Home > Research > Publications & Outputs > Design of a MIMO neuro-controller for ANC syste...

Associated organisational unit

View graph of relations

Design of a MIMO neuro-controller for ANC system with loudspeaker nonlinearity in enclosure

Research output: Contribution in Book/Report/ProceedingsConference contribution

Published
Publication date2007
Host publicationSignal Processing and Communications, 2007. ICSPC 2007. IEEE International Conference on
Place of PublicationNew York
PublisherIEEE
Pages1415-1418
Number of pages4
ISBN (Electronic)978-1-4244-1236-5
ISBN (Print)978-1-4244-1235-8
<mark>Original language</mark>English
EventIEEE International Conference on Signal Processing and Communications - Dubai

Conference

ConferenceIEEE International Conference on Signal Processing and Communications
CityDubai
Period24/11/0727/11/07

Conference

ConferenceIEEE International Conference on Signal Processing and Communications
CityDubai
Period24/11/0727/11/07

Abstract

In this paper optimization of a nonlinear adaptive multi-channel active noise control system in a rectangular enclosure using neural networks is investigated. The model of enclosure is obtained using modal analysis, and the bandwidth of the control system for global reduction of noise is selected between 50-300Hz. Secondary path in modeled offline using MLP and Levenberg-Marquardt algorithm with sum of sinusoid excitation signal. The simulation results for multi-channel case assuming nonlinear loudspeaker in the secondary path, show the performance of feedforward multilayer perceptron neural networks with FxBP algorithm is acceptable in attenuation of noise and its harmonics in the error microphones.