Computation of Worst-case Operation Scenarios against False Data Injection Attacks Considering Load Demand and Generation Uncertainties

False data injection attack (FDI) is one of the most malicious cyber attacks in power systems. However, uncertainties of power systems are neglected in existing works. Focusing on this issue, a bilevel optimization approach to model the worst-case operation of power systems against FDI attacks is proposed, where the attacker is in the upper level to maximize power flow of target line and the control center is in the lower level to adjust generation outputs according to received measurement data. In this model, uncertainties of renewable generations and load demands are mainly considered in the upper level. Then the worst-case operation scenario is determined by transforming the model into a single mixed-integer linear programming problem based on Karush-Kuhn-Tucker (KKT) conditions. The numerical simulations on the modified IEEE 14bus system show that coordination of fluctuations and FDI attacks has more serious effects on the power flow of a target line than only fluctuations or FDI attacks. It is also found that reducing proportion of renewable generations and increasing proportion of fluctuated loads make lines overloaded more easily. Our analysis provides an meaningful insight on the effects of FDI attacks when uncertainties are considered.