Beyond the delay barrier in adaptive feedforward active noise control using Youla–Kučera parametrization

Abstract

Adaptive feedforward broad-band noise compensation is currently used when a correlated measurement with the disturbance (an image of the disturbance) is available. Most of the active feedforward noise control systems feature an internal "positive" acoustical feedback between the compensation system and the reference source (a correlated measurement with the disturbance) which has to be taken into account. Adaptive algorithms for active feedforward noise attenuation have been implemented such that the propagation delay between the compensatory actuator and the measurement of the residual noise (the secondary path) be much smaller than the propagation delay between the reference source (image of the incoming noise) and the measurement of the residual noise (the primary path). Nevertheless, there are potential fields of applications where the propagation delay of the secondary path may be larger than the one of the primary path. The present paper explores the behaviour of the available adaptive feedforward compensation algorithms in this new context. The algorithms have been tested experimentally on a relevant test bench. All the algorithms except the Youla-Kučera finite impulse response (YKFIR) adaptive compensator and the standard FIR adaptive compensator using a stability based filtered adaptation (FUSBA) lead to an unstable behavior. In terms of performance the YKFIR provides the best performance.

Advances in Youla-Kucera parametrization: A Review

“…The use of YK parametrization in disturbance rejection has been compared to different robust techniques Landau and Meléndez (2017) , Landau, Meléndez, Airimitoaie, and Dugard (2019b) , Doumiati et al (2017) . Experimentation proved its efficiency against different disturbance types (known/unknown, narrow/broad band).…”

Improving performance of adaptive feedforward noise attenuators using a dynamic adaptation gain

“…For the purpose of this paper (comparison of various adaptation/learning algorithms), the FIR Youla-Kučera configuration for the feedforward compensator has been selected. This configuration has been described previously in the active vibration context [12], [4] and in the active noise context [13,14]. It offers the great advantage of defining a priori the closed loop poles of the internal positive feedback loop independently of the values of the adaptive FIR filter assuring the attenuation of the incoming disturbance.…”

Handling large model uncertainty in adaptive feedback noise attenuation by overparametrization

“…By simple computation, it can be shown that the poles of the closed loop remain unchanged in the presence of a Q filter having the structure given in Eq. (7).…”

“…Use of nQ < 11 (12 parameters) leads to instabilities for almost all the frequencies within this range. The attenuation in steady state 7 for various frequencies of the disturbance are summarized in row 2 of Table 3 . Fig.…”

“…Excellent models of the compensatory path in active noise attenuation can be obtained by experimental system identification (very small differences between simulations and real experiments). See for example [6,7]. Changes of the physical configurations and of various operational conditions lead to models of the same structure but with different orders and parameters.…”