학술논문

An optimal discrete-time feedforward compensator for real-time hybrid simulation
Document Type
Article
Source
Smart Structures and Systems, An International Journal, 20(4), pp.483-498 Oct, 2017
Subject
토목공학
Language
English
ISSN
1738-1584
Abstract
Real-Time Hybrid Simulation (RTHS) is a powerful and cost-effective dynamic experimental technique. To implement a stable and accurate RTHS, time delay present in the experiment loop needs to be compensated. This delay is mostly introduced by servo-hydraulic actuator dynamics and can be reduced by applying appropriate compensators. Existing compensators have demonstrated effective performance in achieving good tracking performance. Most of them have been focused on their application in cases where the structure under investigation is subjected to inputs with relatively low frequency bandwidth such as earthquake excitations. To advance RTHS as an attractive technique for other engineering applications with broader excitation frequency, a discrete-time feedforward compensator is developed herein via various optimization techniques to enhance the performance of RTHS. The proposed compensator is unique as a discrete-time, model-based feedforward compensator. The feedforward control is chosen because it can substantially improve the reference tracking performance and speed when the plant dynamics is well-understood and modeled. The discrete-time formulation enables the use of inherently stable digital filters for compensator development, and avoids the error induced by continuous-time to discrete-time conversion during the compensator implementation in digital computer. This paper discusses the technical challenges in designing a discrete-time compensator, and proposes several optimal solutions to resolve these challenges. The effectiveness of compensators obtained via these optimal solutions is demonstrated through both numerical and experimental studies. Then, the proposed compensators have been successfully applied to RTHS tests. By comparing these results to results obtained using several existing feedforward compensators, the proposed compensator demonstrates superior performance in both time delay and Root-Mean-Square (RMS) error.