A novel acoustic feedback compensation filter for nonlinear active noise control system. (September 2021)

• Record Type: Journal Article
• Authors: Luo, Lei; Zhu, Wenzhao; Xie, Antai
• Is Part Of: Mechanical systems and signal processing. Volume 158(2021)
• Publication Date: 2021-09
• Subjects: Nonlinear active noise control -- Acoustic feedback -- VREMFN filter -- Stability characteristics -- Leaky NLMS; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2021.107675
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A Volterra recursive even mirror Fourier nonlinear (VREMFN) filter is proposed for NANC system to compensate for acoustic feedback. The stability condition of NANC system using VREMFN is derived. A new leaky normalized LMS update strategy is proposed to accelerate convergence. The effectiveness of VREMFN filter is verified by simulated and experimental studies. Abstract: The acoustic feedback is widely seen in the active noise control (ANC) system for short acoustic duct such as the intake system of an automobile engine, which severely destabilizes the stability of the original ANC system. The conventional acoustic feedback compensation methods based on IIR filter can only be applied to the linear acoustic path and is not suitable for a nonlinear ANC (NANC) system. To solve this problem, this paper proposes a Volterra recursive even mirror Fourier nonlinear (VREMFN) filter, which can effectively suppress both linear acoustic feedback and nonlinear acoustic feedback in the NANC system. In this paper, we show the solutions for the linear acoustic feedback and nonlinear acoustic feedback of VREMFN and discuss its stability condition. Moreover, a new leaky normalized LMS algorithm is also developed to further improve the performance of NANC system based on VREMFN. Simulated results comparing IIR, Volterra and VREMFN based on parameter adaptive algorithm (PAA), FXLMS and FULMS show that the proposed VREMFN dramatically improves the noise reduction performance of NANC system both in linear acoustic feedback path and nonlinear acoustic feedback path. Furthermore, two experiments with different acoustic feedback paths also demonstrate the effectiveness of the proposed VREMFN.