False Data Injection Attack on Power System Data- Driven Methods Based on Generative Adversarial Networks

With the rapid development of power systems, power system models have become more and more complex, which limits the application of traditional analysis methods. Thus, data-driven methods play an increasingly important role in the power system. However, due to the shortcomings of data-driven methods such as poor interpretability and strong dependence on input data, the power system data-driven methods are vulnerable to malicious attacks. This paper proposes a false data injection attack strategy for power system data-driven methods based on generative adversarial networks. In this strategy, the generator is used to generate a disturbance vector, which is superimposed on the original measurement input to construct an attack vector, which has a significant impact on the prediction results of the data-driven model. The method described in this article is implemented and verified in new England IEEE 39-bus system, and the experimental results prove the effectiveness of the method.