Active Noise Control in Ducts: Feedforward/Feedback Design by Blending H∞ and QFT Methods

Abstract

In this paper we take a closer look at the prevalent notion that adaptive filters are preferred to fixed-filters in cancelling noise in ducts. For this purpose, we consider an active noise control (ANC) system with configuration similar to that used in most adaptive implementations; i.e., a duct equipped with both feedforward and feedback sensors and a control speaker. Our design of a linear, time-invariant controller involves three steps: (1) system identification from the experimental duet frequency response, (2) H∞ synthesis using the identified plant and model errors, and (3) use of Quantitative Feedback Theory (QFT) to further improve performance by sequential tuning of the feedback and feedforward filters by using the experimental data. Preliminary experiments indicate that attenuation levels of at least 10-20 dB are achievable over a 400 Hz band. An important consequence of LTI designs is that the performance is guaranteed for wide-band noise as well as pure tone disturbances and, unlike adaptive schemes, stability and performance can be guaranteed beforehand.