基于卷积神经网络的输电线路金具缺陷检测方法

Detection Method of Transmission Line Fittings Defects Based on Convolution Neural Network

针对架空输电线路长期处于恶劣、复杂的自然环境中,线路上的金具受气候、地形、外力作用等影响出现的不同类型缺陷,提出一种基于YOLO V3卷积神经网络的输电线路金具缺陷检测方法。通过YOLO V3卷积神经网络提取不同类型缺陷的特征,并对其进行适应性改进,识别与定位这些缺陷在输电线路上的位置,可提高检测的准确性和实时性,及时发现线路故障,确保输电线路安全稳定运行,提高输电线路巡检的效率和电网的智能化程度。

Aiming at the different types of defects of the fittings on the overhead transmission line,which are affected by the climate,terrain and external force,a method for detecting the defects of the fittings on the transmission line based on YOLO V3convolutional neural network is proposed.The YOLO V3 convolutional neural network is used to extract the characteristics of different types of defects and make adaptive improvement to them.Identifying and locating the positions of these defects on the transmission line can improve the accuracy and real-time of detection,timely detect line faults,ensure the safe and stable operation of the transmission line,improve the efficiency of transmission line inspection and the intelligence of the power grid.