Wavelet packet extraction characteristics of pulsating oil flow pressure vibration signal

This study presents a wavelet packet decomposition (WPD)-based approach for extracting multi-scale vibration characteristics from pressure signals of pulsating oil flow containing random particulate contaminants. Through a multi-criteria evaluation framework – including theoretical screening, frequency component preservation analysis, and root mean square (RMS) reconstruction error minimisation (0.0040) – the sym8 wavelet was selected as the optimal basis function. Four-layer WPD decomposition and energy distribution analysis identified five dominant frequency bands (8.7891 Hz to 437.5 Hz) exhibiting maximal vibration energy and clear quasi-periodicity. Statistical validation confirmed the significance of these bands (analysis of variance (ANOVA): F = 28.46, p < 0.001), with long-term stability deviations below 5.3% over 8 h and repeatability coefficient of variation of 4.12%. The method achieves a processing time of 0.083 s per data segment, supporting real-time deployment. This approach robustly isolates quasi-periodic features despite particulate interference, providing a quantitative foundation for hydraulic equipment condition monitoring and early fault detection in oil transport systems.