A hybrid testing method based on adaptive feed-forward filters

Abstract

Hybrid testing is a technique that allows tests on components of complex systems with realistic boundary conditions. This can be achieved by coupling test rigs with simulation models of the overall system. Applications of hybrid testing to problems involving complex flexible structures demand careful controller design which may be prone to changes in system dynamics or model inaccuracies. Adaptive feed-forward filters offer a robust approach to coupling experimental and virtual subcomponents. In this paper, a framework based on adaptive feed-forward filters with harmonic basis functions is presented. Within the framework, a Least-Means-Squares adaption law and a Recursive-Least-Square adaption law using QR-factorization are proposed. The methods permit an automated work-flow of identification and control, and allow the coupling of structures with multiple degree-of-freedom interfaces. Questions related to measurement and actuation of interfaces are discussed. The methods are successfully applied to a double-clamped beam with an interface composed of two degrees of freedom. The results show the agreement of coupled test and reference system in the frequency domain as well as the ability of the methods to couple systems with arbitrary harmonic excitations. The experiments, furthermore, include a comparison of the different adaption laws in terms of convergence behavior.