Localizing detection of false data injection attack in CPPS based on CFSSA improved multi-layer extreme learning machine

The open environment of cyber-physical power systems (CPPS) exposes the safe operation of power grids to the threat of false data injection attacks (FDIA). Most existing detection methods for such cyberattacks suffer from limitations such as insufficient feature learning ability, slow detection speed, and inability to accurately identify attack locations. Therefore, this paper proposes a FDIA localizing detection method based on sparrow search algorithm with circle chaos initialization and firefly disturbance strategy improved multi-layer extreme learning machine (CFSSA-ELM<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ML</sup>). The multi-layer extreme learning machine (ELM<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ML</sup>) is adopted as the deep feature extraction model and the basic classifier to address the limitation of insufficient feature learning ability of existing detection methods. Moreover, an improved sparrow search algorithm (CFSSA) with strong local search ability is utilized to optimize the initial weight and bias of the multi-layer extreme learning machine, enabling fast and accurate localization of false data injection attacks. The effectiveness of the proposed method is verified through simulations on the IEEE 14-bus and 57-bus systems. It is verified that the proposed method results in better location performance evaluation criteria compared with other detection methods.