基于改进 ZF 网络的航空发动机叶片损伤检测

中国民航大学学报 ›› 2021, Vol. 39 ›› Issue (3): 22-28.

基于改进 ZF 网络的航空发动机叶片损伤检测

邢志伟，李龙浦，侯翔开，施亚中，王浩

（1. 中国民航大学 a. 电子信息与自动化学院；b. 航空工程学院；c.工程技术训练中心，天津 300300； 2. 北京飞机维修工程有限公司西南航线中心，成都 610202）

收稿日期:2020-05-20 修回日期:2020-05-20 接受日期:2020-03-08 出版日期:2021-06-25 发布日期:2021-11-28
作者简介:邢志伟（1970—），男，辽宁新民人，教授，博士，研究方向为民航装备与系统、机场交通信息与控制.
基金资助:
国家自然科学基金项目（U1333121）

Damage detection of aero-engine blades based on improved ZF network

XING Zhiwei , LI Longpub, HOU Xiangkai , SHI Yazhong , WANG Hao

（1a. College of Electronic Information and Automation; b. College of Aeronautical Engineering; c. Engineering Technique Training Center, CAUC, Tianjin 300300, China; 2. Southwest Route Center, AMECO, Chengdu 610202, China）

Received:2020-05-20 Revised:2020-05-20 Accepted:2020-03-08 Online:2021-06-25 Published:2021-11-28

摘要/Abstract

摘要： 为实现航空发动机叶片损伤的准确检测，提出一种基于更快区域卷积神经网络（Faster R-CNN, Faster region-based convolutional neural network）模型的 ZF（Zeiler-Fergus）网络改进方法。首先，将航空发动机叶片损伤检测抽象为目标检测问题，并针对该问题利用 ZF 网络进行损伤位置及类型检测；其次，根据检测效果对原始 ZF 网络进行增加卷积层操作，并调整卷积层卷积核尺寸及池化层步长等超参数，将增加的卷积层纳入共享卷积层以提取更加细致的损伤特征；最后，将已有数据集分为 3 组子数据集，通过实验分别对比每组子数据集改进前后的 ZF 网络性能。仿真结果表明：改进 ZF 网络较原始 ZF 网络平均精度均值最高可提高 5.5%，证实了改进 ZF 网络能更加精准地预测发动机叶片损伤。

关键词: 发动机叶片, Faster R-CNN, ZF 网络, 损伤检测

Abstract: In order to achieve the accurate detection of aero-engine blade damages, a Zeiler-Fergus(ZF) network improvement method based on the faster region-based convolutional neural network(Faster R-CNN) model is proposed. Firstly, the aero-engine blades damage detection is abstracted as the object detection problem, and aiming at the problem, a ZF network is used to detect the damage location and type. Then, some convolutional layers are added into the original ZF network according to the detection effect. Hyper -parameters such as kernel size in the convolutional layers and the stride size in the pooling layers are adjusted, and the added convolutional layers are included in the shared convolutional layers to extract more detailed damage features. Finally, the existing dataset is divided into three sub-datasets, on which the network performance before and after improvedment is experimentally compared. Results show that the mean average precision of the improved ZF network is up to 5.5% compared with that of the original network, which proves that the improved ZF network can predict engine blades damage more accurately

Key words: engine blades, Faster R-CNN, Zeiler-Fergus(ZF) network, damage detection

中图分类号:

邢志伟 , 李龙浦 , 侯翔开 , 施亚中 , 王浩 . 基于改进 ZF 网络的航空发动机叶片损伤检测 [J]. 中国民航大学学报, 2021, 39(3): 22-28.

XING Zhiwei , LI Longpu , HOU Xiangkai , SHI Yazhong , WANG Hao. Damage detection of aero-engine blades based on improved ZF network [J]. Journal of Civil Aviation University of China, 2021, 39(3): 22-28.

参考文献 14

[1]	DYRWAL A, MEO M, CIAMPA F. Nonlinear air-coupled thermosonicsfor fatigue micro-damage detection and localisation[J]. NDT & E International, 2018, 97: 59-67.
[2]	TIWARI K A, RAI譒UTIS R. Refinement of defect detection in the contact and non-contact ultrasonic non-destructive testing of wind turbineblade using guided waves[J]. Procedia Structural Integrity, 2018, 13:1566-1570.
[3]	于霞, 张卫民, 邱忠超, 等. 飞机发动机叶片缺陷的差激励涡流传感器检测[J]. 北京航空航天大学学报, 2015, 41(9): 1582-1588.
[4]	陈棣湘, 潘孟春, 田武刚, 等. 基于柔性电磁传感器的发动机叶片微缺陷检测[J]. 中国测试, 2018, 44(1): 65-68.
[5]	李雅峰, 徐玉秀. 基于模态应变能变化率的大型风力发电机叶片损伤识别与定位[J]. 太阳能学报, 2015, 36(9): 2251-2256.
[6]	丁鹏, 李长有, 马齐爽, 等. 基于小波的航空发动机叶片孔探损伤检测[J]. 北京航空航天大学学报, 2006, 36(12): 1435-1438.
[7]	陈广, 杨震. 基于卷积特征的光纤缺陷检测方法[J]. 中国惯性技术学报, 2019, 27(1): 95-100.
[8]	LI D S, CONG A R, GUO S. Sewer damage detection from imbalancedCCTV inspection data using deep convolutional neural networks withhierarchical classification[J]. Automation in Construction, 2019, 101:199-208.
[9]	CHEN J W, LIU Z G, WANG H R, et al. Automatic defect detection offasteners on the catenary support device using deep convolutional neural network[J]. IEEE Transactions on Instrumentation and Measurement,2018, 67(2): 257-269.
[10]	REDDY A, INDRAGANDHI V, RAVI L, et al. Detection of cracks anddamage in wind turbine blades using artificial intelligence-based imageanalytics[J]. Measurement, 2019, 147: 1-12.
[11]	DENHOF D, STAAR B, LUTJEN M, et al. Automatic optical surface inspection of wind turbine rotor blades using convolutional neural networks[J]. Procedia CIRP, 2019, 81: 1166-1170.
[12]	张维亮, 李楠, 李昂, 等. 基于概率神经网络的航空发动机叶片损伤图像分割[J]. 沈阳航空航天大学学报, 2013, 30(2): 22-26.
[13]	张维亮. 航空发动机叶片损伤图像快速识别技术研究[D]. 沈阳: 沈阳航空航天大学, 2014.
[14]	REN S Q, HE K, GIRSHICK R, et al. Faster R-CNN: towards real-timeobject detection with region proposal networks[J]. IEEE Transactionson Pattern Analysis and Machine Intelligence, 2017, 39(6): 91-99.