摘要
针对设备运行状态和特征参数之间存在的复杂非线性关系,提出了基于主成分分析的RBF神经网络故障诊断方法;该方法用主成分分析方法将高维相关特征参数转化为低维互不相关的特征参数,在此基础上建立了RBF网络分类器;用该网络对某汽轮机减速箱的运行状态进行识别,理论分析和实验结果表明,基于PCA和RBF网络方法的诊断技术具有模型简单、检测速度快等优点,可以在实际应用中发挥有效作用。
Pointing at the complicated relation between running status of equipment and characteristic parameters, PCA and RBF neural network based fault diagnosis technology is put forward. Using PCA technology, multi--dimensional correlated characteristic parameters are transformed into low dimensional independent characteristic parameter, a RBF Neural Network classifier is built. And the neural network is used to recognize a turbine reducer box running status. Theory analysis and experiment result indicates that PCA and RBF neural network based diagnosis technology has simple model, quick detecting speed, and can play effective effect in practical application.
出处
《计算机测量与控制》
CSCD
2008年第7期903-905,共3页
Computer Measurement &Control
基金
陕西科技大学自然科学基金资助项目(ZX05-37)
陕西省工业攻关项目(2006K05-G18)
