摘要
为解决机械智能监测与诊断中缺少故障样本的问题,提出一种机械故障单值分类的新方法——支持向量数据描述法。该方法只需要一类目标样本作为学习样本,而不需要除学习样本以外的其他非目标样本,即可以建立单值分类器,从而将非目标样本从目标样本中区分开。提取机械设备正常运行时振动信号的特征值组成学习样本集,建立单分类模型,应用该模型可以对未来的设备运行状态和故障进行识别诊断。该方法应用于某水泥厂煤灰鼓风机故障诊断的工程实践中,取得满意的结果。
In order to solve the problem of insufficient fault samples in intelligent monitoring and diagnosis for machinery, a new method of one-class classification of mechanical fault-support vector data description was proposed in which the one-class class/tier can be set up by using only one kind of target sample without knowing other outlier samples. In this way, the outlier samples can be distinguished from the target samples. The learning sample set can be bulh up by extracting the vibration signal features of the machinery with the nounal state so that the classifier model can also be established. This model can be used to identify the running states of the equipment and diagnose the faults of it in the future. The method has been applied to the fault diagnosis of the fan in the cement plant, and the satisfactory results are obtained.
出处
《机械强度》
EI
CAS
CSCD
北大核心
2008年第5期697-701,共5页
Journal of Mechanical Strength
基金
国家自然科学基金项目(50675209)
河南省自然科学基金项目(0611022400)
河南省杰出人才创新基金项目(0621000500)~~
关键词
支持向量数据描述
单值分类
故障诊断
Support vector data description
One-class classification
Fault diagnosis
