基于压缩感知的多级风扇周向声模态重构

基于风扇单音噪声周向声模态在叶片通过频率下稀疏的特性，研究压缩感知方法在多级风扇周向主导声模态识别和幅值重构中的应用。开展多级风扇试验器的声场测量试验，利用压缩感知方法对管道周向声模态进行准确识别与分解。建立声模态重构的压缩感知模型，实现少数传声器对主导声模态的阶数识别和幅值精确重构，同时研究传声器数量和布局策略对重构精度的影响。试验结果表明:对于确定声场，当压缩感知常量C大于3.6时，随机布局的传声器有90%以上的概率可以以较高精度重构主导声模态幅值；建立压缩感知模型观测矩阵时应避免传声器的均匀和近似均匀布局，均匀布局的感知模型难以精准重构声模态。 

Azimuthal acoustic mode reconstruction of multi-stage fan based on compressive sensing

Owing to the sparsity of azimuthal acoustic modes at the blade passing frequency, a compressive sensing-based method for azimuthal dominant mode detection of multi-stage fan was proposed to realize the order identification and amplitude reconstruction.Acoustic field experiment for a multi-stage fan was carried out.The compressive sensing model with azimuthal acoustic modes reconstruction was established, which can detect the dominant mode accurately from fewer microphones.The influences of the number and placement strategy of microphones on the reconstruction accuracy were discussed.Under the determined acoustic field, experimental results showed that the signal can be accurately reconstructed with a probability of over 90% when the compressive sensing constant C was greater than 3.6. Both uniform and approximately uniform placement of microphones should be avoided when designing the measurement matrix, which may lead to the failure of acoustic mode reconstruction.