Vehicle vibro-acoustical comfort optimization using a multi-objective interval analysis method

Abstract

Vehicle interior vibro-acoustical comfort, i.e., noise, vibration and harshness (NVH), is important to consumers and automotive manufacturers because it affects the physical and mental health of passengers and impacts the reputations of automotive manufacturers. Although simulations can predict vehicle interior noise and vibration, acoustic treatment and resulting uncertainty may generate considerable errors between the simulation and actual measurement results. Meanwhile, the recursive invocation of a complex simulation model in an optimization leads to a very time-consuming calculation process. These problems make the optimization of vehicle interior vibro-acoustical comfort a challenge for the automotive industry. In this paper, a new method, the multiobjective interval analysis method, was proposed to solve these problems. To implement this method, we carried out a vehicle road test instead of a complicated simulation model to collect data. The vehicle interior noise and vibration were recorded, and the interior sound pressure level (SPL) and total weighted acceleration root mean square (TWA-RMS) values were calculated and used as objectives. Then, a multiobjective interval analysis method was proposed to compute the midpoint and radius of the objective functions under different uncertainties. The preferred average performance and robustness of the vehicle vibro-acoustical comfort was adjusted by tuning the weighting factor. To accelerate the calculation process, a neural network model was developed as an approximate model to construct the numerical relation between the dynamic parameters and the vibro-acoustic metrics. Finally, the effectiveness and accuracy of the proposed method were verified with experimental results.