Speaker
Description
The in-service acoustic performance of fibrous absorbers in aircraft fuselage linings is affected by installation variability and environmental exposure, yet their relative contributions remain poorly quantified. This work reframes the problem as a feature detectability task, where the sound absorption coefficient is used to infer the physical parameters governing its variability. Two experimental datasets are constructed from impedance tube measurements under controlled mounting conditions and humidity exposure, including cyclic moisture loading representative of flight operations. Supervised classification models are employed as qualitative discriminators: classification accuracy is used to assess whether a given feature leaves a detectable signature in the absorption curves. SHAP-based explainability is then used to quantify the frequency-dependent contribution of each feature. Results show that material type, protective layers, and humidity-related parameters are consistently detectable, with classification accuracies up to 99% and 85%, respectively. In contrast, installation effects remain indistinguishable from the random baseline under controlled mounting conditions. Moisture effects are strongly frequency dependent, with dominant contributions grouped in three bands, and exhibit progressive degradation under cyclic exposure. The proposed framework provides a methodology able to isolate and quantify the physical drivers of acoustic variability in porous materials under realistic operating conditions.