Experimental implementation of a self-tuning control system for decentralised velocity feedback

Abstract

Simulations have previously shown that, for broadband excitation, adjusting the gain of a local velocity feedback loop to maximise their absorbed power also tends to minimise the kinetic energy of the structure under control. This paper describes an experimental implementation of multiple velocity feedback loops on a flat panel, whose gains can be controlled automatically by an algorithm that maximises their local absorbed power. Taking care to remove excessive phase shift in the control loop allows a stable feedback gain that is high enough to experimentally demonstrate the transition in control action between optimum damping and pinning of the structure. A simple search algorithm is then used to adapt the feedback gains of two control loops to maximise their local absorbed powers, thus demonstrating self-tuning. By measuring the power absorbed by each of these loops and also estimation of the kinetic energy of the plate from velocity measurements for a wide range of the two feedback gains, it is shown that not only does the adaptive algorithm converge to a set of feedback gains that maximise total power absorbed by the two feedback loops, but also that this set of feedback gains is very close to those that minimise the measured kinetic energy of the panel.