Dual-CyCon Net: A Cycle Consistent Dual-Domain Convolutional Neural Network Framework for Detection of Partial Discharge

  • 2021-10-19 17:27:19
  • Mohammad Zunaed, Ankur Nath, Md. Saifur Rahman
  • 0

Abstract

In the last decade, researchers have been investigating the severity ofinsulation breakdown caused by partial discharge (PD) in overhead transmissionlines with covered conductors or electrical equipment such as generators andmotors used in various industries. Developing an effective partial dischargedetection system can lead to significant savings on maintenance and preventpower disruptions. Traditional methods rely on hand-crafted features and domainexpertise to identify partial discharge patterns in the electrical current.Many data-driven deep learning-based methods have been proposed in recent yearsto remove these ad hoc feature extraction. However, most of these methodseither operate in the time-domain or frequency-domain. Many research approacheshave been developed to generate phase-resolved partial discharge (PRPD)patterns from raw PD sensor data. These PRPD diagrams suggest a correlationbetween partial discharge activities occurring in an alternating electricalwaveform's positive and negative half-cycles. However, this correlationcriterion between half-cycles has been remained unexplored in deeplearning-based methods. This work proposes a novel feature-fusion-basedDual-CyCon Net that can utilize all time, frequency, and phase domain featuresfor joint learning in one cohesive framework. Our proposed cycle-consistencyloss exploits any relation between an alternating electrical signal's positiveand negative half-cycles to calibrate the model's sensitivity. This lossexplores cycle-invariant PD-specific features, enabling the model to learn morerobust, noise-invariant features for PD detection. A case study of our proposedframework on a public real-world noisy measurement from high-frequency voltagesensors to detect damaged power lines has achieved a state-of-the-art MCC scoreof 0.8455.

 

Quick Read (beta)

loading the full paper ...