A Three-Stage Dynamic Game Strategy Incorporating Wind Generation Uncertainty Against FDI Attack of Wind Farm Reactive Power Compensation

This is the authors' abstract. We don't add key points for this paper.

The integration of large-scale wind generation further accentuates the issues of system voltage fluctuations and the increase of line loss. In addition, the uncertainty of wind generation could potentially undermine the efficacy of traditional bad data detection techniques, posing challenges for power gird defenders in accurately assessing the data quality of compensation equipment, thereby increasing the risk of cyber attacks on system. In this paper, based on a proposed day-ahead attack and defense funding allocation strategy, a novel three-stage defender-attacker-operator dynamic game model incorporating wind generation uncertainty is established, aiming to minimize the energy loss caused by false data injection attacks against reactive power compensation equipment of wind farm. In order to effectively address the proposed three-stage dynamic game framework, an equivalent reformulation of a tri-level mixed integer nonlinear mathematical programming model is established. To tackle the middle-lower-level optimization model, a fast self-adaptive piecewise-linear approximation algorithm is proposed associated with the semi-definite relaxation technique. Additionally, a hybrid heuristic search algorithm is devised to solve the upper-level problem. Finally, the validity of the proposed optimization model and solution method is then demonstrated by conducting simulation analysis on the modified IEEE 14-node test system and a real-sized power grid in China.