A threshold based variable step size least mean square algorithm for improved online secondary path modelling in active noise control systems

Abstract

This thesis is a study of active noise control ANC systems with newlinethe view point of adaptive signal processing The discussion is restricted to newlinethe singlechannel feed forward ANC systems A literature survey shows that newlinethe filteredx least mean squareFxLMS algorithm is the most popular newlineadaptive algorithm for ANC systems due mainly to its simplicity and robust newlineperformance However the convergence speed of the FxLMS algorithm is newlineslow Furthermore its performance is degraded when there is a large newlinemeasurement noise in the reference and error signals newlineTo solve these problems this thesis proposes an adaptive filtering newlinewith denoise based algorithm for ANC systems This algorithm uses a similar newlinestructure as that of the FxLMS algorithm based ANC systems In this thesis newlinetwo new methods have been proposed to achieve improved performance in newlineANC systems The first method proposed uses wavelet denoise concept in the newlinesecondary path newline newline