Active noise control : Adaptive vs. robust approach

Abstract

Active noise control is often concerned with the strong attenuation of single or multiple tonal noise disturbances which may have unknown and time varying frequencies. Currently in applications, adaptive feed-forward compensation is used which requires the use of an additional transducer and introduces an instability risk due to a positive internal coupling. However for these types of noise a feedback approach can be efficiently used and this will be illustrated in this paper. One considers the case of two tonal disturbances located in two distinct frequency regions subject to frequency variations within a given range as well the case of interferences between tonal disturbances of very close frequencies. The objective is to minimize the measured residual noise in a predefined location. These problems occurs often in ventilation systems (active silencers). To solve these problems robust and adaptive solutions are considered. A robust controller design using sensitivity function shaping is considered. The maximum achievable attenuation is inverse proportional to the range of frequency variations of the tonal disturbances. To further improve the performance an add-on direct adaptive feedback approach using the Internal Model Principle and the Youla Kucera parametrization is considered. The adaptive approach allows both to improve the performance within the given frequency ranges as well as to extend the admissible domain of frequencies variations. Experimental results obtained on a relevant test bench will illustrate the performance of the two designs.