Novel data augmentation for improved insulation fault diagnosis under nonideal condition
作者：
Highlights：
• This study uses partial discharge data measured from cable joints as the dataset.
• Partial discharge data was used to train 4 convolutional neural networks.
• The system was trained with clean data but tested with data overlapped with noise.
• Novel data augmentation was proposed to improve performance under noisy condition.
• Proposed data augmentation improved classification performance by 15.83–29.05%.
摘要
•This study uses partial discharge data measured from cable joints as the dataset.•Partial discharge data was used to train 4 convolutional neural networks.•The system was trained with clean data but tested with data overlapped with noise.•Novel data augmentation was proposed to improve performance under noisy condition.•Proposed data augmentation improved classification performance by 15.83–29.05%.
论文关键词：Cable insulation,Partial discharge,Data augmentation,Convolutional neural network
论文评审过程：Received 26 April 2022, Revised 5 July 2022, Accepted 1 August 2022, Available online 5 August 2022, Version of Record 9 August 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.118390
