Fault Diagnosis Method of Power Equipment Based on Infrared Thermal Images

Nowadays, infrared imaging equipment, such as FLIR camera, has been widely used for fault diagnosis of power equipment. However, the low quality of infrared images and inconsistent temperature measurement lead to difficulties in identifying power equipment and diagnosing thermal defects. Therefore, this paper proposes a new fault diagnosis method of power equipment based on infrared thermal imaging. Firstly, the convolutional neural network YOLOv5 is introduced and improved to identify different types of power equipment; Then, the suspected heating area of the target equipment is obtained by morphological analysis, where the relative temperature information is calculated by the infrared imaging principle. Finally, the fault of the target equipment is diagnosed according to the heating fault temperature threshold of different power equipment. Experimental results demonstrate that the proposed method achieves an average accuracy of 95% for heat fault diagnosis on the constructed fault dataset of power equipment. Thus, it can be used to improve the efficiency and accuracy of fault diagnosis for power equipment, ensuring the safe and efficient operation of power grids.