A Neural Network Assisted FuLMS Algorithm for Active Noise Control System

In active noise control (ANC) systems, the Filtered-u Least Mean Square (FuLMS) algorithm has better control performance and faster rate of convergence than Filtered-x Least Mean Square (FxLMS) algorithm. However, due to the instability of adaptive infinite impulse response (IIR) filters, the application of FuLMS algorithm is not as extensive as that of FxLMS algorithm using adaptive finite impulse response (FIR) filters. The Equation Error (EE) method for adaptive IIR filtering can solve stability issues caused by poles, so in this paper we use the EE-based adaptive IIR filter to improve the FuLMS algorithm. Moreover, in this work, we introduce a neural network assisted method for designing adaptive IIR filters coefficients for the use in ANC systems. The results, compared with the FuLMS algorithm and neural network only approach, demonstrate the effectiveness of this scheme.