A novel insulator defect detection scheme based on Deep Convolutional Auto‐Encoder for small negative samples

This paper presents a novel insulator defect detection scheme based on Deep Convolutional Auto-Encoder (DCAE) for small negative samples. The proposed DCAE combines the advantages of supervised learning and unsupervised learning. In order to reduce high cost training Neural Networks, this pre-trained Networks (CNN) through open labelled datasets. Through transferring learning, encoder part traditional was replaced by first three layers CNN, number samples were used fine-tune parameters. A threshold discrimination designed evaluate model detection, realising self-explosion judging residual result abnormal score. experimental results show that compared with existing schemes, can time up 40%, recognition accuracy reach 97%. Moreover, does not need large data is especially suitable sample application.