基于加权相空间重构降噪及样本熵的齿轮故障分类

Gear fault classification based on weighted phase space reconstruction and sample entropy

提出了一种基于加权相空间重构降噪及样本熵的齿轮故障分类方法,给出了加权相空间重构降噪及样本熵的原理及计算公式。该方法将一维的时间序列重构到高维的相空间,通过区分吸引子在高维空间的不同的属性与特征,对原始信号进行加权相空间重构降噪,再计算降噪后信号的样本熵从而实现对齿轮故障信号的分类。对该方法进行了仿真与实验研究,结果表明,降噪后的信号有效地抑制了噪声对实验结果的影响,使得样本熵能够对齿轮不同的工作状态进行有效区分。

The vibration signal of the working gear box mixed with a lot of noise is strong nonlinear, which leads to the fact that identification of the working status of the gear is difficult. A fault classification method based on weighted phase space reconstruction and sample entropy is proposed, the principle and calculation formula for the weighted phase space reconstruction noise reduction and sample entropy method is given. It makes the time series continuation from one dimension to higher dimension, which can indicate the feature not well discriminated in one dimension to the attractor easily discriminated in higher dimension. The noise mixed in original vibration signal is reduced, and then the sample entropy of the processed signal is calculated, which realizes classification of the fault signal for gear. The simulation and experiment is carried out. The results show that the processed signal can suppress the influence of noise to the experiment, which make that the different working state for the gear can be discriminated by sample entropy.