摘要
针对设备的视情维修,提出一种将小波包奇异谱熵和支持向量数据描述(SVDD)相结合的滚动轴承性能退化评估方法。先提取轴承全寿命周期内振动信号的小波包奇异谱熵作为轴承状态的特征矢量,然后以轴承正常状态下的特征矢量训练SVDD,得到正常状态下的基准超球体,再计算轴承全寿命周期内的特征矢量与基准超球体之间的相对距离,作为性能退化过程的定量评估指标,并对失效阈值和早期故障阈值进行设定。结果表明,与基于小波包和SVDD的性能退化评估方法相比,该方法的早期故障检测能力更强,对轴承性能退化各个阶段的描述更加准确。最后,利用基于EMD的Hilbert包络解调方法对评估结果的正确性进行了验证。
Aiming at the condition-based maintenance of equipments, a novel assessment method of rolling bearing performance degradation combining wavelet packet singular spectral entropy (WPSSE) and support vector data description (SVDD) was proposed. Firstly, WPSSEs were extracted from bearing full-life-cycle vibration signals as feature vectors to describe a bearing running state. Secondly, SVDD was trained using the feature vectors under normal condition to get the fiducial hypersphere of normal state. Then, the relative distance between full-life-cycle feature vectors of bearing and fiducial hypersphere was calculated as a quantitative index of performance degradation, and the failure threshold and incipient fault threshold were set. Analytical results of experimental data indicated that compared with the degradation evaluation method based on wavelet packet and SVDD, the proposed method had stronger ability for incipient fault detection, and it could describe the stages of bearing performance degradation more accurately. Finally, Hilbert envelope demodulation method based on empirical mode demodulation was used to validate the reliability of evaluation result.
出处
《机械科学与技术》
CSCD
北大核心
2016年第12期1882-1887,共6页
Mechanical Science and Technology for Aerospace Engineering
基金
国家自然科学基金项目(51205130)
江西省科协重点活动项目(赣科协字[2014]88号)资助
关键词
滚动轴承
小波包奇异谱熵
支持向量数据描述
性能退化评估
包络解调
rolling bearing
feature extraction
fault detection
wavelet packet singular spectral entropy
SVDD
performance degradation assessment
envelope demodulation
reliability