Research of Insulator Fault Identification Method Based on Atlas Intelligent Computing Platform

Transmission line monitoring plays a crucial role in the safe operation of the power grid. The application of image recognition technology based on artificial intelligence algorithms can improve the efficiency of fault identification and reduce the labor cost. Object detection is to detect the object of interest in the given picture and is an effective method to recognize the fault location. This paper proposed an insulator fault identification method based on the Atlas intelligent computing platform aiming at the characteristics of small insulator fault data set. This method adopts the SSD300 model for training and inference and analysis is carried out based on the Atlas intelligent computing platform. The experiment results show that the SSD300 model can be ported well to the Atlas intelligent computing platform without reducing the recognition accuracy. At the same time, the model size is decreased.