Enhanced multiple random Fourier features based generalized maximum correntropy algorithm for active noise control

The random Fourier filters-based FxLMS (RFFxLMS) algorithm significantly enhances traditional kernel methods in active noise control (ANC) by effectively reducing memory demands and computational complexity. Additionally, it overcomes the linearity constraints of conventional FxLMS through its nonlinear adaptive capability, enabling superior noise suppression in complex acoustic environments. Therefore, this paper proposes an enhanced multiple random Fourier features based generalized maximum correntropy criterion (EMRFF-GMCC) algorithm. Unlike RFFxLMS and its extensions, the EMRFF-GMCC algorithm incorporates a projection matrix to develop an EMRFF method, improving nonlinear mapping, adaptability, and noise reduction performance in complex scenarios. By leveraging the shared properties of these features, the algorithm achieves a significant improvement in overall performance, making it more effective in handling dynamic and challenging noise environments. Furthermore, the mean square error (MSE) cost function is replaced by the generalized maximum correntropy criterion (GMCC), significantly enhancing robustness against outliers. Simulations and duct experiments validate the effectiveness of the proposed EMRFF-GMCC algorithm, showcasing its superior performance across diverse noise environments and solidifying its potential as a promising candidate for real-world ANC applications.