Article: A fine-tuned YOLOv11-based insulator icing detection algorithm for intelligent inspections of power systems Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.48 pp.1 - 22 Abstract: Ice accumulation on insulators can lead to electrical breakdown, equipment damage, and line outages, making timely and accurate detection essential for maintaining the safe and stable operation of power systems. This paper proposes an ice accretion detection method for insulators based on You Only Look Once version 11 (YOLOv11), integrating image processing and deep learning techniques to achieve automated detection. A self-built dataset was used to fine-tune YOLOv11, enhancing the model's accuracy and robustness in complex environments. Compared to its predecessors, YOLOv11 features an improved backbone network for more efficient feature extraction, advanced attention mechanisms for enhanced focus on critical regions, and an anchor-free detection head that reduces computational complexity while maintaining high precision. Multi-scale feature fusion ensures the accurate detection of ice accretion of various sizes, while dynamic label assignment optimises alignment between predictions and ground truth. Experimental results demonstrate that the fine-tuned YOLOv11-based algorithm achieves high mean average precision (mAP) and F1-scores on the test set, indicating robust detection performance. The proposed method not only enhances detection efficiency but also reduces labour costs, making it well-suited for large-scale power line monitoring. Inderscience Publishers - linking academia, business and industry through research

Title: A fine-tuned YOLOv11-based insulator icing detection algorithm for intelligent inspections of power systems

Authors: Hai Huang; Xun Zhang; Dianli Chen; Yong Du; Shenli Wang; Xiaohua Liu; Quan Fang; Yuhang Xia

Addresses: State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' State Grid Hubei Extra High Voltage Company, Wuhan, 430050, China; Hubei Super-Energic Electric Power Co., Ltd., Wuhan, 430050, China ' Hubei Key Lab of Micro-Nanoelectronic Materials and Devices, Faculty of Microelectronics, Hubei University, Wuhan, 430062, Shanghai, 200240, China

Abstract: Ice accumulation on insulators can lead to electrical breakdown, equipment damage, and line outages, making timely and accurate detection essential for maintaining the safe and stable operation of power systems. This paper proposes an ice accretion detection method for insulators based on You Only Look Once version 11 (YOLOv11), integrating image processing and deep learning techniques to achieve automated detection. A self-built dataset was used to fine-tune YOLOv11, enhancing the model's accuracy and robustness in complex environments. Compared to its predecessors, YOLOv11 features an improved backbone network for more efficient feature extraction, advanced attention mechanisms for enhanced focus on critical regions, and an anchor-free detection head that reduces computational complexity while maintaining high precision. Multi-scale feature fusion ensures the accurate detection of ice accretion of various sizes, while dynamic label assignment optimises alignment between predictions and ground truth. Experimental results demonstrate that the fine-tuned YOLOv11-based algorithm achieves high mean average precision (mAP) and F1-scores on the test set, indicating robust detection performance. The proposed method not only enhances detection efficiency but also reduces labour costs, making it well-suited for large-scale power line monitoring.

Keywords: ice accumulation; insulator icing detection; YOLOv11; ice accretion detection.

DOI: 10.1504/IJICT.2025.150947

International Journal of Information and Communication Technology, 2025 Vol.26 No.48, pp.1 - 22

Received: 18 Jan 2025
Accepted: 19 Mar 2025
Published online: 05 Jan 2026 *

Title: A fine-tuned YOLOv11-based insulator icing detection algorithm for intelligent inspections of power systems

Keep up-to-date