Prediction of Trimmed Body Noise Transfer Function Based on BIW Response Using Machine Learning

This work focuses on the prediction of Trimmed Body Noise Transfer Function (NTF) using Glazed BIW (body in white) structural model characteristics by leveraging Machine Learning (ML) technique. Inputs such as Glazed BIW (GBIW) attachment dynamic stiffness, Body Panel Vibration Transfer Functions (VTF) and Driver Ear level NTFs are employed to predict Trimmed Body NTF for a particular hard point. An iterative process of performing design modifications on the BIW to verify its effect on BIW performance and therefore on Trimmed body NTF is undertaken. BIW geometric parameters are varied in an organized manner to generate hundreds of data points at GBIW level which are provided as input to the train the ML model to predict the trimmed body level NTF. The outcome provides crucial insights of how the trimmed body NTF is closely related to the GBIW design characteristics. This ML approach of predicting trimmed body NTF based on GBIW characteristics provides critical insight about GBIW design during early stages of product evolution, which benefits in quick decision making rather than the conventional approach of evaluating complex trimmed body simulations.