摘要
针对滚动轴承传统上侧重某一时间点的故障类型诊断问题,提出一种注重评估全寿命周期中性能退化趋势的指标提取方法。利用EEMD分解初始信号得到诸多IMF分量,采用相关系数准则提取相关程度高的IMF分量作为有效信息进行重构,实现信号的降噪处理。对降噪信号分别建立时域、频域和时频域共37个性能退化指标,剔除敏感度太弱和信息过于嘈杂等不适合表征滚动轴承退化趋势指标后,利用PCA法进行加权融合剩余的多域性能退化指标,最终得到能全面表征性能退化趋势指标。对Cincinnati大学采集的全寿命周期振动信号进行实验分析,结果表明,基于EEMD和PCA加权融合多域的性能退化指标比单一指标的纯净度更高,敏感度由单一指标的600点提升到490点,为滚动轴承剩余寿命预测奠定了基础。
Traditionally,fault diagnosis of rolling bearings focuses on the type of fault at sometime. This paper proposes an indicator extraction method of performance degradation trends in the full life-cycle assessment to achieve predictive maintenance. EEMD is used to decompose the initial signal to obtain a lot of the IMF components,the high degree of correlation of the IMF components is reconstructed to achieve noise reduction which is extracted as the effective information by correlation coefficient criteria. 37 performance degradation indicators are established in time domain,frequency domain and time-frequency domain,respectively. After excluding the indicators which have too weak sensitivity,too noisy signal or are not suitable for characterizing rolling bearing performance degradation trends,and so on,the comprehensive characterization indicators are obtained through the weighted fusion remaining multi-domain indicators by the use of PCA method. The full life-cycle vibration signals which are collected from Cincinnati University are analyzed. It shows that the purity of the weighted fusion multi-domain performance degradation indicators based on the EEMD and PCA is higher than a single indicator. The sensitivity can be increased from about 600 to 490 points,which laids a good foundation for the residual prediction of the life of rolling bearings.
出处
《江南大学学报（自然科学版）》
CAS
2015年第5期572-579,共8页
Joural of Jiangnan University (Natural Science Edition)
基金
福建省自然科学基金项目(2015J01643)
福建省中青年教师教育科研项目(JA14332)
福建省高校杰出青年科研人才培育计划项目(闽教科[2015]54号)
宁德师范学院"服务宁德区域经济和产业发展"专项课题项目(2013F25
2013F26)
关键词
滚动轴承
集成经验模态分解
性能退化指标
主成分分析
rolling bearings
EEMD
performance degradation indicator
principal component analysis
