Answers are grounded in this paper and cite their sources.
False data injection (FDI) attacks pose a critical threat to power system security by crafting sophisticated attack vectors that evade conventional bad data detection methods. These malicious manipulations corrupt state estimation results, potentially leading to severe operational failures in control centers. To combat this challenge, we present an innovative Generative Adversarial Network framework with Spatial Feature-based Temporal Convolutional Network as the discriminator and Random Forest-Graph Convolutional Generator hybrid model as the generator. The proposed approach leverages a Random Forest-enhanced Graph Convolutional Generator to reconstruct attack-free measurements while employing a Spatial-Temporal Feature-based Discriminator to improve detection accuracy. Through adversarial training, these components synergistically improve both attack detection sensitivity and data reconstruction accuracy. Comprehensive numerical simulations on IEEE 14-bus and 118-bus test systems validate the model’s superior performance, demonstrating significant improvements in both detection robustness and operational resilience against FDI attacks..
No takes yet. Share an insight, caveat, or question.
Wang et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: