A tutorial example of duct acoustics mode detections with machine-learning-based compressive sensing.

Abstract

Acoustic beamforming and mode detections by means of machine learning have potential advantages over conventional strategies, e.g., first-principle based forward acoustic models may be replaced by neural networks. In this work, the machine-learning-based strategy is presented for aeroengine duct acoustic mode detections and the focus is on the associated machine learning implementation. Next, the proposed neural network implementation is incorporated into compressive sensing by taking into account specific acoustic mode detection requirements. The proposed method shall direct the research attention of acoustic measurements to machine learning and particularly benefit mode detections for next-generation aircraft engine problems.