Design And Implementation Of Transmission Line Foreign Object Detection System Based On The Improved YOLOv4

The demand for power resources is gradually increased with the rapid development of our country’s economy.Meanwhile,the transmission line has longer distance and regions when it carries the power of electricity,which makes daily inspections more difficult.During the operation of the transmission line.Some foreign objects on the transmission line would cause a short circuit of the transmission line or even a fire,which will seriously affect the power transmission.Traditional manual inspection is not only time-consuming and laborious,but also unable to conduct timely inspection.In order to ensure the safe and stable operation of power facilities.Intelligent inspection of transmission lines is particularly important.For the characteristics of common foreign objects in transmission lines.The YOLOv4 algorithm is improved to a transmission line foreign objects detection algorithm.Based on this algorithm,a transmission line foreign objects detection system is designed and implemented using Py Qt.The main work of this thesis is as follows:(1)A foreign objects detection algorithm based on improved YOLOv4 is proposed.First of all,in view of the situation that the color and texture of foreign objects in the transmission line are similar to the background information.The SPP(Spatial Pyramid Pooling)module of YOLOv4 is improved.So that it can retain more object information in the case of background interference and reduces object information loss caused by maximum pooling.Secondly,the original BCE(Binary Cross Entropy)loss function is optimized and improved to obtain the GBCE(Gradient-boosting Binary Cross Entropy)loss function,which improves the model’s resolving ability of colors and textures,and reduces the fallout ratio caused by similar colors and textures.The experimental results showed that the improved YOLOv4 transmission line foreign objects detection algorithm has a better comprehensive performance.(2)A foreign objects detection system for transmission lines is designed and implemented.The main functional modules of the system can be divided into three categories:staff information management,foreign objects type detection,and detection data analysis.The staff information management realizes the confidential operation of the working data by the staff.The foreign objects type detection can detect the information of common foreign objects in the transmission line.Then save it in the database.Analysis of the detection data can display the detection results of foreign objects in the transmission line according to the needs of the staff.After testing,the performance of the system meets the design requirements and realizes the foreign object detection function of the transmission line.