基于改进VMD的离心泵空化声发射信号特征提取

针对变分模态分解算法中分解层数和惩罚因子不易确定的问题,提出一种改进变分模态分解(improved variational mode decomposition,IVMD)算法,并将其应用于离心泵空化声发射信号特征提取.应用IVMD算法时,首先根据包络熵差异系数确定变分模态分解的分解层数;然后采用人工蜂群算法优化得出惩罚因子,并将其作为变分模态分解的最佳输入参数.利用IVMD算法对仿真信号进行分析,并与集合经验模态分解结果进行比较.以60%额定流量下采集到的离心泵进口处的声发射信号为例进行IVMD计算,分析携带原信号大量信息的信号分量的频域特征及其绝对能量随离心泵空化状态变化的关系.结果表明:IVMD算法能够择优确定分解层数和惩罚因子,实现非平稳信号的自适应分解.反映离心泵空化状态的声发射信号特征频率集中在50,100 kHz及其附近.随着离心泵空化从无到有、从弱到强的变化,这2个特征频率范围信号分量绝对能量值呈“基本保持不变-减小-增大”的变化规律.

Feature extraction of cavitation acoustic emission signal of centrifugal pump based on improved variational mode decomposition

Aiming at the problem that the decomposition layer and penalty factor in the variational mode decomposition(VMD)algorithm are difficult to be determined, an improved VMD(IVMD)algorithm was proposed and applied to extract the feature of cavitation acoustic emission(AE)signal of centrifugal pump. In the application of IVMD algorithm, first, the envelope entropy diffe-rence coefficient was used to determine the number of decomposition layers, and then the artificial bee colony algorithm was used to optimize the penalty factor which was employed as the optimum input parameter of VMD. The simulated signal was processed via IVMD algorithm, and the results were compared with those by ensemble empirical mode decomposition. AE signals collected from the inlet of centrifugal pump under 60% of its rating flow were chosen and processed via the IVMD algorithm. The frequency domain features of those signal components carrying a great deal of information in the original signals were analyzed, and so was the variation of their absolute energy with changing cavitation stages. The results show that the number of decomposition layers and the penalty factor could be determined optimally via the IVMD algorithm, which helps to decompose the unstable signals adaptively. The cha-racteristic frequency ranges of AE signal focus on 50, 100 kHz and their respective vicinities. With cavitation changing from scratch and from weak to strong, the absolute energies of AE signal components in the above frequency ranges remain at a certain level first, then decrease, and finally increase.