Martin Gomez, Alvaro (Aalborg University), Hassani, Sina (Department of Electronic Systems, Aalborg University), Wisniewski, Rafael (Section for Automation and Control, Aalborg University)

Reference Model-Based Cyber-Attack Detection for Wind Turbine Systems with Polytopic Uncertainties

Scheduled for presentation during the Regular Session "Fault diagnosis I" (WeAC), Wednesday, June 11, 2025, 10:50−11:10, Room B

33rd Mediterranean Conference on Control and Automation, June 10-13, 2025, Tangier, Morocco

This information is tentative and subject to change. Compiled on May 9, 2025

Abstract

In recent years, wind energy infrastructure has become a growing target for cyberattacks. Developing accurate models for anomaly detection poses significant challenges for non-manufacturing companies, as they often lack access to detailed information about specific system parameters. In such cases, standard models from the literature combined with parameter estimation are typically employed. However, this introduces a certain degree of model uncertainty which must be addressed when designing residual generators to avoid false alarms. To address this, an $mathcal{H}_{infty}$ model-matching problem is formulated by representing the parametric uncertainties as a polytope. The $mathcal{H}_i/mathcal{H}_{infty}$ unified solution for an optimal trade-off filter that balances anomaly sensitivity and disturbance robustness serves as the reference residual generator. These residuals are used as tracking targets in the design of a secondary filter, which minimizes sensitivity to model uncertainty while maintaining effective fault detection. This work evaluates the effectiveness of this technique in a wind turbine aerodynamic subsystem for detecting false data injection attacks. Residuals are assessed using a generalized likelihood ratio test. Additionally, the system is subjected to parametric perturbations during testing to demonstrate the enhanced robustness of the proposed approach against model uncertainties.