Delay Sensitivity of Smith Predictor for Real-Time Hybrid Simulation

In a real-time hybrid simulation (RTHS), a multi-story structure is partitioned into numerical and physical substructures, and the vibration behavior of the physical substructure is tested within the real-time simulation. An actuator is employed to apply static and inertial forces to the physical substructure due to forces calculated by the numerical substructure. The actuator dynamic is approximated by a pure time delay, and the time delay in the closed-loop system causes inaccuracy results or even instability. The Smith predictor is adopted to minimize the adverse effect of time delay from the RTHS test results. The delay differential equation (DDE) modeling and Hopf analysis are used to determine the dependence of critical time-delay on mass ratios of the system. The method drives the stability crossing curves in the space of parameters defined by nominal delay, and delay uncertainty. The Smith predictor is a model-based approach for the compensation of time delay in delayed control systems. The Smith delay compensator is sensitive to model uncertainty, particularly for time delay mismatch. The effects of delay-induced uncertainty on the stability of the Smith Predictor control scheme are also analyzed. Sensitivity analysis of Smith predictor to delay mismatch shows a more stable margin for overestimation of delay regarded to underestimation, and the stable region becomes smaller in the area as time-delay increases.