Research On False Data Attack Models And Defense Strategies

The enhancements of advanced power electronic technology,information technology and intelligent management enable smart grid to develop gradually to its version 2.0: Energy Internet.Malicious attackers,however,who have ability to deploy vulnerability of traditional power grid can successfully launch attack to Energy Internet.Once the attack is implemented,it might lead to the collapse of the whole energy system,and the new method of cyber-attack,false data injection attack,can successfully bypass the traditional bad data filter,launching catastrophic attack on smart grid effectively.Therefore it is extremely important for us to research the characteristic of false data injection attack and its defense methods.This paper propose twoattack models in terms of the characteristic of false data attack.Firstly,scenario based complete and incomplete network information minimal attack models are proposed.The model considers the correlation between electric load pattern and time,designing different attack strategies according to different operation scenarios to thus find the optimal attack strategy from attackers' perspective.Then,multi scenarios based dynamic attack model is proposed because normal attack models generally consider single scenario.The proposed attack model gradually tampers power flow data to make the launched attack hardly to be distinguished by most defense methods,thereby launching attack successfully and stealthily.With respect to defense strategies,interval state estimation is proposed firstly.This method consider the uncertainty of electric load forecasting to assess state variables of power system,thus quantifying the exterior bounds of state variables.Then,interval state estimation is utilized to design a cyber-attack defense mechanism.Generally speaking,the fluctuates of state variables might be bigger than normal when an attack is launched,and state variables will possibly not fall in the bound of the proposed state internal.Therefore,checking the state variables whether in the range of state interval and thus to determine whether there is an attack or not.Further,this paper also considers the influence of uncertainty of parameter perturbations due to renewable energy sources,lines aging,climate changes and so on to construct an interval state estimation considering the uncertainty of parameters,and it is applied to cyber-attack defense mechanism as well.Lastly,simulations and analyses are carried out in terms of proposed attack models and defense strategies.We collect the electric load data of Dongguan city and apply it to IEEE benchmarks to make results more realistic.Final results have demonstrated that proposed attack models can successfully bypass the traditional detection and launch attack to power system,meanwhile proposed defense strategies can also detect performed attacks effectively.