基于Inception-CBAM模型的发动机气路部件故障诊断方法
Fault diagnosis method for engine gas path components based on Inception-CBAM model
詹美佳 1徐建国 1张贇 1孟之一1
作者信息
- 1. 南京航空航天大学能源与动力学院,南京 210016
- 折叠
摘要
针对航空发动机气路部件故障,提出一种基于Inception-CBAM模型的航空发动机故障诊断方法.该模型具有包含3个并行分支的Inception模块,以不同大小感受野从传感器数据提取发动机状态特征;每个分支包含1个CBAM模块,自适应调整该分支特征权重;后续展平层进一步将3个分支予以拼接;最后,通过2个全连接层和1个Softmax层输出故障诊断结果.与CNN模型、Inception模型、Inception-SE模型开展对比仿真实验,实验结果表明,Inception-CBAM模型具有优秀的故障诊断性能.
Abstract
A fault diagnosis method based on Inception-CBAM model is proposed to diagnose the faults in gas path components of aero-engine.The diagnosis model incorporates an three-branch Inception module,which extracts engine features from sensor data with receptive field of different sizes.Each branch contains a CBAM module that adaptively adjusts feature weights.And the subsequent flattening layer furtherly concatenate the three parallel branches.Finally,the fault diagnosis results are output through two fully connected layers and a Softmax layer.The Inception-CBAM model was compared with CNN model,Inception model and Inception-SE model in a simulation experiment.The experimental results show that the proposed model has better fault diagnosis performance.
关键词
航空发动机/气路部件/故障诊断/Inception/CBAMKey words
Aero-engine/Gas path component/Fault diagnosis/Inception/CBAM引用本文复制引用
出版年
2024
NETL
NSTL
万方数据