摘要
提出在小样本情况下建立设备状态特征库的方法 ,以解决机械故障诊断中无法采集到大量样本来建立状态特征库的难题。讨论如何根据各种指标在变工况下的稳定程度来选择状态特征的问题 ,提出用指标的稳定行为量来衡量小样本指标的稳定程度。利用火车车轮滚动轴承 3种状态下振动加速度信号的功率谱 ,依据指标的稳定程度 ,确定其特征指标 ,并应用所提出的方法对确定的指标进行统计模拟 ,构建了对应的状态特征库 ,在实际中实现了良好的状态辨识。实践证明 。
In machinery fault diagnosis, knowledge base design should be based on massive data collection, which cannot be implemented easily in practice. Therefore, the bootstrap method is introduced to establish the knowledge base using small sampling. To evaluate the stability of selected state features, a measure steady behavior of the feature is presented. According to their steady behavior, appropriated features corresponding to their states are optimized and selected. Based on these extracted state features, bootstrap method is used to design their state knowledge base. As an example, the vibration signals of a type of rolling bearing are processed and feature-extracted. Practically, it is verified that the method is simple, feasible and effective.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2002年第10期829-832,共4页
China Mechanical Engineering
关键词
故障诊断
特征提取
状态特征库
统计模拟
fault diagnosis feature extraction knowledge-base design bootstrap
