Design And Implementation Of Target Defect Recognition Algorithm For High Voltage Transmission Line

With the improvement of the national economic development level and people’s living standards,the scale of the power grid is also expanding,and how to ensure the safe and reliable operation of the power grid has begun to be widely concerned.The transmission line is one of the important links of the power system.Due to long-term exposure to the field during operation,the wires,fittings,insulators,and other components of the line are prone to defects such as rust,damage,and broken strands.At the same time,the non-standard installation of components also brings hidden dangers to the safe operation of transmission lines.Using UAV inspection data to efficiently and accurately obtain transmission line defect information through deep learning technology has important practical significance for line defect inspection.However,the environment of transmission lines is complex and changeable,which brings huge challenges to target defect identification.Based on the UAV line inspection,this paper uses the inspection pictures obtained by the UAV to study the target defect recognition of the transmission line from two aspects:the defect detection of spacer rods and the detection of multiple types of targets.Among them,the multi-objective defect identification methods for five types of samples,spacer,nest,ground wire,shock-proof hammer,and ground wire clamp,are mainly elaborated.The specific research contents are as follows:This paper first conducts a thorough investigation on the detection algorithm of the transmission line target defect,introduces the algorithm used in the power target defect-recognition in recent years and its shortcomings,proposes the use of drones to patrol and take pictures,and the use of deep learning algorithms to achieve the realization of the transmission line Artificial intelligence recognition of ontological defectsIn view of the poor recognition effect of the existing target detection algorithm on the spacer bar,this article improves on the SSD algorithm and uses the fully convolutional neural network DenseNet-32 to replace the original VGG-16 network to avoid pooling The key position information is discarded,and the attention mechanism is introduced to enhance the semantic information of the feature map to improve the network feature extraction ability and algorithm detection accuracy.The experimental results show that the algorithm is better than other algorithms for identifying single-type targets on transmission lines,specifically the gap bar tilt defect.Due to the addition of the attention mechanism network,the recognition rate is slightly lower than the SSD algorithm 7FPS,reaching 91.2%Average accuracy.In the detection of multiple types of targets,first of all,aiming at the problem of insufficient data sets,the image data augmentation method is used to expand the samples.To make the image clearer and features more obvious,piecewise linear transformation,histogram equalization,and smoothing and denoising image preprocessing methods are used to highlight the target or gray-scale interval of the region of interest.Furthermore,in terms of target detection and defect-recognition,a transmission line multi-target detection technology based on attention mechanism fusion is proposed,and the attention fusion method is used to fuse features of different scales to improve the segmentation performance.Experiments show that the improved algorithm proposed in this paper is superior to other classic target detection algorithms in terms of recognition accuracy.In order to further improve the transmission line UAV inspection combined with the AI intelligent defect recognition operation system process,the deployment and application of the improved YOLOv3 algorithm has been realized.In practical applications,the compressed algorithm is moved to the Jetson AGX Xavier platform,and the post-deployment algorithm is used to identify defects in the pictures of the transmission line components taken by the drone.And on the designed transmission line defect recognition platform,the recognition result display and data analysis application verify that the multi-objective defect-recognition algorithm in this paper meets the application requirements of actual scenarios.