,降低了电力系统人为故障事故的风险,提高了运行的安全性。然而假数据注入攻击的提出和近几年使用网络技术恶意攻击电力系统事件的发生,使电力系统安全面临新的严峻威胁。本文以此为背景,研究了直接攻击电力系统的基于通信协议的网络攻击手段,测试了传统不良数据检测方法对假数据注入攻击的检测效果,最后提出了能够检测假数据注入攻击的基于数据关系图的不良数据检测方法。主要研究成果如下:对电力系统正向隔离装置进行了传输功能测试和各种攻击测试,尝试突破正向隔离装置进入电力系统生产控制区。正向隔离装置在各种攻击测试中正常工作,发挥了隔离作用,不过测试过程中发现了正向隔离装置可能的潜在漏洞。分析了假数据注入攻击的原理,并在四节点16测点模拟系统上实现了注入攻击,同时也证明了基于残差的传统不量数据检测方法无法有效检测出假数据注入攻击。提出了基于数据关系图的不良数据检测方法,利用正常量测条件下状态关系图和含有不良数据或假数据的量测条件下状态关系图的不一致检测并辨识不良数据或假数据发生点。并在四节点16测点模拟系统上对不良数据和假数据注入攻击进行了检测实验,实验表明基于数据关系图的不良数据检测方法能有效避免残差污染和残差淹没现象,正确检测出不良数据和假数据注入攻击。
[Abstract]:With the wide application of computer technology and communication technology in power system, the traditional measurement system characterized by dial type is gradually upgraded to digitization. Automated and highly information-based modern measurement system. The detection and monitoring based on operator's experience is replaced by efficient automatic data acquisition and detection system, which reduces the risk of human fault in power system. However, the false data injection attack and the occurrence of malicious attacks on power system using network technology in recent years make the security of power system face new and severe threat. The network attack method based on communication protocol for direct attack on power system is studied, and the detection effect of traditional bad data detection method on false data injection attack is tested. Finally, a method of detecting bad data based on data relation diagram is proposed, which can detect false data injection attacks. The main research results are as follows: the transmission function test and attack test of forward isolating device in power system are carried out. Attempts to break through the forward isolator into the production control area of the power system. The forward isolator works normally in various attack tests and plays an isolating role. However, the potential vulnerabilities of the forward isolation device were found during the test. The principle of the false data injection attack was analyzed, and the injection attack was realized on the four-node 16-point simulation system. At the same time, it is proved that the traditional data detection method based on residual error can not detect false data injection attack effectively, and a bad data detection method based on data relational graph is proposed. The inconsistency of the state diagram under normal measurement condition and that of measurement condition containing bad data or false data is used to detect and identify the point of occurrence of bad data or false data. The detection experiment of bad data and false data injection attack is carried out. The experimental results show that the bad data detection method based on the data relational graph can effectively avoid residual pollution and residual submergence and detect the bad data and false data injection attack correctly.
【学位授予单位】：上海电力学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP309;TM73

【参考文献】