摘要
研究了设备噪声监测中主分量特征表示的提取及应用。在设备噪声时域和频域统计模式特征基础上,通过主分量分析探讨了设备噪声模式的主分量特征表示方法,引入相关度的概念分析了主分量特征表示对设备状态的表征能力,提出了选择有效维数的主分量特征表示进行设备噪声监测的方案。通过在EQ6100型发动机上预先模拟四种连杆轴承磨损故障,测取噪声信号,实例分析显示了低维主分量特征表示可以有效表征设备状态。实验最后对测试集样本进行状态监测得到了100%的准确率,表明了主分量特征表示用于设备噪声监测的有效性。
The extraction and application of principle component representations were studied for machine noise monitoring. On the base of the time- and frequency domain statistical features extracted from the machine noise signal, the principle component representations of machine noise pattern were explored, the ability of principle component representations to represent the characteristic of machine condition by introducing the idea of correlation was analyzed, and the machine noise monitoring scheme using the effective low-dimensional principle component representations was proposed. Through previously designing four wearing faults on the connecting rod bearing of EQ6100 model gasoline engine, experimental analysis shows that the low-dimensional principle component representations of engine noise can be conveniently and effectively used for representing the machine conditions. Finally, the excellent result with the accuracy of 100% was obtained by monitoring the conditions of the testing samples, which demonstrates the availability of principle component representations for machine noise monitoring.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2006年第6期1093-1099,共7页
Optics and Precision Engineering
关键词
设备噪声
状态监测
主分量分析
连杆轴承
磨损
machine noise
condition monitoring
principle component analysis
connecting rod bearing
wearing
作者简介
何清波（1980-），男，河南台前人，中国科学技术大学精密机械与精密仪器系博士研究生，研究方向为盲信号处理、统计信号处理、设备状态监测；E-mail：heqb@ustc．edu．cn;
孔凡让（1951-），男，安徽人，教授，博士生导师，中国科学技术大学工程科学学院副院长，研究方向为智能信息处理、振动分析、测控技术、设备状态监测与故障诊断等。
