Active Vibration Control Paradigm for Stay Cable Vibration: A Hybrid Imitation and Deep Reinforcement Learning Approach
Structural Control and Health Monitoring · 2026-01 · 2026 (1)
DOI: 10.1155/stc/5588873 ↗Kurz gefasst
Fachartikel in der Zeitschrift Structural Control and Health Monitoring, online erschienen am 31.03.2026. Verfasst von Yi-Ang Zhang und 1 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Research Grants Council, University Grants Committee.
Abstract (Originalsprache)
Although classical active control techniques demonstrate superior performance in controlling structural vibrations, deriving the optimal control strategy often faces challenges due to the difficulties in accurately modeling complex dynamic systems in practical scenarios. Learning‐based control methods eliminate such a requirement by directly learning control strategies from structural behavior. However, past studies on learning‐based control algorithms primarily focused on their applicability without guaranteeing optimal control performance, resulting in a performance gap between learning‐based vibration control and model‐based optimal control. A large number of samples required in the learning process present another practical issue in real implementations. In response, this study presents a groundbreaking learning‐based control framework, which combines imitation learning (IL) and deep reinforcement learning (DRL) to mitigate structure vibrations. This approach involves initially training a deep neural network controller through behavior cloning of an extremely simple control strategy, followed by fine‐tuning using model‐free DRL. The framework’s feasibility and effectiveness are extensively examined in simulations by using a verified numerical model of a full‐scale stay cable system with an active damper. The control performance of the proposed model‐free method closely approaches that of the model‐based full‐state feedback controller, regardless of the number of observation states or measurement noises, and achieves better sample efficiency than other state‐of‐the‐art model‐free DRL algorithms. This innovative approach to structure vibration mitigation can revolutionize efficiency in practice, and its unparalleled effectiveness makes it a promising alternative in controlling stay cables or other high‐degree‐of‐freedom structures.
Förderung
- Research Grants Council, University Grants Committee
Zitieren
APA
Zhang, Y., & Zhu, S. (2026). Active Vibration Control Paradigm for Stay Cable Vibration: A Hybrid Imitation and Deep Reinforcement Learning Approach. Structural Control and Health Monitoring, 2026(1). https://doi.org/10.1155/stc/5588873
Vancouver
Zhang Y, Zhu S. Active Vibration Control Paradigm for Stay Cable Vibration: A Hybrid Imitation and Deep Reinforcement Learning Approach. Structural Control and Health Monitoring. 2026;2026(1). doi:10.1155/stc/5588873
BibTeX
@article{Zhang2026,
author = {Zhang, Yi-Ang and Zhu, Songye},
title = {Active Vibration Control Paradigm for Stay Cable Vibration: A Hybrid Imitation and Deep Reinforcement Learning Approach},
journal = {Structural Control and Health Monitoring},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/stc/5588873}
}
EUVIMED