基于改进的SqueezeNet直升机滚动轴承故障诊断

余志锋; 熊邦书; 李新民; 欧巧凤

doi:10.13224/j.cnki.jasp.20210317

基于改进的SqueezeNet直升机滚动轴承故障诊断

doi: 10.13224/j.cnki.jasp.20210317

1.
南昌航空大学图像处理与模式识别省重点实验室，南昌 330063
2.
中国航空工业集团有限公司中国直升机设计研究所直升机旋翼动力学重点实验室，江西景德镇 333001

基金项目: 国家自然科学基金（61866027）；航空科学基金（2016ZD56008；20185756006）

详细信息

作者简介:
余志锋（1995-），男，硕士生，主要从事直升机自动倾斜器滚动轴承故障研究。

通讯作者:
熊邦书（1968-），男，教授，博士，主要从事计算机视觉、直升机故障诊断研究。E-mail：xiongbs@126.com

中图分类号: V233+5;TH133.33
计量
- 文章访问数: 104
- HTML浏览量: 6
- PDF量: 48
- 被引次数: 0
出版历程
- 收稿日期: 2021-06-22
- 刊出日期: 2022-06-28

Fault diagnosis of helicopter rolling bearing based on improved SqueezeNet

1.
Provincial Key Laboratory of Image Processing and Pattern Recognition，Nanchang Hangkong University，Nanchang 330063，China
2.
Science and Technology on Rotorcraft Aeromechanics Laboratory，China Helicopter Research and Development Institute，Aviation Industry Corporation of China，Limited，Jingdezhen Jiangxi 333001，China

摘要

摘要: 针对现有基于卷积神经网络的故障诊断方法存在模型参数量和计算量大的问题，提出一种改进的SqueezeNet模型应用于直升机滚动轴承故障诊断。该模型借鉴VGG16模型的思想，在经典的SqueezeNet基础上，采用3个 $3 \times 3$ 卷积核代替1个 $7 \times 7$ 卷积核，实现了在相同感知野条件下增加网络容量、增强非线性、减少网络参数量，采用卷积层、池化层和Fire模块、池化层两大结构交替的方式组成模型特征提取层，在保障特征提取能力的情况下，进一步减少了网络参数量。通过轴承数据开展模型故障诊断实验，结果表明该模型诊断精度达到99.65%，与传统卷积神经网络及经典的SqueezeNet模型相比诊断精度相当，而计算量与参数量最大缩减约6倍和36倍。
- 直升机滚动轴承 /
- 卷积神经网络 /
- VGG16模型 /
- 轻量化 /
- SqueezeNet模型
Abstract: In order to solve the problem of large amount of model parameters and calculations in the existing fault diagnosis methods based on convolutional neural networks，an improved SqueezeNet model was proposed to be applied to the fault diagnosis of helicopter rolling bearings.By drawing on the idea of the VGG16 model based on the classic SqueezeNet，the model used three sizes of $3 \times 3$ convolution kernels instead of one size of $7 \times 7$ convolution kernel，and realizes the increase of network capacity，enhancement of nonlinearity，reduction of network parameters amount under the same perceptual field conditions.To further reduce the amount of network parameters，the convolutional layer，pooling layer，Fire module and pooling layer were alternated to form the model feature extraction layer.While ensuring the feature extraction capability，the amount of network parameters was further reduced.The model faults diagnosis experiment was carried out through the bearing data onto the research group.The results showed that the diagnosis accuracy of the model reached 99.65%，which was comparable to the traditional convolutional neural network and the classic SqueezeNet model.The calculation amount and the parameter amount were reduced by about 6 times and 36 times.
- helicopter rolling bearings /
- convolutional neural network /
- VGG16 model /
- lightweight /
- SqueezeNet model

HTML

参考文献(19)

[1]	万齐杨,熊邦书,李新民,等.基于DCAE-CNN的自动倾斜器滚动轴承故障诊断[J].振动与冲击,2020,39（11）:273-279.
[2]	HINTON G E,SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313（5786）:504-507.
[3]	WANG Huaqing,LI Shi,SONG Liuyang,et al.A novel convolutional neural network based fault recognition method via image fusion of multi-vibration-signals[J].Computers in Industry,2019,105（1）:182-190.
[4]	DENG Zhipeng,SUN Hao,ZHOU Shilin,et al.Multi-scale object detection in remote sensing imagery with convolutional neural networks[J].ISPRS Journal of Photogrammetry and Remote Sensing,2018,145（4）:3-22.
[5]	PANG Shuchao,LUACES O,YU Zhezhou,et al.Deep learning to frame objects for visual target tracking[J].Engineering Applications of Artificial Intelligence,2017,65（1）:406-420.
[6]	ZHAO Rui,YAN Ruqiang,CHEN Zhenghua,et al.Deep learning and its applications to machine health monitoring[J].Mechanical Systems and Signal Processing,2019,115:213-237.
[7]	YANG Bin,LEI Yaguo,JIA Feng,et al.An intelligent fault diagnosis approach based on transfer learning from laboratory bearings to locomotive bearings[J].Mechanical Systems and Signal Processing,2019,122:692-706.
[8]	ZHU Jun,CHEN Nan,PENG Weiwen.Estimation of bearing remaining useful life based on multiscale convolutional neural network[J].IEEE Transactions on Industrial Electronics,2019,66（4）:3208-3216.
[9]	杨平,苏燕辰.基于卷积门控循环网络的滚动轴承故障诊断[J].航空动力学报,2019,34（11）:2432-2439.
[10]	陈果,杨默晗,于平超.基于深度学习的航空发动机不平衡故障部位识别[J].航空动力学报,2020,35（12）:2602-2615.
[11]	钟诗胜,李旭,张永健.基于DBN的不均衡样本驱动民航发动机故障诊断[J].航空动力学报,2019,34（3）:708-716.
[12]	李坤伦,魏泽发,宋焕生.基于SqueezeNet卷积神经网络的车辆颜色识别[J].长安大学学报（自然科学版）,2020,40（4）:109-116.
[13]	BOUVRIE J.Notes on convolutional neural networks[J].Neural Nets,2006,10（6）:43-51.
[14]	余志锋,熊邦书,熊天旸,等.基于VMD-CWT和改进CNN的直升机轴承故障诊断[J].航空动力学报,2021,36（5）:948-958.
[15]	SUN Wei,ZHANG Zhenhao,HUANG Jie.RobNet:real-time road-object 3D point cloud segmentation based on SqueezeNet and cyclic CRF[J].Soft Computing,2020,24（8）:5805-5818.
[16]	ZHOU Guoxu,ZHAO Qibin,ZHANG Yu,et al.Linked component analysis from matrices to high-order tensors:applications to biomedical data[J].Proceedings of the IEEE,2016,104（2）:310-331.
[17]	ALHICHRI H,BAZI Y,ALAJLAN N,et al.Helping the visually impaired see via image multi-labeling based on squeezenet CNN[J].Applied Sciences-Basel,2019,9（21）:1-20.
[18]	霍爱清,张文乐,李浩平.基于深度残差网络和GRU的SqueezeNet模型的交通路标识别[J].计算机工程与科学,2020,42（11）:2030-2036.
[19]	SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[EB/OL].[2021-06-01].http:∥citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.740.6937&rep=rep1&type=pdf.