Home > Publication > Ph.D. Dissertation  
Title Feedforward control using IIR based adaptive filter and modified error path
Author S.H.Oh
Type KAIST Ph.D. Dissertation
Year of Pub. 2001
Adaptive feedforward control algorithms based largely upon LMS approach have been developed in recent two decades, and have been widely applied to practical sound and vibration control problems where the reference signal is available. In this study, FIR adaptive filter structure that has been used mostly in adaptive feedforward control strategy was changed into IIR and the dynamics of the error path were modified in order to increase the feedforward control performance. IIR based adaptive filter, which uses a linear combination of IIR filter bases, is proposed to overcome the shortcomings of FIR filter while keeping its advantages. It is based on a linear combination of IIR structures but there are neither instability nor nonlinearity problems on weight updating process. The multiple (constraint) Filtered-X LMS algorithm is derived with the proposed filter. Through the experiment of road booming noise attenuation, we demonstrate the effectiveness of the proposed methods. The change of error path dynamics by feedback controller may improve the feedforward control performance. Thus the design criteria and methodology of feedback controller for enhancement of convergence speed and steady-state error of feedforward control are proposed via the eigen-analysis of filtered-x signal. Moreover, three design criteria based on the Kharitonov theorem are also proposed in order to guarantee the stability of feedforward controller combined with feedback loops in the presence of parametric uncertainty. Two modified criteria reduce the conservatism of Kharitonov theorem because it gives a highly conservative result for a small number of parametric uncertainties. Simulation and experiment show the feasibility and efficiency of the hybrid controller that was designed based on the proposed design method.