Secondary path modeling for narrowband active noise control systems

Abstract

This work examines the performance of both offline and online secondary path modeling algorithms that are used in narrowband active noise control (NANC) systems. Theoretical analysis reveals that disturbances that are sensed by error sensors reduce the convergence rate and accuracy of adaptive system identification. In a parallel-structure narrowband active noise control system, a filterbank is applied to partition the full-band excitation and error signals to utilize an independent and lower-order secondary-path modeling filter for every channel. This method increases convergence speed and modeling accuracy. The results of an analysis and improved performance are confirmed by simulations in which measured transfer functions are used.