偽データ注入攻撃下の非線形浮体式洋上風力タービンの信頼性ゲート付きGRUベースのセキュア制御

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:Provides a novel application of GRU-based reliability gating for secure control of nonlinear FOWT under FDI attacks, offering insights into cyber-physical security for renewable energy systems.

🏢実務担当者:Offers a framework for enhancing the cybersecurity of offshore wind turbines, which can be integrated into control systems to ensure reliable operation and protect against sensor attacks.

🏛政策担当者:Highlights the importance of cybersecurity measures in renewable energy infrastructure, informing policies for securing critical energy assets and ensuring reliable power generation.

<title>Abstract</title> <p>The offshore floating wind turbines are coupled with strong wind-wave disturbances and need to rely on feedback from the sensors for the regulation of rotor speed, monitoring of the platform motion, and power regulation. But the FDI attacks may taint the critical measurements, such as rotor speed, platform pitch and generated power, resulting in poor feedback-dependent control behaviour and inaccurate operational assessment. In this paper, a secure-control framework based on a reliability-gated gated recurrent unit (GRU) is proposed to control a nonlinear model of a floating offshore wind turbine (FOWT) designed based on a reference turbine from the NREL (National Renewable Energy Laboratory) 5-MW reference series. The proposed framework combines a residual based attack detection, GRU based safe sensor prediction, channel-wise reliability assessment and reliability gated compensation prior to the corrected measurements being provided to a baseline Region II/III controller. The proposed reliability gate decreases over-reliance on the weakly predicted channels of the sensors and conservatively corrects the uncertain feedback from the platform-motion channels. A set of false measurements (rotor speed, platform pitch, and generated power biases) is added to the various measurement channels during the time period between 120 and 210 s. The simulation results indicate that the total sensor level RMSE of the reliability-gated GRU compensation is 85.95%, of which the rotor speed error is 85.71%, the platform pitch error is 50.01%, and the generated power is 86.33%. Closed-loop secure-control validation results show that the proposed method can detect the attack with 97.27% accuracy, and the RMSE of the rotor-speed and generated-power decreases by 26.96% compared with the unprotected attacked case with the platform-pitch response being almost unchanged. Reliability gating is also robust and ablator robust to show a good trade-off between sensor recovery, prediction uncertainty, and closed-loop control safety. The results show the capability of reliability-aware learning compensation to recover sensors securely and control rotor speed/power in nonlinear floating wind turbine systems when subject to cyber-physical sensor attacks.</p>

• Research Square https://doi.org/10.21203/rs.3.rs-10760246/v1first seen 2026-09-05 04:47:51 · last seen 2026-09-18 04:20:58