Vibroacoustic properties specify among others the quality of vehicles and contribute to customer satisfaction. In the early design phase, a finite element (FE) analysis is usually carried out to predict and design vibroacoustic properties.
Nevertheless, the inevitable lack of information adversely affects the model-based prediction like vibroacoustic vehicle properties. For instance, a deterministic one-point model parameter like Young’s Modulus, which ignores the variance due to manufacturing tolerance, leads to among others an incomplete determination of real vibroacoustic behavior of the vehicle in FE simulation.
At this point, uncertainty is only partially taken into account in vehicle vibroacoustic. This paper contributes to mastering uncertainty in model-based prediction of vibroacoustic vehicle properties. A category comprising data-, model- and structural uncertainty and their mathematical description is preliminarily introduced to provide an insight into uncertainty in the early design phase of vehicle vibroacoustic. In regard to this, this paper introduces the concept and framework to treat uncertainty in vehicle vibroacoustics. It is composed of the following three steps: (1) Data uncertainty is quantified and can be reduced by a probabilistic approach like Bayesian Inference based statistical model calibration using the information of existing data. (2) Model uncertainty can then be treated by capturing and calibrating the discrepancy function. (3) Structural uncertainty can be treated by exploring the topology of structure through the occurrence probability of equipment variants. Possible challenges of the proposed concept like computational effort are highlighted and the future implementation of the approach is briefly described.