华中科技大学学报（自然科学版）2025，Vol.53Issue(5)：24-30,7.DOI:10.13245/j.hust.250033
基于MSF和I-InceptionNet的变工况滚动轴承故障诊断
MSF and I-InceptionNet based fault diagnosis of rolling bearings with variable operating conditions
摘要
Abstract
A fault diagnosis method based on multisensor fusion(MSF)and improved InceptionNet(I-InceptionNet)was proposed to address the issue of low diagnostic accuracy and poor robustness in existing fault diagnosis methods under limited fault data and varying operating conditions.The method first resampled the acquired multi-source signals using a multi-phase anti-alias filter and converted them into red,green,blue(RGB)images as input to the model,preserving the multi-dimensional characteristics of the signals.Then,the attention feature fusion(AFF)technique was used to improve the connection layer of the InceptionNet network,fusing multi-sensor image features to enhance the overall classification performance.Finally,the fused images were classified for fault states.Experimental results show that the proposed method significantly outperforms single-source and other comparative methods in fault diagnosis under variable operating conditions.Particularly,in the cases of limited data,the diagnostic accuracy reaches 98.5%,demonstrating superior diagnostic precision and robustness.关键词
滚动轴承/故障诊断/InceptionNet网络/特征级融合/多传感器Key words
rolling bearings/fault diagnosis/InceptionNet networks/feature-level fusion/multi-sensor分类
计算机与自动化引用本文复制引用
王进花,曹文宝,周德义,曹洁..基于MSF和I-InceptionNet的变工况滚动轴承故障诊断[J].华中科技大学学报（自然科学版）,2025,53(5):24-30,7.基金项目
国家自然科学基金资助项目(62063020,61763028) （62063020,61763028）
甘肃省自然科学基金资助项目(20JR5RA463). （20JR5RA463）
