Insulator Fault Detection Method for Transmission Line Based on Deep Learning
In view of the problem that the insulator target of high voltage transmission line is susceptible to complex background and partial occlusion,a model based on improved YOLOv5 is developed.First,the detection box optimization algorithm of GA+K is used to improve the selected model to improve the recognition accuracy;then,the CBAM module is integrated in the YOLOv5 algorithm framework to improve the salience of the fault target area in the image;second,the Gaussian function is used to improve the non-maximum suppression method in YOLOv5 to improve the recognition accuracy of the occlusion target;finally,the UAV inspection image provided by a power grid company in Liaoning Province is used,and the proposed algorithm is compared with four classical target detection algorithms.The experimental results show that compared with the four comparison algorithms,this algorithm can guarantee high detection accuracy with good real-time performance,and the average detection accuracy can reach 95.1％.The detection time of each image is 0.04 s,which has both the accuracy and real-time performance of target detection.
万方数据
维普
