A New Hybrid Active Noise Control System with Convex Combination of Time and Frequency Domain Filtered-X LMS Algorithms

摘要

A conventional single-channel feedforward active noise control (ANC) system encounters noise from multiple sources. In many cases, there is a reference signal that is independently available which is correlated with the ambient noise. However, in some situations, a portion of the noise generated is uncorrelated with the reference signal. In this paper, a new hybrid ANC (HANC) system is proposed using a convex combination of time and frequency domain Filtered-X LMS (FXLMS) algorithms. The feedforward part uses a time domain FXLMS (TDFXLMS) algorithm to control disturbances that are correlated with the reference noise and the feedback part uses a multiresolution analysis-based frequency domain wavelet packet FXLMS (WPFXLMS) algorithm to control the uncorrelated disturbances encountered during the operation of an ANC system. An extra adaptive filter is used to smoothen the error signal in the HANC system. An effort is made to exploit the benefits of the structural design and time-frequency domain signal processing techniques for active control of disturbances that are both correlated and uncorrelated with the reference noise. As a result of the convex combination of the TDFXLMS and WPFXLMS algorithms in a HANC system, the proposed method is successful in canceling the disturbances encountered by the ANC system from multiple noise sources.