GFANC-RL: Reinforcement Learning-based Generative Fixed-filter Active Noise Control

• Published in: Neural networks Vol. 180; p. 106687
• Main Authors: Luo, Zhengding, Ma, Haozhe, Shi, Dongyuan, Gan, Woon-Seng
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.12.2024
• Subjects: Active Noise Control; Algorithms; Computer Simulation; Convolutional Neural Network; Generative Fixed-filter ANC; Humans; Machine Learning; Neural Networks, Computer; Noise; Reinforcement Learning; Reinforcement, Psychology; Reinforcement Learning; Convolutional Neural Network; Generative Fixed-filter ANC; Active Noise Control
• ISSN: 0893-6080; 1879-2782; 1879-2782
• DOI: 10.1016/j.neunet.2024.106687

The recent Generative Fixed-filter Active Noise Control (GFANC) method achieves a good trade-off between noise reduction performance and system stability. However, labelling noise data for training the Convolutional Neural Network (CNN) in GFANC is typically resource-consuming. Even worse, labelling errors will degrade the CNN’s filter-generation accuracy. Therefore, this paper proposes a novel Reinforcement Learning-based GFANC (GFANC-RL) approach that omits the labelling process by leveraging the exploring property of Reinforcement Learning (RL). The CNN’s parameters are automatically updated through the interaction between the RL agent and the environment. Moreover, the RL algorithm solves the non-differentiability issue caused by using binary combination weights in GFANC. Simulation results demonstrate the effectiveness and transferability of the GFANC-RL method in handling real-recorded noises across different acoustic paths.22The code will be accessible at https://github.com/Luo-Zhengding/GFANC-RL. •GFANC-RL uses RL to address challenges in GFANC and improve its performance.•We model GFANC as a Markov Decision Process, enabling the use of RL techniques.•RL algorithm trains CNN with unlabelled data to enhance exploration capabilities.•GFANC-RL attenuates real noises and exhibits good robustness and transferability.