A Markov data-based approach to system identification and output error covariance analysis for tensegrity structures

This paper introduces a data-driven approach to address the long-standing challenge of modeling complex tensegrity systems. The proposed approach focuses on approximating unknown black box systems and estimating their output error covariance using input/output (IO) information. First, an approximation system that mirrors the input–output relation of the black box system is obtained. Next, output error covariance between the approximation and the black box system is calculated, which evaluates the accuracy of the identified model. This two-step approach relies exclusively on the black box system’s Markov parameter sequence, eliminating the need for dynamics knowledge of the system. Nonlinear examples of a NACA 2412 tensegrity morphing airfoil and a 3D tensegrity prism are studied for validation. The proposed approach successfully identified approximation systems in state space realization in both cases with insignificant output error covariances. Compared to the widely-used Mode Displacement Method (MDM), the proposed approach exhibits an advantage in identifying velocity outputs for tensegrity systems. The developed approach in this paper applies to other tensegrity structures and structural identification problems.