Handling plant model uncertaintiesin adaptive regulation using dual Youla-Kucera parametrization. Application to Active Noise Control

International audienceAdaptive feedback regulation is concerned with the strong attenuation of unknown and time varying disturbances. This approach has proven to be very efficient for active noise and active vibration control among other domains of application. This technique uses the Internal Model Principle (IMP), the Youla – Kucera (YK) parametrization of the controller and Parameter Adaptation Algorithms (PAA). However in adaptive regulation it is supposed that the plant model of the compensatory path is known and almost constant. The present talk explore the problem of handling large plant model uncertainties which may occur in an number of application. The solution proposed uses overparametrization of the YK disturbance compensation filter. A theoretical analysis of this approach is done establishing the framework for its practical use. The basic ingredients for the analysis of this approach are the Dual Youla-Kucera parametrization, the small gain theorem and the use of parameter adaptation algorithms with projection. Extensive experimental tests has been conducted on a test bench for active noise control.The talk will start with a brief review of the basic adaptive regulation techniques followed by the analysis of the adaptive regulation schemes in the presence of large plant model uncertainties. Experimental results will illustrate the potential of the approach proposed. Open theoretical problem will be mentioned in the conclusion of the talk