基于改进YOLOv5s的架空输电线路鸟类入侵检测方法  

Bird Invasion Detection Method for Overhead Transmission Lines Based on Improved YOLOv5s

作　　者：裴少通[1] 张善驰 PEI Shaotong;ZHANG Shanchi(School of Electrical and Electronic Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《智慧电力》2023年第6期100-105,共6页Smart Power

基　　金：国家自然科学基金资助项目(61871182);中央高校基本科研项目(2020MS093)。

摘　　要：鸟类在输电线路周围活动愈加频繁,导致输电线路跳闸与停电等事故。为快速准确地检测出入侵输电线路区域中的鸟类目标,提出了一种基于改进YOLOv5s的架空输电线路鸟类入侵检测方法。首先,将特征提取网络中的普通卷积模块替换为嵌入CBAM注意力机制的卷积模块;其次,利用SPPF模块替代原有SPP模块,加强特征提取效果;最后,将Mish激活函数嵌入三次卷积(Conv×3,C3)模块中,提高模型的非线性学习能力。结合模型训练与测试结果验证了所提方法的有效性。Increasingly frequent bird activities around transmission lines cause the line tripping and blackouts.In order to detect the bird targets quickly and accurately in the invasion area of the transmission lines,this paper proposeds a bird invasion detection method for the overhead transmission lines based on improved YOLOv5s.Firstly,a common convolutional module in the feature extraction network is replaced by a convolutional module embedded into the convolutional block attention module(CBAM)attention mechanism.Secondly,a spatial pyramid pooling-fast(SPPF)module instead of a original spatial pyramid pooling(SPP)module is used to enhance the feature extraction effect.Finally,the Mish activation function is embedded into the cubic convolution(Conv×3,C3)module to improve the nonlinear learning ability of the model.The model training and test results show the effectiveness of the proposed method.

关 键 词：输电线路 鸟类入侵 改进YOLOv5s CBAM SPPF 

分 类 号：TM755[电气工程—电力系统及自动化]