Analysis and Experiment of the Compressive Sensing Approach for Duct Mode Detection

摘要

This paper describes a new mode detection method based on the compressive sensing approach, which has been developed in information technology to reduce samples required by the classical Nyquist-Shannon sampling theory. A revised approach is proposed here to detect azimuthal spinning modes from aeroengine fan noise by using a microphone array outside the bypass duct. A number of numerical simulations is prepared to examine the associated test performance, such as the reconstruction accuracy, robustness, background noise interference, and coherence impact. The mode sparsity s is usually much smaller than the highest order of the modes, so the azimuthal mode is sparse at the frequency of interest to ensure a satisfactory compressive sensing-based mode detection. The current simulations show 1.5s log N/s sensors shall be generally sufficient rather than N sensors required by the Shannon-Nyquist sampling theory. An experimental system is designed and implemented to demonstrate the proposed method. The test system contains a straightened and rigid duct with an enclosed spinning mode synthesizer to emulate fan noise propagation inside and scattering from a bypass duct. An outer sensor array is employed to demonstrate that the compressive sensing-based mode detection method can significantly reduce the required number of sensors, especially for modes of high order, which results in a much simplified sensor array design for aerospace noise tests.