摘要
提出一种结合小波包分解和模糊神经网络的故障诊断方法,采用小波包分解与重构提取各频带的能量作为故障特征向量,并以此为学习样本,再利用正交最小二乘学习算法训练模糊神经网络,确定故障诊断系统模型,对轴承故障进行诊断和识别.仿真结果及与其它一些方法比较表明:该轴承故障诊断方法可以有效识别和预测轴承的状态,且学习效率、准确性和可靠性等方面均有较大提高.
A novel method for fault diagnosis combining wavelet packet decomposition and fuzzy neural network（FNN）was proposed.The eigenvectors were extracted with energyofeach band bywavelet packet decomposition,and were taken as learning sample.Then training fuzzy neural network by orthogonal least squares（OLS）learning algorithm,and building model of fault diagnosis system to diagnose and recognize bearing fault.Simulation results and comprehensive comparisons with some other approaches prove the proposed method efficiently recognize and predict the state of bearing fault diagnosis,and learning efficiency,accuracy and reliability are greatly enhanced.
出处
《湖南科技大学学报（自然科学版）》
CAS
北大核心
2010年第2期28-31,共4页
Journal of Hunan University of Science And Technology：Natural Science Edition
基金
教育部重点科研基金项目(208098)
湖南省教育厅重点科研基金项目(07A056)
