A novel active noise control using neural networks without the secondary path identification

In this paper, a novel active noise control (ANC) scheme based on neural networks is presented for nonlinear ANC systems without the identification of secondary path by introducing virtual primary noises. The ANC system is analyzed in the form of discrete-time state equations. The proposed controller employs neural networks to attenuate the noises. The proposed scheme does not require the dynamical knowledge of the primary and secondary path model compared to classical ANC approaches. The stability of the proposed scheme is analyzed by the Lyapunov theory. Simulation results show that the proposed strategy performs well for attenuating the noises.