Selective preconditioning of an active sound control system with virtual sensing

Abstract

The convergence speed of multichannel adaptive active noise control algorithms can be improved if the reference signals are preconditioned, so it is advantageous to whiten them and reduce their cross-correlations. However, this preconditioning depends on the properties of the reference signals, which will generally change over time. It may still be possible to approximate the properties of the measured reference signals using a limited number of exemplar cases. By classifying the measured reference signals into the closest of these cases, the most appropriate preconditioning can be selected for the adaptive algorithm at any one particular time. Also, one method of implementing virtual sensors is to use the additional filter method, which provides targets for the measured error signals to follow, generated from the reference signals. The performance of this virtual sensing is known to be degraded if the properties of the reference signals change, so this classification method could also be used to select the most appropriate additional filter.