基于LMD和SVDD的滚动轴承健康状态评估

Roller Bearing Health Condition Assessment Based on LMD and SVDD

为了提高滚动轴承健康状态评估的分类精度,提出了基于局部均值分解(Local mean decomposition,简称LMD)和具有故障样本的支持向量数据描述(Support Vector Data Description,简称SVDD)相结合的滚动轴承故障状态识别方法。该方法首先将利用LMD方法进行滚动轴承振动信号的分解,得到一系列PF(乘积函数,product function)分量之和并具有物理意义,接下来对含有主要故障信息的PF分量进行能量计算并构造特征向量,最后将其输入SVDD分类器,进行滚动轴承的健康状态评估。实验结果证明该方法的可行性和有效性。

In order to improve the classification accuracy of rolling bearing health assessment,a fault state recognition method based on local mean decomposition(LMD)and support vector data description(SVDD)with fault samples is proposed. The method first decomposed rolling bearing vibration signals into several sum of PF(product function,PF) component with physical significance by the LMD method,then calculated the energy of PF component containing the main fault information and structured characteristics,finally input them into SVDD classifier and assessed the health condition of rolling bearing.The experimental results prove that the proposed method is feasible and effective.