Abstract
Keywords: Squeaks and Rattles, Separation, Detection, Cluster Analysis, NVH
Traditionally, annoying squeak and rattle events (S&R) in an in-car environment are detected and analysed by well trained employees. Such subjective ratings are used while driving a car as well as in a car on a test stand (hydropulse, shaker). However, it is difficult to document annoying patterns consistently and numerically by using arbitrary subjective evaluation. The qualitative and quantitative effectiveness of countermeasures are as well, difficult to prove. The reproducibility of S&R provides a further problem, since road excitation leads to a random and sporadic S&R occurrence.
By means of an increasing amplitude swept sine input via a shaker and automatic airborne sound analysis, it is feasible to track down S&R sounds of different material contact points. Furthermore the resonance frequency of each squeak and rattle event is captured and all events are prioritized according to their unpleasantness index and probability of occurrence. The order of ranking corresponds to the appearance at different excitation levels. Thus an annoying noise, which already occurs at low excitation levels, gets a higher priority index than a noise, which arises for the first time at high excitation levels.
This paper addresses the signal processing of airborne sound and the separation of squeak and rattle events that occur. In contrast to road excitation, swept sine excitation leads to S&R occurrence according to its resonance. However, the resonance ranges of many different squeaks and rattles events overlap each other. This causes an additive combination of different S&R root sounds. Overlapping noise impulses of varying origin (material contact points) arise almost simultaneously - in accordance with the excitation frequency. The intention is to split up the total airborne noise into its constituent S&R events. So the number, character and resonance of the different S&R sources can be found. Squeaks and rattles with two different resonance frequencies cause a further challenge. A contact point with two resonant vibrating material contact pairs has one and the same sound at different excitation frequencies. The signal processing of airborne sound detects such double noises as a single S&R event and returns its two resonances.