Robust and high fidelity real-time hybrid substructuring

Abstract

In many engineering applications, mechanical contact leads to unwanted dynamic phenomena, such as excitation of high frequency modes. To investigate the induced dynamics, systems need to be tested already in the early development stage. Real-Time Hybrid Substructuring (RTHS) is a Hardware-in-the-Loop approach, that enables testing of critical components with realistic boundary impedance. In RTHS, the critical components are tested experimentally, while the remaining system is simulated numerically in a co-simulation. The dynamics of the transfer system, which couples the experimental part with the numerical simulation render the test outcome inaccurate or even make it unstable unless they are compensated for. The aim of this work is to tackle two major shortcomings of RTHS, namely stability and fidelity of RTHS tests. This is achieved by the combined use of Normalized Passivity Control and Iterative Learning Control. Normalized Passivity Control guarantees passivity of the transfer system and Iterative Learning Control improves the actuator tracking performance iteratively in order to increase the fidelity of the test outcome. Furthermore, a convergence condition is proposed for Iterative Learning Control in this setup such that an optimal tuning of the learning law can be achieved. We tested the proposed compensation scheme experimentally for an RTHS test, where contact occurs. A special focus is put on investigating unstable RTHS tests, as their stabilization and accurate testing is a particular challenge. The results show that the test fidelity, which is measured by the relative root-mean-square error, can be improved by a factor of about 6.5. In addition, it is shown that the combination of velocity feedforward, Iterative Learning Control and Normalized Passivity Control improves the tracking and thus the test fidelity even further. The results reveal that the proposed method has the potential not only to overcome the nonlinear effects of friction in the actuator, but also to improve the tracking accuracy in highly dynamic motion tasks. The main advantages over existing compensation techniques are that only little knowledge about the transfer dynamics of the actuator and experimental part is needed and that the implementation is simple. Hence, it is assumed that this method will be of interest for testing with RTHS in many engineering applications, with or without contact.