Nonlinear Active Noise Control Using Adaptive Wavelet Filters

This paper deals with nonlinear active noise control using adaptive wavelet filters. The ability of wavelets in signal reconstruction and function approximation make them appealing for black box system identification. Moreover, the intrinsic similarity between wavelet filters and noise/vibration signals implies that better approximation of these signals can be achieved by employing wavelet filters. Here, a new simple structure for using in active noise control system is proposed comprises a nonlinear static mapping cascaded with an IIR filter to take care of the dynamics of the system. With this strategy, one can avoid using multi-dimensional wavelet networks and thus eliminate curse of dimensionality. The performance of the proposed ANC system is examined for typical linear/nonlinear cases. The simulation results demonstrate superior performance of this method in terms of fast convergence rate and noise attenuation as well as computational complexity reduction while avoiding curse of dimensionality.