Power System Cyber-Attack Event Recognition Method Based on Extreme Random Trees
Xichao Zhao, Xingshen Wei, Zeji Sun, Xiaoming Zhou, Wei Shen, Jian Zhou · 2023
The establishment of a new intelligent control mode for the power system has improved the stability and economy of the power grid, but at the same time, the access of a large number of networked devices has also brought more cyber-attack threats to the power grid. Unlike the impact caused by natural events, the impact of cyber-attacks on the power grid is more persistent and covert, with greater destructive power. Therefore, operators urgently need to accurately identify the types of cyber-attacks suffered by the new power system and make targeted decisions. This article analyzes the public dataset of power grid attacks and proposes a power system anomaly detection model using extreme random tree method. The model has strong random exploration characteristics, making it highly adaptable to solving such problems. Validation experiments were conducted on public datasets, and compared with other machine learning methods, its classification accuracy is high, the false alarm rate is low, and more importantly, its generalization ability is excellent, which can ensure classification effectiveness in complex scenarios.