New neural network algorithms for nonlinear active noise cancellation with nonlinear secondary path

In this paper, a feedforward nonlinear active noise cancellation (NANC) system employing a neural network (NN) based filtered-X least mean square algorithm is developed. NN-NANC with linear secondary path (LSP) and nonlinear secondary path (NSP) are unified using virtual secondary path concept. The developed NN-NANC system is also extended to incorporate filtered-E based algorithm which results in reduced computational algorithm. Performance of the proposed algorithms is validated through computer simulations.