风浪流多场耦合下车-桥动力响应预测模型研究

The cross-sea long-span bridges often serve in the coupled wind-wave-current environment. The constant interactions between the cross-sea long-span bridge and the coupled wind-wave-current environment not only have significant influence on the dynamic characteristics of the bridge, but also impose huge impact on the vehicle running safety and ride comfort. Nevertheless, the research on the dynamic characteristics of the vehicle-bridge system under the coupled wind-wave-current actions is quite limited in the literature currently. This project proposes an innovative framework aiming to systematically investigate the dynamic characteristics of the vehicle-bridge system in the coupled wind-wave-current field through the integration of the Copula Theory and the Machine Learning Theory. This project contains three parts: (1) we build high-dimensional time-varying Vine Copula models in order to capture the tail dependence and the time-varying dependence structures among the multivariant wind-wave-current parameters; (2) we establish a sophisticated scheme to predict the wave-current force exerted on the bridge substructure in the coupled wind-wave-current field, by combing the numerical simulations and the wave tank experiments; (3) we construct a forecast model to facilitate the prediction of the dynamic responses of the vehicle-bridge system under the coupled wind-wave-current actions based on the Machine Learning Theory. The forecast model enables the investigation into the influence of each field (wind, wave and current field) on the vehicle-bridge dynamic responses, and allows to pinpoint the most influential field as well as the dominant parameters. The research achievements of this project will gain a deeper insight into the dynamic characteristics of the vehicle-bridge system under the coupled wind-wave-current actions. It will also provide important theoretical supports for the design and operation management of the cross-sea long-span bridges as well as to ensure the traffic is running smoothly and safely.

跨海大桥长期处于风浪流多场耦合的复杂环境中，这些环境荷载对桥梁的振动特性以及行车安全性和舒适性均有重要影响。然而目前国内外针对风浪流多场耦合下车-桥动力响应的研究相当有限。本项目将Copula理论和机器学习理论在风浪流-车-桥耦合振动研究领域进行创新，深入探究了风浪流多场耦合下车-桥动力响应的规律。主要研究内容为：（1）建立多维时变风浪流要素联合分布的Vine Copula模型，揭示了风浪流要素间的尾部相关性和时变相关性；（2）结合数值模拟和物理模型实验精确模拟了风浪流耦合场中桥梁下部结构的波流力；（3）基于机器学习理论建立了风浪流多场耦合下车-桥动力响应的预测模型，揭示了风浪流各物理场对车-桥动力响应的影响规律、探明了影响车-桥动力响应的关键物理场及主导因素。本项目的研究成果将加深对风浪流多场耦合下车-桥系统动力行为的认识，为跨海大桥的设计和运营管理，以及车辆的安全平稳运行提供理论支持。

风—浪—车—桥系统是多重随机激励作用下的时变耦合振动系统，是多方向、多学科的交叉。本项目围绕多维时变风浪流要素联合分布模拟、风浪流多场耦合中桥梁下部结构波流力的精确模拟、以及风浪流多场耦合下车—桥动力响应的预测模型三个方面展开工作，主要的研究成果包括：1. 基于Vine Copula理论建立了能模拟风浪要素多维联合概率分布的模型；2. 提出了基于修正Morison方程的孤立波作用下跨海桥梁桥墩波浪荷载模型；3. 构建了基于改进元胞自动机车流模型的风—车流—桥耦合系统；4. 充分考虑风、浪、车、桥之间的相互作用，建立了风—浪—车—桥耦合振动数值模型，并研究了该系统的动力特性；5. 探明了地震动的空间变异性对风—浪—车—桥耦合振动系统动力特性，并提出了桥上行车安全的评估方法；6. 揭示了基础冲刷对风—浪—车—桥耦合振动系统中桥梁自振频率和车—桥系统动力响应的时频特性；7. 阐明了随机风、车流联合作用下大跨公路悬索桥纵向振动特性，并开展了纵向减振及阻尼器参数优化研究；8. 建立了大跨度公路悬索桥涡振条件下风—车—桥耦合振动模型，在此基础上，通过引入国际标准评价法ISO 2631-1:1997(E)，提出了大跨度公路悬索桥涡振条件下人体全身振动评价方法。通过本研究加深了多动力（风、浪、地震等）共同作用下车—桥耦合振动系统的理解，研究成果具有广泛的应用前景。