基于改进YOLOv4的航空发动机叶片损伤检测  

Aeroengine Blade Damage Detection Based on Improved YOLOv4

作　　者：王倩岚 刘文波[1,2] 滕子煜 单永奇 WANG Qianlan;LIU Wenbo;TENG Ziyu;SHAN Yongqi(College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;Non-destructive Testing and Monitoring Technology for High-speed Transport Facilities Key Laboratory of Ministry of Industry and Information Technology,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China)

机构地区：[1]南京航空航天大学自动化学院,江苏南京211106 [2]南京航空航天大学高速载运设施无损检测监控技术工信部重点实验室,江苏南京211106

出　　处：《机械制造与自动化》2025年第2期223-226,共4页Machine Building & Automation

基　　金：国家自然科学基金项目(61871218);国家重点研发计划项目(2018YFB2003304);中央高校基本科研业务费项目(NJ2019007,NJ2020014)。

摘　　要：针对孔探设备检测航空发动机叶片损伤时会出现的漏检和人力物力耗费过大的问题,在YOLOv4网络的基础上提出一种基于扩张卷积和注意力机制的目标检测算法。使用CSPDarknet53作为特征提取网络;引入混合注意力机制并融合扩张卷积来增强网络的特征提取能力;采用Focal Loss函数优化原有的损失函数。实验结果表明:改进后算法网络的检测精度提高了5.71个百分点,更能满足发动机叶片损伤检测不漏检的需求。In order to solve the problems of missed detection and excessive cost of human and material resources when detecting aeroengine blade damage with hole detection equipment,a target detection algorithm based on dilated convolution and attention mechanism was proposed based on YOLOv4 network.CSPDarknet53 was used as the feature extraction network,and the Mixed attention mechanism and the fusion expansion convolution were introduced to enhance the feature extraction capability of the network.The Focal Loss function was applied to optimize the original loss function.The experimental results show that the detection accuracy of the improved algorithm network is increased by 5.71 per cent,better meeting the requirements of the engine blade damage detection without missed detection.

关 键 词：发动机 叶片损伤 目标检测 YOLOv4 注意力机制 扩张卷积 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]