With the removal of the combustion engine as a masking component, the noise of the electric powertrain and the tire and wind noise get more into focus. In particular, the high-frequency tonal components of the powertrain have a negative impact on the perceived pleasantness of the driver. Therefore, it is essential to comprehensively analyze the components of the electric vehicle interior noise. The aim of this research work is to develop suitable and efficient methods for automated noise separation and allocation to the emitting components in order to reduce measurement and analysis efforts. In doing so, the amount of component-specific input data should be kept as low as possible. For this purpose, this work presents the developed methods for an efficient acoustic analysis as a basis for an optimization of the vehicle interior noise. A targeted detection of tonal noise components of the electric motor, gearbox, inverter and tires is based on binary images of spectrograms and order spectrograms using the Hough transformation, followed by a separation using the Vold-Kalman filter. The detection accuracy is in the range of 87 % to 94 % depending on the component. For the orders of the inverter and the tire cavity, the component assignment is unambiguous due to the geometric appearance in the image area. In the case of the electric motor and gearbox orders, these are assigned to the respective emitting component using a machine learning process with preceding parameter selection with a classification accuracy of 86 %. In addition, to further reduce the measurement effort, methods are developed to determine the electric machine speed and, for vehicles with asynchronous machines, the slip from an acoustic signal. Finally, the residual noise remaining after separation of all tonal noise components of the powertrain, can be approximately separated into a tire and a wind share. All methods developed in this work are physically reasonable and not vehicle specific and can therefore be assumed to be robust. Since the methods are mainly based on mathematical relations and only a few empirical parameters and statistical processes are integrated, an applicability for future measurements and measuring vehicles is expected. Furthermore, it has to be emphasized that the developed methods are set up without any input of component-specific parameters.
Automatisierte Trennung und Zuordnung von Geräuschanteilen im Innenraum von Elektrofahrzeugen
Automated separation and allocation of the interior noise of electric vehicles
2023-01-01
1 Online-Ressource : Illustrationen, Diagramme pages
Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2023; Aachen : RWTH Aachen University 1 Online-Ressource : Illustrationen, Diagramme (2023). doi:10.18154/RWTH-2023-08448 = Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2023
Miscellaneous
Electronic Resource
German
Automatisierte Trennung und Zuordnung von Geräuschanteilen im Innenraum von Elektrofahrzeugen
TIBKAT | 2023
|Zuordnung einer Ladeeinheit zu einer Mehrzahl von Elektrofahrzeugen in einem Netzwerk
European Patent Office | 2018
|Klare Trennung - digitale Isolation bei Hybridfahrzeugen und Elektrofahrzeugen
Automotive engineering | 2010
|