Active sound-profiling for automobiles
Active sound-profiling for automobiles
Taking the engine air inlet as the primary source and a loudspeaker mounted within the inlet as the secondary source, active control simulations were conducted using a distributed point source array model. It was found that the optimised acoustic configuration has a quadrupole like radiation efficiency, and that the attenuation achievable is not affected when the primary and secondary sources in the simulation were reversed, which led to a new development in the theory of reciprocity. The total sound power output, however, is not the same for the two configurations, with the original configuration radiating significantly less power.
The thesis also proposes 3 novel adaptive algorithms that are based on the FXLMS algorithm, but have been modified such that the output can be set to a predetermined level. These algorithms are compared analytically, by simulation and experimentally against current active sound-profiling algorithms. It was found that the phase scheduled command-FXLMS and the closely related automatic phase command-FXLMS algorithms out-performed existing algorithms in the areas of stability and control effort.
When controlling multiple tones, as required in the active sound-profiling of engine noise, the cost, which is critical in automotive applications, is partly determined by the power of the DSP hardware required for real-time operation. It is proposed that a psychoacoustic model, based on that used in MPEG encoding, can be used to reduce the number of engine orders that require control to reach the target engine spectrum, and thus reduce the DSP processing, without a loss in sound quality. As an example, the psychoacoustic model was used to create new audio files of engine noise with only a small loss in sound quality, but reducing the number of controlled engine orders needed by 50%.
University of Southampton
Rees, Lewis E
9ffc57fc-5b5f-4220-aaa8-d2ab3c797091
2005
Rees, Lewis E
9ffc57fc-5b5f-4220-aaa8-d2ab3c797091
Rees, Lewis E
(2005)
Active sound-profiling for automobiles.
University of Southampton, Doctoral Thesis.
Record type:
Thesis
(Doctoral)
Abstract
Taking the engine air inlet as the primary source and a loudspeaker mounted within the inlet as the secondary source, active control simulations were conducted using a distributed point source array model. It was found that the optimised acoustic configuration has a quadrupole like radiation efficiency, and that the attenuation achievable is not affected when the primary and secondary sources in the simulation were reversed, which led to a new development in the theory of reciprocity. The total sound power output, however, is not the same for the two configurations, with the original configuration radiating significantly less power.
The thesis also proposes 3 novel adaptive algorithms that are based on the FXLMS algorithm, but have been modified such that the output can be set to a predetermined level. These algorithms are compared analytically, by simulation and experimentally against current active sound-profiling algorithms. It was found that the phase scheduled command-FXLMS and the closely related automatic phase command-FXLMS algorithms out-performed existing algorithms in the areas of stability and control effort.
When controlling multiple tones, as required in the active sound-profiling of engine noise, the cost, which is critical in automotive applications, is partly determined by the power of the DSP hardware required for real-time operation. It is proposed that a psychoacoustic model, based on that used in MPEG encoding, can be used to reduce the number of engine orders that require control to reach the target engine spectrum, and thus reduce the DSP processing, without a loss in sound quality. As an example, the psychoacoustic model was used to create new audio files of engine noise with only a small loss in sound quality, but reducing the number of controlled engine orders needed by 50%.
Text
968054.pdf
- Version of Record
More information
Published date: 2005
Identifiers
Local EPrints ID: 465504
URI: http://eprints.soton.ac.uk/id/eprint/465504
PURE UUID: a7a22d53-4af7-477e-abc1-f9d2293dcd97
Catalogue record
Date deposited: 05 Jul 2022 01:29
Last modified: 16 Mar 2024 20:13
Export record
Contributors
Author:
Lewis E Rees
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics