基于faster RCNN与Mean-Shift的电缆附件缺陷红外图像自动诊断方法

Infrared thermography is an effective tool in timely detection of overheating defects in cable accessories. However, the traditional manual diagnosis method is time-consuming and laborious, and overplays experiences in engineering. Most of the existing studies reveal that the specific features are extracted as the input of the constructed neural network to realize the intelligent recognition of infrared images of electrical equipment. However, the selection of features depends on manual selection. Consequently, we proposed an autonomous recognition method based on the Faster RCNN network and the Mean-Shift algorithm. Firstly, the trained Faster RCNN network was applied to identify and locate the objects to be diagnosed. Then, the Mean-Shift algorithm was used for image segmentation to extract the overheating area. Finally, the temperature parameters were calculated, and the diagnosis results were obtained by comparing with the relevant condition assessment criteria. Case studies given in the paper show that the proposed method can be adopted to locate the cable ac-cessories and their overheating regions at different shooting angles and under various background conditions, and to realize the autonomous diagnosis. The research has a certain reference value for the defect diagnosis of cable accessories in actual engineering. © 2020, High Voltage Engineering Editorial Department of CEPRI. All right reserved.