基于Hankel－DMD的立式混流泵压力脉动特性分析OA

To ensure the safe operation of vertical mixed-flow pumps in low-head pumping stations,it is necessary to accurately identify the pressure pulsations in the vaneless space.Based on numerical simulation results,time-series data was constructed and a dynamic mode decomposition approach en-hanced by the delayed-embedding property of the Hankel matrix was employed to enable the extraction of dominant dynamic features from one-dimensional signals.The frequency components obtained from this method were compared with those derived from Fourier transform analysis to verify identification accuracy.Three typical flow conditions were examined to analyze the dominant frequencies,energy dis-tribution,and modal characteristics of pressure pulsations in the vaneless region.Signal reconstruction using a limited number of low-order modes was conducted to evaluate error levels and the stability of the method.The results show that the proposed approach effectively identifies the primary frequencies induced by rotor－stator interaction and outperforms the Fourier transform in extracting low-energy com-ponents.The low-order reconstruction maintains a high degree of consistency with the original signal,demonstrating good error control.The study concludes that this method can accurately characterize pressure pulsation characteristics of the vaneless region of vertical mixed-flow pumps,and provides valuable technical support for analyzing complex time-series data and assessing the operational safety of pumping stations.