engleski

Active Noise Control with Adaptive IIR Filter Based on Discrete Legendre Functions

There is often problem of acoustical feedback from secondary source to the reference sensor in feedforward active noise control systems. Presence of such a feedback substantially influences on performance of active control, so countermeasures have to be applied, specially when dealing with uniformly distributed random noise signals. Considering feedback as a part of the overall plant, infinite impulse response (IIR) adaptive filters have ability of modeling its transfer function with poles and zeros. Adaptive IIR filters are not unconditionally stable, and their performance surface is generally nonquadratic, so they can converge to a local minimum. Because of its simplicity, filtered-u recursive least mean square (FURLMS) algorithm has been widely applied, although its convergence has never been proved formally, and doubts were raised on its stability. In this paper we propose a filtered-x algorithm for use in active noise control based on Perez-Tsujii’s Legendre adaptive filter (LADF) derived from Legendre orthogonal functions. This IIR filter has unimodal performance surface and it is stable. Its structure is simple and modular so it is easy to implement even when large filter order is required. Proposed filter is compared with FURLMS algorithm, and computer simulations were performed.

active noise control; adaptive recursive filters; discrete Legendre functions

nije evidentirano