A novel approach to feedforward higher-harmonic control

Abstract

A novel adaptive filter structure is proposed for the control of systems characterized by higher harmonic response. The control approach has been designated the higher harmonic least-mean squares (HLMS) algorithm to differentiate it from the standard multifrequency filtered-x version of the least-mean squares approach. In the HLMS algorithm, a single frequency reference at that of the fundamental is all that is required to implement the controller. The remaining harmonics are generated internally based upon simple trigonometric relationships. The filtered-x LMS algorithm is implemented in parallel for each frequency to be controlled, minimizing or totally eliminating the contribution of the time-varying terms during the convergence process and increasing the rate of convergence for the higher harmonic control application. Results from the simulation demonstrate that the HLMS approach is far superior to the standard multifrequency, filtered-x LMS algorithm in adaptive, feedforward, higher-harmonic control.