基于复杂度的轴流风机动叶偏移时声压信号的表征

Characterization on Sound Pressure Signal of an Axial Flow Fan with Abnormal Blade Angle Based on Complexity Algorithms

以OB-84型单级动叶可调轴流风机为研究对象,在大涡模拟LES基础上,引入FW-H声学模型,模拟得到动叶安装角异常时风机各区域的声压信号分布,并与正常工况进行了比较.基于功率谱密度特征,对原始声压信号进行小波分解与重构,并提取重构信号的特征熵,探讨了3种复杂度算法及其表征能力.结果表明:不同区域内声压时域特征各异,当动叶安装角异常时,声压信号的脉动形态、脉动区间和幅值均发生改变;该风机噪声信号在低频处呈现出明显的离散峰值特征,且能量逐级递减,而随偏离度增加,宽频噪声特性更加突出;动叶区声压信号特征熵最低,规律性很强,集流区和扩压区的气动特性和噪声信号相对稳定,可用近似熵、样本熵和Lempel-Ziv复杂度来表征风机单动叶偏离程度.

Taking a variable pitch axial flow fan of OB-84 type as an example, the distribution of sound pressure in each region was simulated by large eddy simulation with FW-H noise model, and the results under abnormal installation angles were compared with that under normal conditions. Based on the analysis results of power spectral density, wavelet decomposition and reconstruction were conducted to the original sound pressure signal, while characteristic entropy of the reconstructed signal was abstracted, so as to study the characterization capabilities of three complexity algorithms. Results show that the sound pres- sure in various regions of the axial flow fan is obviously different in time domain, and the pulsation mor- phology, interval and amplitude vary with the abnormal installation angle. The noise signal is featured by discrete peaks in low-frequency region, with energy level gradually reduced, while broadband noise be- comes notable with the rise of deviation degree. The characteristic entropy in blade region is the least due to uniform regularity, and the aerodynamic characteristics and noise signal remain relatively stable in afflux section and diffusion section, thus the abnormal degree of blade installation angle could be evaluated by ap- proximate entropy, sample entropy and Lempel-Ziv complexity.