自组织映射法在机械故障诊断中的应用  

Application of the Self-Organizing Map to Fault Diagnosis of Machinery

作　　者：黄士涛[1] 孙惠国[1] 

机构地区：[1]郑州工业大学机械系

出　　处：《郑州工业大学学报》1998年第2期9-13,共5页Journal of Zhengzhou University of Technology

摘　　要：自组织映射法是由Ｔ．Ｋｏｈｏｎｅｎ提出的一种人工神经网络模式，它能把高维的信息数据以有序方式映射到低维的网络上，形成一种拓扑意义上的有序图。由于其有序性，使得信号空间图成为许多输入信息及其关系的群落图。每个图可用灰度深浅不等的方块阴影图表示出来，这种可视的图形表示了机械运行的状态。人们可以非常直观地从图上判断机械运行状态，从而使得机械运行状态的监测和故障诊断更为简便。The self-organizing map is an artificial neural networks model and algorithm that implememts a characteristic nonliner projection from the high-dimensional space of signal data into a low-dimensional array of neurons in an orderly fashion, which is made by T. Kohonen. The mapping tends to preserve the topological relationships of signal domains. Owing to the order, the image of the signal space tends to manifest clusters of input information and their relationship on the map. The map can be shown by square shades of gray, so it makes machinery operating conditions be visualized. The map will vary with changes of machinery operating conditions, which helps us to detect and identify the faults . 

关 键 词：神经网络 自组织映射法 故障诊断 机械 

分 类 号：TH17[机械工程—机械制造及自动化]