Active sound profiling of engine noise inside a vehicle based on the HPSC-FxLMS algorithm

Active sound profiling (ASP) can mitigate the negative impacts of excessive interior noise and satisfy people’s pursuit of acoustical comfort. However, in ASP of engine noise, the existing algorithms are susceptible to the modeling errors of the secondary paths and suffer from insufficient accuracy due to frequency fluctuation. This paper introduces a phase-scheduled command filtered-x least mean square algorithm based on the Hilbert transform (HPSC-FxLMS) to address the above issues. The proposed algorithm modifies the internal model structure of the PSC-FxLMS algorithm by integrating an autonomous PSC system. This avoids the impact of secondary paths modeling error on the stability of the ASP system. Moreover, the HPSC-FxLMS algorithm identifies the desired orders within primary noise by bandpass filters, and align the command signal phases to the desired order phases more accurately. This synchronization enables the secondary source to achieve the predefined objective with less control effort. Numerical simulations demonstrate that the HPSC-FxLMS algorithm is robust to the frequency fluctuation of engine noise because of the precise phase scheduling of the command signal phases. Further, the multi-channel on-board experiments are conducted. Compared to the conventional Command-FxLMS algorithm, when the vehicle is driven in the semi-anechoic chamber at 1700 rpm, the 2nd order secondary output of the HPSC-FxLMS is reduced by over 50 %. Whereas when the vehicle is driven on roads at 2550 rpm, the 2nd order and 6th order secondary outputs of the HPSC-FxLMS are reduced by 40 %. The results verify the effectiveness and applicability of the HPSC-FxLMS algorithm.