A hybrid feedforward/feedback multi-channel active control system with optimization for cancelling road noise inside a vehicle cabin

Abstract

A hybrid feedforward/feedback multi-channel system was developed for active road noise control (ARNC) inside a vehicle cabin. First, a centralized feedforward subsystem was present with a multi-channel normalized weighted error FxLMS algorithm, including a simplified reference signal normalization method, a reconstructed filter bank for filtering the reference signals, and a newly defined cost function of A-weighting error signals. Second, a distributed feedback subsystem was presented with multiple feedback single-channel simplified normalized FxLMS algorithms. By combining the two subsystems, a hybrid ARNC system was developed. Furthermore, data related to the ARNC system were collected from the test vehicle. Next, a nondominated sorting genetic algorithm-II was introduced to optimize the system parameter. Multi-objective optimization models of the feedforward and hybrid ARNC systems were established respectively, and the optimal Pareto solution sets for their parameters were obtained. Real-time experiments show that, when the test vehicle driving at 60 km/h on the small brick pavement, the developed hybrid ARNC system can achieve an overall noise reduction of 5.87 dBA in the time domain, and a peak noise reduction of 7.43 dBA in the frequency-domain. Compared with the feedforward ARNC system, road noise reduction is much improved. It still has a good noise reduction effect at other speeds and on different roads.