A filtered-X RLS based narrowband active noise control system in the presence of frequency mismatch

Rotating machines, such as fans, engines, large-scale cutting machines etc., generate noise signals of a sinusoidal nature. Narrowband active noise control (ANC) systems are effective in suppressing these noise signals. However, when the reference timing signal sensor and the cosine wave generator contain errors, the frequencies of the reference sinusoids fed to each ANC channel are different from the primary noise signal frequencies. This difference is referred to as frequency mismatch (FM). We first introduce a fast filtered-X RLS (FXRLS) algorithm to replace the filtered-X LMS (FXLMS) algorithm in the conventional parallel form narrowband ANC system, and demonstrate its improved performance without FM. Second, through extensive simulations, it is revealed that this FXRLS-based ANC system may suffer from significant performance degradation in the presence of FM. Third, the FXRLS-based system is modified to mitigate the influence of the FM by incorporating second-order AR model based adaptive blocks that are updated by an LMS or RLS algorithm. Simulations are provided to demonstrate that the modified system is robust to the existence of FM and superior to a recently developed FXLMS-based ANC system in terms of capabilities of suppressing the undesirable FM.