A On-Line Secondary Path Modeling Frequency Domain Active Noise Control Algorithm Based on the Kalman Filter

摘要

A new frequency domain ANC algorithm based on the Kalman filter is presented. This algorithm has on-line secondary path modeling and is designed to minimize the effects of large secondary path changes on the acoustic noise level. It is based on the Overall Modeling Algorithm with a careful controller and estimation of the secondary path changes. Due to its frequency domain implementation the system will not work well as a predictor but is suitable for feed-forward broadband active noise control applications. The Kalman filter estimates the primary and secondary paths. The estimates and their variances are used to determine the optimal controller. Because the controller is only based on secondary and primary path models, it is much more robust to secondary path changes than other approaches based on the FX-LMS, allowing for faster convergence.