This study introduces IntakeSQ, a novel Sound Quality (SQ) metric for predicting intake noise perception during the early design phase. Utilizing 1D simulation and experimental measurements, various resonator configurations were implemented and evaluated. Subjective tests were conducted with a jury of 50 individuals, and a final Multiple Linear Regression (MLR) model was developed using a minimal set of independent metrics: Speech Interference Level (SIL), Sharpness ( S d ), and Tonality ( T ). The model achieved high predictive accuracy with a total coefficient of determination ( R 2 ) of 0.922 and an Adjusted R 2 of 0.888. To ensure robustness, the metric was rigorously verified through Leave-One-Out Cross-Validation (LOOCV), yielding a LOOCV R 2 of 0.836 and a Mean Squared Error (MSE) of 0.160. Furthermore, the model was successfully validated across 12 different vehicles, confirming its generalizability. By enabling perception prediction directly from engine compartment data, this standalone assessment tool bypasses the need for late-stage full-vehicle prototypes and complex transfer function modeling, allowing for effective intake system optimization at the component level.
Özkan et al. (Fri,) studied this question.