A generalized exponential functional link artificial neural networks filter with channel-reduced diagonal structure for nonlinear active noise control

The nonlinear adaptive exponential functional link artificial neural networks (E-FLANN) filter has been introduced to improve the noise reduction capability of the functional link artificial neural networks (FLANN) in nonlinear active noise control (NANC) system. It, however, suffers from a heavy computational burden at the nonlinear secondary path (NSP) and poor convergence performance in strong nonlinearity systems. To surmount these shortcomings, a computationally efficient generalized E-FLANN filter with the channel-reduced diagonal structure (GE-FLANN-CRD) for NANC system is developed in this paper. Based on introducing the suitable cross-terms and adaptive exponential factor into the trigonometric functional expansions, the nonlinear processing capability of the filter is enhanced in NANC. Also, by applying the filtered-error least mean square (FELMS) algorithm to the GE-FLANN-CRD, it substantially decreases the computational cost to update the exponential factor. Computer simulations demonstrate that the proposed filter-based the NANC system performs better than the FLANN, E-FLANN and Generalized FLANN (GFLANN) filters-based NANC system in the presence of strong nonlinearity.