Model Predictive Control for Real-Time Hybrid Testing of Multibody Dynamical Systems

Abstract

Real-time hybrid testing provides a practical way to evaluate engineering systems under realistic operating conditions by combining numerical simulation with physical experiments. For multibody dynamical systems, however, nonlinear kinematics, coupled motion, and actuator delay make it difficult to maintain both kinematic compatibility and force equilibrium between the numerical and physical substructures. Existing model predictive control approaches for real-time hybrid testing mainly address trajectory tracking. This study proposes a model predictive control framework for real-time hybrid testing of multibody dynamical systems in which the coordination problem is formulated as a regulation problem. The framework directly targets simultaneous satisfaction of kinematic compatibility and interface force equilibrium. The general formulation is first assessed in a virtual real-time hybrid testing environment using a three-dimensional six-degree-of-freedom multibody dynamical system. The virtual results show accurate coupling, recovery of the nonlinear system response despite linearized predictive models, and robustness to moderate mismatch in the predictive model. For experimental implementation, the framework is reduced to a real-time realization with a closed-form solution. Implemented on a multi-axial simulation-table setup, the reduced controller executes in 95.6 μs, corresponding to 19.6 % of the available control period at 2048 Hz, confirming real-time viability. The experiments show stable and bounded operation, satisfactory kinematic compatibility and force equilibrium, and physically consistent interaction in the presence of nonlinear and dissipative physical-substructure behavior. These results demonstrate the potential of regulation-based predictive coordination for real-time hybrid testing of nonlinear multibody dynamical systems.