An online modeling and control algorithm for narrowband active noise control

Abstract

The narrowband FxLMS (NBFxLMS) algorithm is specifically designed for narrowband active noise control (NANC) systems and requires only the secondary path’s frequency response at the noise frequency, rather than the entire secondary path. The NBFxLMS algorithm converges only when the phase deviation of the secondary path’s frequency response is less than 90°. However, in some scenarios, variations in the secondary path may cause the phase deviation to exceed 90°, leading to algorithm divergence. To address this issue, this paper proposes an online modeling and control algorithm for NANC systems. Unlike conventional approaches that use white noise as auxiliary noise, the proposed algorithm employs two single-frequency signals near the noise frequency as auxiliary noise. It models the secondary path’s frequency response online and obtains the frequency response at the noise frequency through interpolation, which is then used for the control filter coefficient iteration. Simulation results demonstrate that the proposed algorithm can accurately model and converge even in cases of abrupt secondary path changes. Compared to existing algorithms, it exhibits faster convergence, superior noise reduction performance, lower computational complexity, and avoids the issue of inconsistent results under the same parameters encountered in conventional auxiliary white noise online modeling algorithms. Real-time experiments are carried out in a common room and the results validate the effectiveness of the proposed algorithm.