基于CYCBD和参数自适应DMD的滚动轴承微弱故障诊断

Weak fault diagnosis for rolling bearings based on CYCBD and parameter-adaptive DMD

在旋转机械故障诊断中,传感器采集到的滚动轴承振动信号容易受到多个激励源的耦合影响,特别是微弱故障特征往往淹没在背景噪声中,给故障诊断带来挑战。为了抑制无关激励源并增强故障特征,本文提出一种基于最大二阶循环平稳盲解卷积(CYCBD)和参数自适应动态模式分解(DMD)的滚动轴承微弱故障诊断方法。首先根据包络谐波乘积谱(EHPS)估计CYCBD算法的关键参数循环频率α,将原始信号经CYCBD处理使故障特征得到增强;然后将CYCBD处理后的信号通过参数自适应DMD进一步消除残余噪声,其中DMD的最优截断秩参数r通过遗传算法自适应确定;最后通过包络谱分析来提取故障特征频率及倍频。仿真和试验结果证明了本方法的有效性。

In fault diagnosis of rotating machinery,the rolling bearing vibration signals acquired by sensors are susceptible to coupling effects of multiple excitation sources.Especially,the weak fault features are often submerged in background noise,which presents challenges to the diagnosis.To suppress irrelevant excitation sources and enhance fault features,this paper proposed a weak fault diagnosis method for rolling bearings based on maximum second-order cyclostationary blind deconvolution(CYCBD)and parameter-adaptive dynamic mode decomposition(DMD).Firstly,the key parameterαcalled cyclic frequency of CYCBD algorithm was estimated via envelope harmonic product spectrum(EHPS),and the original signal was processed through CYCBD to enhance the fault features.Then,the treated signal was further denoised by using parameter-adaptive DMD,where the optimal truncation rank parameter r in the DMD procedure was adaptively determined by genetic algorithm.Finally,envelope spectral analysis was performed to extract fault feature frequency and its harmonic frequencies.Both simulation and experimental results demonstrate the effectiveness of this method.