基于WVPMCD和层次模糊熵的液压泵故障诊断方法研究

Method of Fault Diagnosis of Hydraulic Pump Based on WVPMCD and Hierarchical Fuzzy Entropy

摘要:

为了更准确地对液压泵进行故障诊断,提出了基于WVPMCD(WLS-Variable predictive mode based class discriminate,WVPMCD)和层次模糊熵(hierarchical fuzzy entropy,HFE)的故障诊断方法;由于液压泵振动信号比较复杂,基于变量预测模型的模式识别(variable predictive mode based class discriminate,VPMCD)方法在对模型参数进行估计时会出现异方差的现象,从而导致参数估计出现病态,估计所得参数不稳定,从而降低预测精度；WVPMCD作为VPMCD的改进,采用更先进的加权最小二乘参数估计法代替最小二乘参数估计法,消除异方差的影响,提高参数估计的精度,进而提高液压泵故障诊断准确率；此外,在层次熵(HierarchicalEntropy,HE)的基础上提出了层次模糊熵的概念,模糊熵作为样本熵的改进,在衡量时间序列复杂度上并比样本熵更优越；运用WVPMCD和层次模糊熵对液压泵进行故障诊断,实验结果验证了该方法的有效性。

Abstract:

In order to diagnose hydraulic pump more accurately,a new approach based on WVPMCD and HFE is proposed. Because hydraulic pump's vibation signal is complex,VPMCD’s parameter estimation will be heteroscedastic.And then,it will resault in morbidity and the gained parameter will be unstability.Thus, the prediction accuracy will decrease.As improvement of VPMCD,WVPMCD uses the advanced weighted least square parameter estimation method to replace ordinary least square parameter estimation method,which remove the affect of heteroscedasticity and improve accuracy of parameter estimation.Thus,accuracy of pattern recognition will be also increase.Furthermore,HFE is proposed on the base of HE.As improvement of SE,FE have advantage in measurement of time series’ complexity.The proposed algorithm is verified by experimental data analysis.