Research On Detection And Defense Methods Against False Data Attack On Smart Grid

The future SG (Smart Grid) will provide reliable, inexpensive and sustained power services. The Control Center of SG can make a series of intelligent decision to maintain the stability and security power grid responding to diverse power consumption and urgent emergence. The EMS (Energy Management system) and SCADA (Supervisory Control and Data Acquisition) are the core components of smart grid and give real-time orders to maintain the normal function of SG.The FDA (False Data Attack) is discovered in recent years, which introduce great errors into PSSE (Power System State Estimation), core component of SCADA. It can evade the detection of system, influence the state values estimated by PSSE and finally lead the wrong decision made by the Control Center. The attack is primarily introduced into linear DC power system but has extended to AC power system and more complicated PSSE. Current detection and protection methods against the specific attack are initialized and require more research to maintain the security of SG. It is very important and significant to research on FDA and come up with defense methods.The paper concentrates on the FDA problem in SG, including:1. The paper introduces the basic principle of FDA and analyzes the attack strategies and the impact of FDA on SG under different conditions. The paper also studies the FDA on other components of SG.2. Considering data detection, the paper undertakes research on detection method based on historical data and using the conventional statistic such as measurements residuals to protect the power system. Then by analyzing the impact of injection data, State Summation Detection method are proposed to detect the attack using state variables’ distributions. In addition, we formalize the problem as a hypothetical test of standard normal distribution with empirical data and compare with traditional bad data detection methods.3. The paper analyzes the conditions on which the FDA depends if the attacker conducts a successful attack, and proposes a protection method of securing a set of measurements to prevent the attack to the extent that the optimally selected and complementary sets can be implemented to be resistant to power system’s topology changes and failure. 4. We demonstrate the effectiveness of our detectors comparing with traditional detectors on IEEE14-bus and118-bus systems.