New robust adaptive algorithm for multichannel adaptive active noise control

Abstract

The adaptive active noise control problem of a multichannel system is studied. The general structure of the multichannel active noise control system is clarified and the perfect cancellation condition by feedforward control is analyzed. A new LMS type of robust multichannel adaptive algorithm for updating the feedforward controllers is proposed which can assure the stability of the adaptation and keep each cancelling error within a tight bound by efficiently using prior information of each upper bound of the uncertainty terms. The algorithm is scalar normalized and does not need to calculate the inverse matrix. A fully adaptive scheme for dealing with the case when all the channel dynamics are unknown and changeable is also presented. The proposed robust algorithm is examined and validated in the experiment of noise attenuation for an air duct system.