基于RBF网络和ARX模型的液压系统故障诊断方法  

Fault Diagnosis Approach of Hydraulic System Using RBF Network and ARX Model

作　　者：贺湘宇[1,2] 何清华[2] 邹湘伏[2] 谢习华[2] 黄志雄[2] 

机构地区：[1]长沙理工大学汽车与机械工程学院,长沙410004 [2]中南大学机电工程学院,长沙410083

出　　处：《系统仿真学报》2009年第1期282-285,共4页Journal of System Simulation

基　　金："863"国家高技术研究发展计划项目(2003AA430200)

摘　　要：提出了基于径向基函数(Radial Basis Function,RBF)网络和有源自回归(Auto-Regressive with Extra Inputs,ARX)模型的液压系统的故障诊断方法。作为一种性能优越的网络分类器,RBF网络比传统的反向传播(Back Propagation,BP)网络表现出更好的分类效果,非常适合于故障特征识别。故障诊断方法首先针对目标故障状态建立ARX模型,提取ARX模型的自回归系数作为故障特征向量。然后将故障特征向量作为RBF网络训练样本,建立RBF网络故障分类器,进一步根据RBF网络的输出结果来判断故障的类型。通过建立挖掘机铲斗部分液压系统仿真模型,验证了于基于RBF网络和ARX模型的故障诊断方法的有效性。A fault diagnosis approach of hydraulic system using radial basis function （RBF） network and auto-regressive with extra inputs （ARX） model was put forward. As an excellent nonlinear classifier, RBF network could reach better performance than could be achieved by classical back propagation （BP） network, therefore RBF network is feasible to be applied to fault classification. Firstly, several ARX models were established for target faults. Auto-regressive parameters extracted from the model were regarded as fault feature vectors. Secondly, those vectors were served as training data and the RBF network fault classifier was established. Then, fault patterns could be identified by the output of the classifier. A simulation model of hydraulic system for excavator’s bucket was established to verify the proposed fault diagnosis approach and simulation result shows that the approach is effective.

关 键 词：故障诊断 液压系统 挖掘机 径向基函数网络 有源自回归模型 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]