Design and Evaluation of Frequency Weighted LQG—Maximum Entropy Controllers on an Experimental Truss Structure

Abstract

This paper will discuss the experimental evaluation of a structural control system designed using a combination of Frequency Weighted LQG and Maximum Entropy. The experimental implementation was performed on an experimental truss structure, the Sandia Truss, which is described in the paper. The control design model was obtained via experimental system identification using the eigensystem realization algorithm with data correlation. The control design used frequency weighting to stabilize the unstructured uncertainty of the system due to low signal-to-noise and uncertain system dynamics in various frequency ranges. Maximum entropy is used to provided robustness for structured uncertain system dynamics within the controller bandwidth. The experimental implementation of the controllers designed with this approach show the ability to design controllers with a specified bandwidth, gain stabilization of unstructured uncertainty, and robustness to structured uncertainty within the controller bandwidth.