Infrared Temperature Detection Method for Substation Equipment Based on Improved CenterNet
Infrared detection can detect the abnormal temperature of power equipment in substations,and reduce the probability of safety accidents.Therefore,an object detection algorithm model of CenterNet_PRO based on improved CenterNet is proposed.This algorithm adopts ShuffleNet V1/V2 as a backbone network,and introduces a feature pyramid network(FPN)to extract multi-scale features.In order to overcome the difficulties of target detections at different scales,the algorithm increases the rotation angle regres-sion branches,predicts the rotation angle of the target,and optimizes with the improved IoU Loss,further improves the model detec-tion speed and accuracy.The threshold segmentation method is adopted to extract the surface temperature of power equipment and an-alyze and calculate the surface temperature,which designs and defines the temperature defect judgment specification and temperature warning threshold of power equipment,and judges the related defects of power equipment according to the specification.The experi-mental results show that the average accuracy of the improved CenterNet model reaches up to 90％,compared with the traditional CenterNet model,the average accuracy improved by 1.3％,which can meet high requirements for infrared detection of power equip-ment in actual substation scenarios.
centerNetshuffleNetpower equipmentinfrared image temperature defect detectionmulti-scale feature extraction
NETL
NSTL
万方数据
