基于SGMD-Autogram的液压泵故障诊断方法研究

Hydraulic pump fault diagnosis method based on SGMD-autogram

辛几何模态分解方法(Symplectic Geometry Mode Decomposition,SGMD)存在特征信息分布过于分散问题、Autogram方法中的最大重复离散小波变换(Maximal Overlap Discrete Wavelet Packet Transform,MODWPT)存在特征提取能力不足问题,针对上述两问题,提出了基于SGMD-Autogram的新方法。对实测液压泵多模态故障振动信号进行SGMD分解;针对分解后产生的特征信息分布过于分散问题,提出基于最大无偏自相关谱峭度法,筛选含有丰富运行特征信息的模态分量为数据源,进而取代MODWPT,实现最优故障特征提取;对数据源进行阈值处理,并基于频谱实现对液压泵故障的诊断。通过对比分析仿真和实测液压泵斜盘故障振动信号,验证了该方法可以有效地诊断斜盘故障。

The symplectic geometry mode decomposition(SGMD)method has the problem of feature information distribution being too scattered,and Autogram method has the problem of the feature extraction ablility of the maximal overlap discrete wavelet packet transform(MODWPT)being not strong.Here,aiming at above problems,a new method based on SGMD and Autogram was proposed.Firstly,SGMD was applied to decompose measured multi-mode vibration fault signals of hydraulic pump.Secondly,aiming at the problem of too scattered feature information distribution after decomposition,the spectral kurtosis method based on maximum unbiased autocorrelation was proposed.Some mode components containing rich operating feature information were screened out as the data source to replace MODWPT,and realize the extraction of the optimal fault features.Finally,the data source was processed with threshold,and the fault diagnosis of hydraulic pump was realized based on spectrum.By comparing and analyzing simulated and measured vibration fault signals of hydraulic pump swash plate,it was verified that the proposed method can effectively diagnose hydraulic pump swash plate faults.