Research on Construction of Infrared Image Classification Model of Substation Equipment Based on CNN

Infrared Intelligent detection is the key research direction of infrared fault detection of power equipment. However, different types of equipment and different processing purposes lead to a variety of fault detection methods. If the manual method is used to classify the equipment, the efficiency of fault detection will be greatly reduced. In this paper, based on convolutional neural networks, based on RGB and HSV color space conversion, a classification model suitable for infrared images of power equipment is constructed. Firstly, the structure characteristics and training process of CNN are introduced. After that, based on RGB and HSV color space conversion, the infrared image of substation equipment is processed, and the target area of suitable size is extracted to establish network training and test set. Finally, a CNN-based infrared image classification model is established, and its good applicability is verified by case analysis.