False data injection attack detection in smart grid based on CNN-BiGRU
Yuan Zeng, Yue Huang, Yaling Chen · 2024
False Data Injection Attacks (FDIAs) are a common security threat to power systems, where an attacker injects false data into the system to interfere with its normal operation. Most of the current research on false data injection attack detection has been conducted in the physical grid environment, while relatively little research has been conducted on new and advanced sensing and measurement technologies to achieve the efficient and smooth operation of smart grids. In this paper, we first provide a comprehensive understanding of FDIAs and investigate the incidents and impacts of FDIA on power systems worldwide. By searching the literature, the existing detection methods for FDIAs are collated, and it is found that most of the existing detection methods use rules or statistical models, which are vulnerable to deception by attackers. Secondly, CNN is used to reduce the dimensionality and extract the spatial features of the high-dimensional measurement information of the power system in the smart grid, and then combined with BiGRU to fully explore the time series features in the information, thus proposing a CNN-BiGRU based detection method of FDIAs in smart grids, and then detecting the measurement information input model of power system and classifying it. This paper also conducts comparison experiments with three commonly used algorithms, CNN, RNN and MLP, and obtains that the research in this paper has high accuracy and certain feasibility.