基于双谱和支持向量数据描述的故障诊断

INTELLIGENT FAULT DIAGNOSIS BASED ON BISPECTRUM AND SUPPORT VECTOR DATA DESCRIPTION

双谱是处理非线性、非高斯性信号的有力工具,而支持向量数据描述(support vector data description,SVDD)是一种单值分类方法,可以解决故障诊断中故障样本缺乏的问题。基于此,提出一种基于双谱和SVDD相结合的故障智能诊断方法。该方法采用双谱对振动信号进行处理并提取特征向量,以此作为SVDD的输入参数进行训练和分类。将该方法应用于滚动轴承的故障诊断中,结果表明,该方法可以有效提取轴承信号的特征信息,提高SVDD在故障诊断中的准确性。

Bispectrum is a powerful processing tool for non-linear and non-Gaussian signal.Support vector data description（SVDD） is a one class classification method.It can solve the problem of insufficient fault samples in fault diagnosis.Combining bispectrum and SVDD,a new fault diagnosis approach using bispectrum and SVDD is proposed.In the new approach,bispectrum is used as signal preprocessing to extract the feature vector.Then the feature vector can be served as the input parameters of SVDD for classification.Applying this approach to the rolling bearing fault diagnosis,the results show that the proposed approach is very efficient to extract the signal feature and improve the veracity of SVDD in fault diagnosis significantly.