Work

Prediction of Vehicle Interior Wind Noise Based on Shape Features Using the WOA-Xception Model

Year: 2025

Type: article

Abstract: In order to confront the challenge of efficiently evaluating interior wind noise levels in passenger vehicles during the early stages of shape design, this paper proposes a methodology for predicting interior wind noise. The methodology integrates vehicle shape features with a whale optimization Xception model (WOA-Xception). A nonlinear mapping model is constructed between the vehicle shape features and the wind noise level at the driver’s right ear. This model is constructed using key exterior parameters, which are extracted from wind tunnel test data under typical operating conditions. The exterior parameters include the front windshield, A-pillar, and roof. The key hyperparameters of the Xception model are adaptively optimized using the whale optimization algorithm to improve the prediction accuracy and generalization ability of the model. The prediction results on the test set demonstrate that the WOA-Xception model attains mean absolute percentage error (MAPE) values of 9.78% and 9.46% and root mean square error (RMSE) values of 3.73 and 4.06, respectively, for sedan and Sports Utility Vehicle (SUV) samples, with prediction trends that align with the measured data. A comparative analysis with traditional Xception, WOA-LSTM, and Long Short-Term Memory (LSTM) models further validates the advantages of this model in terms of accuracy and stability, and it still maintains good generalization ability on an independent validation set (mean absolute percentage error of 9.45% and 9.68%, root mean square error of 3.77 and 4.15, respectively). The research findings provide an efficient and feasible technical approach for the rapid assessment of in-vehicle wind noise performance and offer a theoretical basis and engineering references for noise, vibration, and harshness (NVH) optimization design during the early shape phase of vehicle development.

Source: Machines

DOI: 10.3390/machines13060497