Segmentation Method for Infrared Images of Substation Equipment Based on DeepLabv3+ Neural Network

This study focuses on developing a high-precision image segmentation method for substation equipment using the DeepLabv3+ neural network. With the increasing automation level of substations, accurate monitoring of equipment status and fault prediction becomes increasingly important. Critical equipme...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2024 5th International Seminar on Artificial Intelligence, Networking and Information Technology (AINIT) S. 1947 - 1951
Hauptverfasser: Xuan, Wenchao, Wang, Hongliang, Peng, Haichao, Chen, Hongtao, Guo, Yanchun, Wang, Xianda, Song, Lin, Wang, He
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 29.03.2024
Schlagworte:
Accuracy
Current transformers
Deep learning
DeepLabv3
Image segmentation
Neural networks
Seminars
Substation equipment
Substations
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This study focuses on developing a high-precision image segmentation method for substation equipment using the DeepLabv3+ neural network. With the increasing automation level of substations, accurate monitoring of equipment status and fault prediction becomes increasingly important. Critical equipment such as current transformers directly affect the safe and stable operation of the power grid. Traditional manual detection methods have many limitations in terms of accuracy and efficiency, while deep learning provides an effective automated solution. In this study, the advanced image segmentation network DeepLabv3+ in deep learning is employed to process infrared images of substation equipment, leveraging its excellent feature extraction and segmentation capabilities. Through the collection and annotation of a large number of current transformer images, a model capable of accurately identifying and segmenting key parts of the equipment was trained[1]. The successful implementation of this study not only showcases the enormous potential of deep learning technology in substation equipment monitoring but also provides new insights and methods for the intelligent management and maintenance of power systems in the future. Future work will focus on further optimizing model performance, expanding to more types of power equipment, and exploring its application in real-time monitoring systems.
DOI:10.1109/AINIT61980.2024.10581821