摘要
针对航空发动机故障诊断过程缺乏大量实际故障数据的问题,提出了一种基于支持向量机和小波包相结合的滚动轴承的早期故障诊断方法。该方法利用有限的故障样本,以结构风险最小原理为基础,建立滚动轴承早期故障特征与其运行状态之间的对应函数关系,即故障分类器,并以该函数的输出判定轴承的早期故障类型。实验结果表明,小波包分析能够有效的提取滚动轴承中微弱的早期故障特征,支持向量机可以对这些早期故障特征进行准确识别。
Aim at the problem of lacking fault data on aero-engine fault diagnosis, a rolling bearing initial fault diagnosis method based on support vector machine and wavelet packet is proposed. Through a finite learning samples, the function relationship between the initial fault character and it＇s running state is established based on Structural Risk Minimization, which is so-called faults classifier. By the faults classifier＇s output, the type of rolling beating initial fault is determined. The experiment shows that wavelet packet analysis can extract the initial fault character effectively and the support vector machine method can identify those initial faults of rolling bearing correctly.
出处
《科学技术与工程》
2009年第13期3884-3888,共5页
Science Technology and Engineering
基金
国家自然科学基金(50675178
60472116)资助
关键词
早期故障诊断
支持向量机
滚动轴承
小波包
initial fault diagnosis
support vector machine （SVM）
rolling bearing
wavelet packet
