Research on fast optimal reference sensor placement in active road noise control

Abstract

The noise reduction performance of the active road noise control (ARNC) system highly depends on the location of the reference sensors. Generally, the optimal sensor locations are evaluated by calculating the multiple coherence function (MCF) between all possible reference signal combinations with road noise. However, this trial-and- error method becomes time-consuming when the number of candidate locations is large. The transfer path analysis method can select the optimal sensor locations quickly while with low accuracy. Therefore, this article proposes two fast optimal sensor placement (FOSP) methods, namely, Wiener filter (WF)-FOSPand the MCF-FOSP, respectively. In both methods, the sensors are iteratively extended to the desired number, and each added sensor maximizes the predicted noise reduction of this iteration loop. Numerous ARNC simulations based on measured signals are conducted to illustrate the performance of the proposed two methods in terms of efficiency and accuracy. The results demonstrate that the WF-FOSP method provides the best comprehensive performance. The data analysis for one operating condition takes three minutes, and the absolute error is within 5% with respect to the benchmark. In addition, two schemes are discussed to obtain a set of sensor locations compatible with the noise reduction requirement of different operating conditions. The sensor locations can achieve a maximum average noise reduction of 7.29 dB(A) under four operating conditions.