Expert Decision Support System for aeroacoustic source type identification using clustering

TL;DR

This paper introduces an expert decision support system that uses spectral and spatial features to cluster aeroacoustic sources, aiding experts in identifying and understanding different source types across various flow conditions.

Contribution

The paper presents a novel clustering-based decision support system that uses interpretable features independent of Mach number for aeroacoustic source identification.

Findings

Clusters mostly match expert-identified source types

Provides transparent clustering with confidence measures

Effective across different flow configurations

Abstract

This paper presents an Expert Decision Support System for the identification of time-invariant, aeroacoustic source types. The system comprises two steps: first, acoustic properties are calculated based on spectral and spatial information. Second, clustering is performed based on these properties. The clustering aims at helping and guiding an expert for quick identification of different source types, providing an understanding of how sources differ. This supports the expert in determining similar or atypical behavior. A variety of features are proposed for capturing the characteristics of the sources. These features represent aeroacoustic properties that can be interpreted by both the machine and by experts. The features are independent of the absolute Mach number which enables the proposed method to cluster data measured at different flow configurations. The method is evaluated on deconvolved beamforming data from two scaled airframe half-model measurements. For this exemplary data, the proposed support system method results in clusters that mostly correspond to the source types identified by the authors. The clustering also provides the mean feature values and the cluster hierarchy for each cluster and for each cluster member a clustering confidence. This additional information makes the results transparent and allows the expert to understand the clustering choices.