摘要
提出一种利用盲源分离技术对齿轮箱混合故障进行诊断的方法。该方法以最小互信息量为准则,采用自然梯度的自适应算法求解统计独立源信号的估计值,并根据分离信号的频谱成功地提取了混合故障的特征信息,有效地诊断出齿轮箱所处的故障状态。
A method based on blind source separation was developed for mixing faults diagnosis of gearbox. By using adaptive algorithm of natural gradient based on minimum mutual information, the estimate of statistics independent sources was figured out. According to the separated signal＇s FFT, the fault information was picked up successfully and the fault states were diagnosed effectively.
出处
《机械设计与制造》
北大核心
2006年第6期87-89,共3页
Machinery Design & Manufacture
关键词
齿轮箱
故障诊断
盲源分离
最小互信息
自然梯度
Gearbox
Fault diagnosis
Blind source separation
Minimum mutual information
Natural gradient
