Machine learning for bridge wind engineering

• DOI: 10.1016/j.awe.2024.100002
• Publication Date: 2024-07-03T21:10:18Z
• Authors: Zeyu Zhang; Shanwu Li; Hui Feng; Xuxi Zhou; Nan Xu; Hui Li; Shujin Laima; Wenli Chen
• Keywords: Bridge (graph theory); Wind engineering

Modeling and control are primary domains in bridge wind engineering. The natural field characteristics (e.g., non-stationary, non-uniform, spatial-temporal changing characteristics) the wind-bridge interaction processes physically complex exhibit strong nonlinearity. lack of analyzing these physical based on first principles makes it difficult for traditional modeling methods to accurately characterize or wind-induced vibration bridges. Data-driven automatic is a new expansion direction engineering due extensive valuable historical data available complex, high-dimensional character state space action agent active control. Machine Learning (ML) has been shown versatile various tasks its excellent ability automatically extract features efficiently make wonderful decisions facing complicated without manual interval. Starting from an overview basic concepts ML, this review will comprehensively analyze summarize application ML aspects engineering, including fields, static aerodynamic reconstruction, vibrations reinforcement learning, order enable readers have full knowledge potential subfields. essence rely data-driven minds solve tricky problems that cannot be solved principles. continue integrate coexist with bridges generate revolutions both theories applications owing powerful information-processing modeling-learning ability.