Proceedings of the First International Conference on Computing, Communication and Control System, I3CAC 2021, 7-8 June 2021, Bharath University, Chennai, India

Research Article

Transmission Power Line Fault Detection using Convolutional Neural Networks

Download38 downloads
  • @INPROCEEDINGS{10.4108/eai.7-6-2021.2308661,
        author={Kalanidhi  K and Baskar  D and Vinod  Kumar D},
        title={Transmission Power Line Fault Detection using Convolutional Neural Networks },
        proceedings={Proceedings of the First International Conference on Computing, Communication and Control System, I3CAC 2021, 7-8 June 2021, Bharath University, Chennai, India},
        publisher={EAI},
        proceedings_a={I3CAC},
        year={2021},
        month={6},
        keywords={transmission lines fault analysis cnn alexnet vgg16 resnet},
        doi={10.4108/eai.7-6-2021.2308661}
    }
    
  • Kalanidhi K
    Baskar D
    Vinod Kumar D
    Year: 2021
    Transmission Power Line Fault Detection using Convolutional Neural Networks
    I3CAC
    EAI
    DOI: 10.4108/eai.7-6-2021.2308661
Kalanidhi K1,*, Baskar D2, Vinod Kumar D2
  • 1: Research Scholar - Electrical & Electronics Engineering, 3Professor & Head - Biomedical Engineering,Vinayaka Mission's Kirupananda Variyar Engineering College, VinayakaMission’sResearch Foundation (Deemed to be University), Salem, Tamil Nadu, India.
  • 2: Assistant Professor - Electrical & Electronics Engineering, Annai Teresa College of Engineering, Viluppuram, Tamil Nadu, India.
*Contact email: kalanidhik@gmail.com

Abstract

In an electrical power system, most of the faults occurs in overhead transmission lines because of most of the conductor exposure to the atmosphere. Therefore, Insulated Overhead Conductors (IOCs) are widely used. To overcome this, a robust real-time PD fault analysis system is required. To analyze and classify the raw voltage signal for detection of PD's in IOC's a Convolutional Neural Network (CNN) based fault classification algorithm is proposed in this paper. The CNN is implemented using popular pre-trained CNN architectures such as AlexNet, VGG16 & ResNet are applied to the voltage signals in the dataset. From the values of Precision, Recall & F1-Score it is observed that ResNet architecture provides the best prediction and classification results.