Investigation and Evaluation of Adaptive Algorithms for Multichannel Active Noise Control System

TL;DR

This paper evaluates adaptive algorithms, specifically FxLMS and pre-trained control filters, for multichannel active noise control, demonstrating FxLMS's superior performance in various real-world noise scenarios.

Contribution

It provides a comprehensive analysis and comparison of FxLMS and pre-trained control filters, highlighting their effectiveness and responsiveness in noise reduction.

Findings

FxLMS exhibits strong noise reduction and quick adaptation.

Pre-trained control filter performs better initially.

FxLMS achieves higher average noise reduction over time.

Abstract

This dissertation focuses on the investigation and evaluation of adptive algorithms for multichannel active noise control system. The aim of the research is to investigate the effectiveness of the FxLMS algorithm and the pre-trained control filter in attenuating various types of noise. The study begins with a comprehensive review of the existing literature on active noise control, highlighting the significance of noise reduction in different applications. The theoretical foundations of the FxLMS algorithm and the pre-trained control filter are then presented, including their underlying principles and mathematical formulations. Through a comprehensive analysis, a clear understanding of these methods is established. To assess the performance of the FxLMS algorithm and the pre-trained control filter, extensive simulation experiments are conducted using real-world noise signals. The experiments include scenarios such as aircraft noise, traffic noise, and mixed noise. The results of the simulations are used to compare the noise reduction capabilities of the two methods and to evaluate their effectiveness under different noise conditions. The findings indicate that the FxLMS algorithm exhibits remarkable noise reduction performance. It demonstrates a strong ability to track and respond quickly to varying noise patterns. In the initial stages of noise reduction, the pre-trained control filter shows better performance. However, as time progresses, the FxLMS algorithm consistently achieves higher average noise reduction levels compared to the pre-trained control filter. Additionally, the FxLMS algorithm shows faster responsiveness during transitional periods when noise characteristics change. These findings contribute to the field of noise control and provide valuable insights for designing efficient noise reduction systems.