Toward a probabilistic approach of vehicle noise assessment Seminar

In this context, noise assessment against objective criteria remains difficult mostly because today simulation load cases only cover a small portion of the operating range of vehicles. Thus, it is proposed to build vehicle interior noise indicators, based on classical annoyance/exposure criteria such as the equivalent level or statistical level indicators.

Within the current simulation process low frequency noise –either booming or rolling noise- is considered as a product of forces and vibroacoustic frequency response functions (FRFs). Both terms are subject to uncertainties and variability that enforce a probabilistic approach to apply the proposed indicator. For this purpose, forces as well a FRFs require a probabilistic modeling.

First, a single vehicle’s noise in real driving conditions will be analyzed as a random process and modelled using random functions of operating conditions parameters such as rpm or delivered torque. Since only one vehicle is considered, noise dispersion is mostly caused by the dynamic excitation dispersion that is observed indirectly.

FRFs dispersion can be observed more directly through end-of-line measurements. After experimental results are presented, a non-parametric modeling is introduced considering uncertainties as a whole. Such a method provides an efficient way to build the probability laws of a vehicle structural-acoustic FRF.