Abstract:

In view of the fact that the traditional manual line inspection method is not suitable for short-range monitoring of insulator status of distribution lines, and the existing methods have the problems of low accuracy, a new method based on deep transfer learning for insulator status monitoring of distribution lines is proposed. Firstly, the intelligent distribution terminal collects the insulator images obtained by the camera on the distribution line, extracts the image features by oriented FAST and rotated BRIEF(ORB) algorithm, and adopts gray centroid method to ensure that the properties of the image feature points do not change after rotation. Then, according to the acquired image features, the depth learning and transfer learning algorithm are combined to train the image features and realize the insulator state classification. Finally, based on Matlab simulation platform, the proposed method and other combination methods are tested and analyzed in common scenes. Experimental results show that compared with other combination methods, the proposed method can accurately monitor insulator status in different environments, and the classification accuracy is higher.

Key words: Insulator state monitoring, ORB algorithm, feature analysis, deep transfer learning, state classification