Non-Collocated Flexible Beam Motion Control based on a Delayed Adaptive Inverse Method

Abstract

The viability of a delayed adaptive inverse method for motion control of the tip of a flexible beam is demonstrated in this paper. This method is based on an adaptive linear FIR filter which provides a stable and close approximation to the inverse plant dynamics. Such FIR filters can be used to control non-minimum phase systems, certain nonlinear systems, or plants of unknown dynamics and can be implemented using real-time interrupt driven high performance computational devices such as digital signal processing (DSP) based hardware. The effects of varying parameters of the delayed adaptive inverse controller (such as the values of the initial weights, the gain constants, and the length of delay) on the system performance are presented. Since the method is computationally efficient, it can be used in high bandwidth applications for control of complex plants given only the desired and the actual measured outputs for the plant. A feedback controller is used in conjunction with the delayed adaptive inverse controller to eliminate the residual errors that would occur if only feedforward control were used. This delayed adaptive inverse method is then compared to an H¿-based loopshaping controller.