Year: 2025
Type: article
Abstract: ABSTRACT Wind energy plays an essential position in the renewable energy sector and is frequently deployed remotely, which makes them susceptible to intrusions that can compromise their operational system. This paper introduces a novel method T–IDS leveraging the interconnected thermal behaviours of wind turbine modules to identify the abnormal imprints that signify security breaches. Our approach consists of three key components: a graph model that outlines the dependencies among the thermal variables of the turbines, a random forest‐based prediction strategy for these variables within the thermal graph and an anomaly detection method that assesses the predicted thermal values with actual observations. We performed extensive experiments using three real‐world wind turbine supervisory control and data acquisition (SCADA) log datasets: one dataset collected over six months and two additional datasets covering 12‐month operational durations from distinct wind turbine installations for rigorous cross‐validation. The results demonstrate that T–IDS achieves an overall anomaly detection accuracy of 97.3% when detecting unusual thermal activities such as physical model damage leading to overheating or tampering temperature readings.
DOI: 10.1049/cps2.70024
Authors: Ngoc Que Anh Tran, Liang He
Institution: University of Nebraska–Lincoln
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,420
Funder: National Science Foundation
OpenAlex ID: W4411236966