Azimuthal Mode Characteristics of Rotating Instability in Axial Compressor Using Compressed Sensing Method
Abstract A signal reconstruction algorithm based on the compressed sensing (CS) theory with dual-uniform sampling point (DUSP) distribution is developed and applied to identify the circumferential mode of axial compressor. A regular failure signal pattern is found and the corresponding explanation is presented with validation. Circumferential mode analysis is applied to both numerical and experimental pressure fluctuation signals of rotating instability in the axial compressor tip region. For numerical calculations, the signal in the circumferential mode domain is reconstructed by the CS with random measurement points and DUSP respectively. The success rates and the reconstruction errors are discussed in details. It is shown that the circumferential mode reconstruction method based on CS combined with DUSP is capable to identify the complex flow modes in tip region of axial compressor. For the experimental results, high circumferential mode numbers are reconstructed based on dynamic pressure signals measured by DUSP. Circumferential mode analysis efficiency is thereby significantly improved. The time-resolved characteristics of the rotating instability (RI) is discussed. Moreover, a robustness analysis is conducted, demonstrating the ability of the CS-based method with DUSP to address fault sensor problems.