Abstract
The vulnerabilities of power system state estimation have been widely analyzed recently. However, most of the existing attack models can only pass the bad data detection (BDD) in power system state estimation, where the state-of-the-art neural attack detectors (NADs) are not fully considered. To generate a stealthy attack in power system with high-penetrated renewable energy sources (RESs), the principle of constructing a spatiotemporally stealthy attack is analyzed, where a spatiotemporality principle is proposed to ensure the stealthiness of the attack. A spatiotemporal correlation generation framework is proposed in the improved Wasserstein generative adversarial network with gradient penalty (WGAN-GP) framework, which can generate spatiotemporally stealthy false data injection (FDI) attack in power system state estimation deployed with the state-of-the-art NADs. Simulations in the IEEE 14-bus, the IEEE 57-bus, and the IEEE 118-bus test systems verify the stealthiness and effectiveness of the proposed spatiotemporally stealthy FDI attack.
Funding Information
  • Taishan Scholars Program (tsqn202211203, tsqn202408241)
  • National Natural Science Foundation of China (62293502, 62293502)
  • Qilu University of Technology (Shandong Academy of Sciences) “Outstanding Young Talents” (2025QZJH01)
  • General Program of Natural Science Foundation of Shandong Province (ZR2025MS1020)
  • General Program of National Natural Science Foundation of China (62073138)
  • Pairing Project of Qilu University of Technology (2024JDJH06)

This publication has 32 references indexed in Scilit: