| Summary: | The scope of this thesis is to investigate methods of recording, processing and analysing sound data from wheel brake testing in dynamometers with focus on detecting and measuring squeal. The desired outcome is a method that Scania can use to record and analyse brake sound. A literature study was made to find relevant methodologies and tools proposed in papers, books and industry standards. These methods were tried and evaluated by recording and analysing real sound data and other signals from one of Scanias dynamometers. The resulting method includes directions on what hardware to use, how to set it up and an algorithm that computes a spectral limit based on normal sound data. This limit is then used as reference when evaluating other recordings. To increase signal to noise ratio, an adaptive filter is proposed to attenuate background noise in the recordings, in particular from the dynamometer and ventilation system. The conclusion is that it is possible to find squeal using spectral limits based on normal data. The performance of the algorithm is a compromise between being very effective but rather complex, or slightly less effective but also less complex. Its performance is also highly dependent on how squeal is defined. A very narrow definition will only find certain types of squeal while a more broad definition will find more squeal, but also potentially mislabel some recordings.
|
|---|
