Analysis and DSP implementation of an ANC system using a filtered-error neural network

In this paper, feedforward active noise control (ANC) using a neural network (NN) based on filtered-error back-propagation (BP) algorithm is considered. The filtered-error BP NN (FEBPNN) algorithm is first derived, and the difference between the FEBPNN algorithm and the filtered-X BP NN (FXBPNN) algorithm is given to show that the FEBPNN algorithm offers computational advantage over the FXBPNN algorithm. Computer simulations are carried out to compare the FEBPNN algorithm with the filtered-X least mean square (FXLMS) algorithm and the FXBPNN algorithm. The controllers based on the FEBPNN algorithm and the FXLMS algorithm are implemented on a Texas Instruments digital signal processor (DSP) TMS320VC33. The simulations and the experimental verification tests show that the FEBPNN algorithm performs as well as the FXLMS algorithm for a linear control problem, and better for a nonlinear control problem, at the same time, the simulations and the experimental verification tests also show that the convergence rate of the FEBPNN is acceptable, and the FEBPNN has better tracking ability under changes of the primal signal, the primary path and the secondary path. The experiments also lead to the conclusion that more work is required to improve the predictability and consistency of the performance of the NN controller based on the FEBPNN algorithm.