基于分段累计近似与自适应噪声辅助集成经验模态分解的滚动轴承故障诊断方法

Fault Diagnosis Method for Rolling Bearings Based on PAA and CEEMDAN

提出了一种基于分段累积近似(PAA)与自适应噪声辅助集成经验模态分解(CEEMDAN)的适用于长信号处理的滚动轴承故障诊断方法,首先对高频采样得到的长信号进行包络解调,再对得到的低频包络信号进行PAA处理,实现对包络信号的数据压缩,最后通过CEEMDAN分离出包含故障信息的内禀模态函数(IMF)分量,进而实现对轴承故障的诊断。该方法通过数据压缩,解决了CEEMDAN因计算复杂而无法处理长信号的问题;先解调再PAA压缩,不会削弱信号中的低频包络信号(即故障特征信号)。实际案例的处理结果验证了该方法的有效性。

A fault diagnosis method for rolling bearings based on piecewise aggregate approximation(PAA) and complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN) is proposed for long signal processing. Firstly, the envelope demodulation is performed on long signal obtained from high-frequency sampling. Then, PAA is performed on low-frequency envelope signal to achieve the data compression of envelope signal. Finally, the intrinsic mode functions(IMF) components containing fault information are separated through CEEMDAN, and the fault diagnosis for bearings is realized. By data compression, the proposed method solves the problem that CEEMDAN cannot process long signal due to its complex computation. Demodulation followed by PAA compression will not weaken the low-frequency envelope signal(i.e. fault characteristic signal) in signal. The processing results of an actual case verify the effectiveness of the method.