基于局部均值分解与快速独立成分分析的潜水泵故障诊断

针对潜水泵故障诊断与分析时的信号干扰问题,提出使用局部均值分解与快速独立成分分析相结合的方法。将采集的振动信号进行局部均值分解,得出一系列的乘积函数分量,计算出每个分量与振动信号的互相关系数,选取互相关系数较大的分量进行重构并与振动信号组成矩阵,对矩阵进行快速独立成分分析处理,将处理后的有效数据进行频谱分析,得到最后的频谱图。通过具体实验,对比局部均值分解与快速独立成分分析处理后的正常运行和故障运行的频谱图,实现故障分析与诊断。实验结果表明,经过局部均值分解与快速独立成分分析处理的频谱图能够清晰地分辨出故障特征频率,提高了故障诊断的准确性。

Fault Diagnosis of Submersible Pumps Based on Local Mean Decomposition and Fast Independent Component Analysis

Aiming at the signal interference problem in fault diagnosis and analysis of submersible pumps, a method of combining local mean decomposition with independent component analysis was proposed. A series of product function components were obtained by decomposing the local mean value of the collected vibration signal, and the cross-correlation coefficient between each component and the vibration signal was calculated. The components with large cross-correlation number were selected to reconstruct and a matrix with the vibration signal was formed. The matrix was processed by fast independent component analysis, and the processed effective data was analyzed by spectrum analysis to get the final spectrum. Through specific experiments, the spectrum diagram of normal operation and fault operation after local mean decomposition and independent component analysis were compared, and the fault analysis and diagnosis were realized. Experimental results showed that the spectrum diagram processed by local mean decomposition and independent component analysis can clearly distinguish the fault characteristic frequency and improve the accuracy of fault diagnosis.