Active Noise Control Based on Reinforcement Learn-ing

Abstract

Active Noise Control (ANC) systems are used in order to reduce the sound noise level by generat-ing anti-noise signals. M-Estimators are widely used in ANC systems in purpose of updating the adaptive FIR filter taps used as systems controller. Up to now evaluation of M-Estimators capabili-ties show that there exists a need for further improvements. In this paper, Reinforcement Learning (RL) methods are used to generate the controller output. The sensitivity of the constant parameter in RL method is checked. The effectiveness of proposed method is proven by comparing the re-sults with the previous studies. Simulations show the fast initial convergence of the proposed algo-rithm.