Aiming at the problem of slow speed of electrical equipment diagnosis and high dependence on artificial experience by manually reading infrared images of substation fault electrical equipment

propose to use image based on deep learning retrieval for power equipment fault diagnosis. By establishing the image of the typical faulty electrical equipment as a typical dataset

the image retrieval technology is used to retrieve the most similar N images from the typical dataset

and obtain the corresponding image fault resolution method according to the image name. Test verification is performed on the experimental dataset. At the same time

the performance reliability verification of the system is carried out on the CIFAR—10 and Oxford5 k datasets. Experiments show that it is feasible to apply image retrieval technology to image retrieval of faulty electrical equipments.