Robustness of adaptive active noise control systems.

Abstract

In recent years, adaptive control techniques have been increasingly employed in active acoustic noise control systems. The aim in using an adaptive methodology is to produce a noise control system that is entirely self-calibrating and that can maintain optimal attenuation despite changes in the acoustic environment. However, a cost of the increased sophistication of adaptive controllers as compared to fixed control methods is the introduction of some unexpected performance limitations. In particular, active noise control systems based on adaptive methods can suffer from instability, long-term drift and suboptimal steady-state performance. In a recent paper [M. Johnson and G. Dodd, Proc. Internoise 91, Sydney, 1137–1140 (1991)], some of the inherent instability mechanisms in an adaptive active noise control system were identified. The proposed paper will provide both theoretical support for these results and confirmation by way of simulation and real-time examples. It is possible to improve the controller performance without greatly increasing the computational burden by making some straightforward modifications to the adaptation algorithm. Results obtained using the modified algorithm demonstrate improved tolerance to disturbances and changing conditions in the acoustic environment.