摘要
提出了一种基于J散度的谱系聚类算法,对不同状态下信号的AR模型进行了分类,以分类结果建立正常信号的标准样本,比较待检样本与标准样本之间J散度和设定阈值的大小,实现对待检样本的分类。将该方法应用于往复式压缩机气阀的故障诊断中,比较J散度与欧氏距离和相关系数在分类中的效果,证实了基于J散度的模式分类方法的分类结果更加准确。
J divergence was calculated as the measure of Hierarchical Clustering Method and it was applied to the classification of the AR model of the signals acquired from the different working conditions. The standard AR model was build up based on the classification results. Then the method was applied to the fault diagnosis of the reciprocation compressor valve. Compared with other measures, J divergence is proved to be more effective.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2007年第4期431-433,共3页
China Mechanical Engineering
关键词
J散度
谱系聚类法
AR模型
故障诊断
J divergence
hierarchical clustering method
AR model
fault diagnosis
