A generalized collaborative functional link adaptive filter for nonlinear active noise control

Abstract

This paper derives a collaborative control algorithm called generalized collaborative functional link adaptive filter (GCFLAF) for nonlinear active noise control (NANC). The algorithm adopts the collaborative scheme by handling the noise cancellation problem divisionally and parallelly, and working coordinately. It takes the characteristics of different nonlinear components into consideration and handles them separately. And a hierarchical adaption law is adopted in which linear filter is always prior to nonlinear filters no matter whether nonlinear distortions arise from the primary path as required by the filtering scenario. Thus the proposed scheme can tailor to different nonlinear characters arising from the primary path and can hierarchically adapt to the influences of the distinct nonlinearities. Particularly, since the zero-memory nonlinearity and memory nonlinearity are often unavoidable in NANC, it employs two shrinkage parameters to regulate the contribution of the different degrees of memory and zero-memory nonlinearity as required by the filtering scenario, aiming to achieve a better convergence performance and robustness to nonlinear distortions. Some numerical simulation results in the context of random noises as well as the real noise signals verify the improved convergence behavior of the presented architecture in NANC scenarios.