基于LMD和RBF结合的齿轮箱故障诊断

Gearbox Fault Diagnosis Based on Local Mean Decomposition( LMD) and Radial Basis Function Neural Network( RBF)

局域均值分解(Local Mean Decomposition,LMD)是近年来出现的一种新的时频分析方法,在机械设备故障诊断领域中的应用日益广泛。针对齿轮箱振动故障信号的非平稳性和非线性,提出了一种基于局域均值分解和径向基函数神经网络(Radial Basis Function Neural Network,RBF)相结合的齿轮箱故障诊断方法。该方法利用小波包对原始信号进行消噪;利用LMD对处理后信号进行分解,得到一系列PF分量(Product Function,PF);选取包含主要故障信息的PF分量并从中提取偏度系数等特征参数对RBF神经网络进行训练,并对齿轮箱故障进行识别和分类。通过实例验证了该方法的有效性。

Local mean decomposition（ LMD） is a new time-frequency analysis method appeared in recent years,the field of application in fault diagnosis increasingly widespread. A fault diagnosis approach for Gearbox based on local mean decomposition and Radial Basis Function Neural Network（ RBF） was proposed for non-steady and nonlinear signal. Wavelet packet was carried out to preprocess the signal which contains large of background noise. The LMD was used to decompose the preprocessed signal into many of Product Function（ PF） components. The PF component was selected which contains main fault information and feature vector such as the coefficient of skewness was extracted,and feature vector was put into the RBF neural network for the fault identification and classifier. The examples shows that this method is effective.