Detection Method for Tolerable False Data Injection Attack Based on Deep Learning Framework
Sizhe He, Yadong Zhou, Xiaoliang Lv, Wei Chen
Abstract
The electric power system is one of the major infrastructures of a country. The ordinary False Data Injection (FDI) is a classical attack on it, which can obstruct normal operation of the electric power system. But the planning and implement of FDI need strict condition. Therefore, we aim at detecting Tolerable False Data Injection (TFDI) attack, which is more likely to implement. Nonetheless, the detection of such an attack is still an unsolved problem . In this paper, we propose a method based on deep learning framework to detect the TFDI attack. We test our model based on the IEEE 14-bus and 30-bus standard datasets for set load condition and mixed load condition. The results demonstrate the good performance of the model distinguishing the TFDI attack. Besides, it has an excellent generalization ability on different load scenarios.