基于改进迁移学习的行星齿轮箱故障自动诊断
Automatic Fault Diagnosis of Planetary Gearbox Based on Improved Transfer Learning
欧振杰 1成兴 1覃仕明1
作者信息
- 1. 广西壮族自治区特种设备检验研究院,广西南宁 530200
- 折叠
摘要
传统机器学习故障诊断方法依赖专业经验选取统计特征,导致诊断结果误差较大.为此,提出了基于改进迁移学习的行星齿轮箱故障自动诊断.采集行星齿轮箱的振动信号,对振动信号进行去噪处理,利用深度学习改进迁移学习构建故障诊断模型,对采集信号进行分类识别,实现了行星齿轮箱故障自动诊断.结果表明,该设计方法下不同类型的行星齿轮箱故障诊断精度为96.09％,证实了该方法的性能良好.
Abstract
The traditional machine learning fault diagnosis method relies on professional experience to select statistical features,the diagnosis results are greatly affected by human factors.Therefore,the automatic fault diagnosis of planetary gear boxes based on improved transfer learning is proposed.The vibration signal of the planetary gearbox is collected,the vibration signal is denoised,and the deep learning is improved by transfer learning to build the fault diagnosis model,classify and identify the collected signal,and realize the automatic fault diagnosis of the planetary gearbox.The results show that the fault diagnosis accuracy of different types of planetary gearbox is 96.09％,which confirms the good performance of this method.
关键词
改进迁移学习/行星齿轮箱/故障诊断/自动诊断方法Key words
improved transfer learning/planetary gearbox/fault diagnosis/automatic diagnosis method引用本文复制引用
出版年
2024
NETL
NSTL
万方数据