摘要
本文从统计意义上分析了机械设备在不同运行状态下振动信号的特征;选择信号幅值的概率分布的前n阶矩作为特征向量来进行状态信息凝聚.在此,将设备的状态分为“完好”和“故障”两种,应用模式识别技术进行状态分类.最后通过对试验数据的分析,证实了特征参数的稳定性和对故障的敏感性.结果表明,分类判据是有效的.
In this paper, the characteristics of vibration signal of machinery in different running conditions are statistically analysed, and some moments of statistical distribution of signals are selected as the eigenvector to condense the state information. Here, we divide the states of machinery into two: 'good' and 'faulty', the pattern recognition techniques are used to classify the running conditions of machinery. At the end of this paper, the authors present some test data, and from the results obtained, it's verified that the eigenvector selected is reliable and sensible to faults. And the results also show the effectiveness of classification rule.
出处
《应用数学和力学》
EI
CSCD
北大核心
1991年第8期697-701,共5页
Applied Mathematics and Mechanics
关键词
模式识别
机械设备
故障诊断
pattern recognition, condensed state information, divergence index, inter- object distance, intra-object distance
