Variable Step-Size Hybrid Filtered-x Affine Projection Generalized Correntropy Algorithm for Active Noise Control

Active Noise Control (ANC) is frequently utilized to minimize noise in industrial environments. However, the powerful pulses in industrial noise pose challenges to its application. Consequently, ANC systems necessitate a high-performance algorithm as a core component. In this process, the variable step-size strategy is the main approach for enhancing the ANC algorithm's performance but ensuring robustness while improving performance remains a challenge. To address this problem, we propose a new ANC algorithm with a variable step size. This algorithm is derived from the Affine Projection Generalized Maximum Correntropy (APGMC) method, featuring a hybrid step-size and a new step-size approach achieved by modifying the mean square deviation (MSD). To showcase the practical effectiveness of the proposed algorithm, noisy audio from a real construction site was used for noise reduction control. Results show that the proposed algorithm effectively manages noise across frequency bands, with an improvement of approximately 16% to 19.2% compared to existing similar algorithms.