Kohonen网络故障诊断方法及试验

Fault Diagnosis Means and Experiment Based on Kohonen Neural Network

根据 Kohonen神经网络诊断的工作原理、诊断特征 ,提出了渔船轴系模拟试验台的系统结构和振动监测方法 ,并通过自行开发的数据采集系统和诊断软件 ,对故障特征矢量进行识别和诊断。

In this paper the fault diagnosis system based on Kohonen neural network is introduced. The principal, characteristics and mathematical model of this network are studied and it is applied to fault diagnosis system of fisher ship shafting. The structure and principal of a ship shafting simulation test bed are discussed. By using the data acquisition system and developing the software system, the shafting vibration signal and its FFT chart are obtained. Then the frequency chart is analyzed to obtain its eigenvector. The Kohonen neural network is used to identify the eigenvector and conclude the fault reason. A real example authenticates that Kohonen neural network is a meritorious fault diagnosis means of ship shafting.