In response to the weak signal in the early stage of bearing failure,which makes it difficult to extract and identify,NGO(northern goshawk optimization)is proposed to optimize VMD(variational mode decomposition)parameters,and the support vector machine method is combined with DBO(dung beetle optimizer)to optimize SVM(support vector machine)for fault extraction and classification recognition.Firstly,the NGO is used to search for the optimal parameters of the VMD,and the signal is decomposed into several intrinsic mode functions(IMF)using the VMD.Then,the kurtosis is used to select the optimal IMF.Finally,it is input into the DBO-VMD model for fault classification and recognition.The experimental results show that the NGO-VMD method has certain advantages in terms of iteration times and convergence accuracy.The kurtosis is used to select the optimal eigenmode function and the envelope demodulation analysis has the best ability to extract the fault features of early weak fault signals.Under the background of weak fault signal,the SVM diagnosis model can improve the classification and recognition rate of fault diagnosis to a certain extent.This method has good ability of fault feature extraction and classification recognition,and provides technical support for early fault diagnosis of rolling bearings.
关键词
滚动轴承/故障诊断/北方苍鹰算法/变分模态分解/蜣螂优化算法/支持向量机
Key words
rolling bearings/fault diagnosis/NGO/VMD/DBO/support vector machine
维普
万方数据