基于多尺度局部最大样本熵的液压泵故障特征提取

The Feature Extraction of Hydraulic Pump Based on Multiscale Local-maximum Sample Entropy

提出了一种新的表征时间序列复杂度的方法——多尺度局部最大样本熵。多尺度局部最大样本熵不仅克服了样本熵只能在单一尺度上衡量时间序列复杂度的缺点,而且与多尺度熵相比,既提高了每个时间尺度上样本熵的精度,又抑制了振动信号中的噪声和干扰成分。通过对仿真信号的对比分析,验证了多尺度局部最大熵在处理振动信号上的优势,将其应用到液压泵振动信号的特征提取中,很好地区分出了液压泵的不同故障。

A new method, multiscale local-maximum sample entropy is proposed to measure the complexity of time series. It not only overcomes the drawback of sample entropy, but also can suppress the noise and interference of vibration signals and improves the precision of sample entropy on each time scale compared with multi-scale entropy. Firstly, we verify the advantage of multiscale local-maximum sample entropy on processing vibration signal through the contrastive analysis of simulation signal. Then, we apply it to feature extraction of hydraulic pump＇ s vibration signal. It can identify different features of a hydraulic pump quite obviously.