Adaptive parallel filter method for active cancellation of road noise inside vehicles

Abstract

Active noise cancellation (ANC) is an efficient and effective method for suppressing low frequency disturbances generated by tyre–road interaction in a vehicle. There are significant differences in road noise inside the vehicle under different driving conditions. However, fixed-filter ANC strategies are unable to achieve superior noise attenuation for different types of noise, and conventional adaptive algorithms require prolonged adaptation which is detrimental to dynamic noise reduction. Thus, this study proposes an adaptive parallel filter method (APFM) that quickly responds to and suppresses varying vehicle interior road noises. The APFM is divided into two phases: the tuning phase and control phase. In the tuning phase, a set of potential noises is adopted to train the corresponding optimal fixed filters. During the control phase, the most appropriate parallel control filter is first selected for each frame based on a selection mechanism to minimise the energy of the estimated error signal. Then, a delayless variable step size normalised frequency-domain block adaptive filter algorithm is adopted to update the selected control filter. In addition, the computational complexity of the proposed APFM is analysed. Numerous simulations and real time in-vehicle experiments conducted using multi-channel ANC headrests have confirmed the feasibility and effectiveness of the proposed algorithm. It is evident that the APFM achieves fast noise reduction response and low steady-state error in attenuating varying in-vehicle road noise, and retains superior performance in the presence of non-pre-trained disturbances.