Transformer Defects Detection Method Based on Visible and Infrared Fusion Images

摘要

Transformer is one of the most important power equipment in substation. Generally, workers in substation are supposed to inspect the transformer regularly and try to avoid abnormal operation of equipment, including the equipment state inspection and infrared detection. This contributes to the waste of manpower and the inspection task cannot be maintained frequently. This paper has proposed a transformer defects detection method based on visible and infrared fusion images. The method consists of the deep learning model which focuses on the fusion images. It is able to detect abnormal overheating and visible defects of the transformer automatically using the binocular camera in the substation without any manual intervention. The method has been operated in a 220kV substation in Hangzhou and the test results are shown in this paper.