Transmission lines, the core part of the transmission and distribution system in the smart grid, require effective, efficient, and reliable protective measures against faults to avoid severe damage to physical infrastructure and financial losses. Due to their growing popularity, machine learning models are used in fault detection and classification, whose performances can be severely affected by cyber‐attacks due to their data dependency, posing a critical concern. Hence, this paper introduces false data injection attacks to address the vulnerability of machine learning‐based fault detectors and classifiers. A comparative study of 9 detection models and 6 classification models under normal conditions and during a combination of two models of false data injection attacks is presented to evaluate the severity of cyber‐attacks. Experimental results show that highly accurate models in normal conditions are more susceptible to cyber‐attacks, with up to 69% and 28% degradations in accuracy for fault detectors and classifiers, respectively. Furthermore, the detection models are found to be more vulnerable to cyber‐attacks than the classification models. With no robust detectors and classifiers being found, this work addresses the importance of developing attack‐resilient fault detection and classification schemes considering their academic and industrial significance.
28 April 2025 in IEEE Transactions on Dielectrics and Electrical Insulation
This work addresses the dual challenge posed by label-flipping attacks and high data imbalance in Dissolved Gas Analysis (DGA)-based Power Transformer Fault Diagnosis (PTFD). While existing literature focused on improving the performance of data-driven methods by utilizing clean training samples with a combination of data-balancing methods, however, to date, the combined challenges of data imbalance and label-flipping attacks in the training phase of Machine Learning (ML) and Deep Learning (DL) models remain unaddressed. Therefore, this work bridges this gap by introducing a customized and extended semi-supervised learning framework. The proposed method identifies potentially correct and incorrectly labeled samples, processes them with Gaussian augmentation to handle the unlabeled data, and incorporates a re-weighting mechanism to handle class imbalance. Experimental results demonstrate that the existing data balancing methods fail to provide support for ML and DL models under label-flipping attacks, and the proposed method achieves an overall accuracy of 90%, a precision of 92%, and an F1-score of 89%, significantly outperforming state-of-the-art ML and DL models. Under label-flipping attacks and data imbalance scenarios, the proposed method demonstrates at least a 12.5% improvement in accuracy, 8.2% in precision, and 14% in F1-score, reflecting a significant relative performance gain over state-of-the-art models in such adversarial scenarios. Moreover, the proposed method exhibits resilience to Gaussian noise and False Data Injection Attacks (FDIA), ensuring robust DGA-based PTFD against physical and cyber anomalies..
Accurate and efficient insulator defect detection is critical for power grid reliability, but it can be affected by the presence of noises in captured images and can be difficult to employ for real‐time operation due to the slow processing of the detection scheme. This paper proposes a novel framework based on the YOLOv8x object detection scheme, addressing the challenge of detecting small defects in complex aerial images and providing a noise mitigation scheme. A Gaussian blur and Laplacian sharpening‐based hybrid scheme is proposed to mitigate the impacts of noises in insulator images. Experimental results indicate that the proposed framework can achieve a mean average precision (mAP) of 98.4% on noise‐free images, surpassing benchmark models, such as YOLOv5x and YOLOv7 by 2.1% and 3.9%, respectively. Also, while the performance of a conventional system can decrease to a mAP of 93.3% in the worst case, the implementation of the proposed mitigation scheme ensures a mAP of 96.7% for that case. With an inference speed of 56.9 ms per image, this approach offers a promising solution for real‐time power line inspection, contributing to enhanced power grid maintenance and safety.
1 January 2025 in International Journal of Energy Research
Accurate short‐term solar power forecasting (nowcasting) facilitated by smart devices and cyberinfrastructure, which uses sky images and artificial intelligence (AI)–based models, is susceptible to cyberattacks. This study investigates the vulnerabilities of deep learning (DL) and artificial neural network (ANN)–based sky image–based nowcasting models to adversarial attacks such as fast gradient sign method (FGSM), projected gradient descent (PGD), and a mixed attack template, along with proposing a feature extraction–based multi‐unit solar (FEMUS)‐Nowcast model. Results reveal that adversarial attacks significantly degrade all models’ accuracy and lead them to an unusable state. Moreover, FGSM is found to be the most severe attack, with root mean square error (RMSE) increasing by 5–16 times and mean absolute error (MAE) increasing by 4–12 times compared to the normal scenario under maximum perturbation. As the proposed FEMUS‐Nowcast outperforms models of existing literature, reducing RMSE by 48% and 25% under normal conditions, adversarial training is adapted to enhance its robustness in the presence of cyberattacks. Furthermore, adversarially trained (AT) FEMUS‐Nowcast shows no RMSE or MAE trade‐offs under all scenarios. Additionally, the AT FEMUS‐Nowcast model demonstrates high resilience against advanced attacks, including iterative FGSM (I‐FGSM) and momentum I‐FGSM (MI‐FGSM), confirming its reliability and robustness across diverse attack scenarios.
16 January 2025 in Signal Processing: Image Communication