ADAPTIVE ACTIVE ATTENUATION OF NOISE USING MULTIPLE MODEL APPROACHES

The adaptive active noise control problem is considered, in which the response of the system under control varies with time. A new algorithm for system identification in such circumstances, using multiple model s of the secondary path transfer function, is presented. The computati onal burden of the proposed algorithm can be much smaller than the exi sting methods, particularly in the multi-channel case. An infinite imp ulse response structure is assumed for the control filter, and the rec ursive least mean square algorithm introduced by Feintuch is used to a djust the filter coefficients. By combining the proposed identificatio n technique and the filtered-x recursive least mean square algorithm, the multiple model adaptive controller for active noise control proble ms is presented. This method requires a priori information about the p lant to select models of the secondary path transfer function, so it c an be used when we have some information on the plant and operating si tuations. Computer simulations were performed to show the effectivenes s of the proposed algorithm in a simple duct problem. (C) 1995 Academi c Press Limited