Structural shape optimization plays an important role in the design of wind‐sensitive structures. The numerical evaluation of aerodynamic performance for each shape search and update during the optimization process typically involves significant computational costs. Accordingly, an effective shape optimization algorithm is needed. In this study, the reinforcement learning (RL) method with deep neural network (DNN)‐based policy is utilized for the first time as a shape optimization scheme for aerodynamic mitigation of wind‐sensitive structures. In addition, “tacit” domain knowledge is leveraged to enhance the training efficiency. Both the specific direct‐domain knowledge and general cross‐domain knowledge are incorporated into the deep RL‐based aerodynamic shape optimizer via the transfer‐learning and meta‐learning techniques, respectively, to reduce the required datasets for learning an effective RL policy. Numerical examples for aerodynamic shape optimization of a tall building are used to demonstrate that the proposed knowledge‐enhanced deep RL‐based shape optimizer outperforms both gradient‐based and gradient‐free optimization algorithms.

Inherent time-varying trends of wind records are frequently captured in recent field measurements of tropical cyclones. These observations indicate a nonstationary process that deviates from the traditional stationary assumption. In this study, a strong wind event from a landfall typhoon recorded at Sutong Bridge site (Jiangsu Province, China) in 2008 is selected for the investigation of stationary and nonstationary wind characteristics. Since the extraction of the underlying trend is usually determined by experience, a self-adaptive method is presented to automatically isolate the time-varying mean based on the signal stationarity. Once the trend is removed, the residual turbulence will naturally be stationary. Accordingly, two categories of fluctuating wind speeds are then obtained by subtracting constant or time-varying means from original wind samples. Based on the traditional stationary model and two recent nonstationary models, stationary and nonstationary turbulent wind characteristics, e.g...

Advances of the analytical, numerical, experimental and field-measurement approaches in wind engineering offers unprecedented volume of data that, together with rapidly evolving learning algorithms and high-performance computational hardware, provide an opportunity for the community to embrace and harness full potential of machine learning (ML). This contribution examines the state of research and practice of ML for its applications to wind engineering. In addition to ML applications to wind climate, terrain/topography, aerodynamics/aeroelasticity and structural dynamics (following traditional Alan G. Davenport Wind Loading Chain), the review also extends to cover wind damage assessment and wind-related hazard mitigation and response (considering emerging performance-based and resilience-based wind design methodologies). This state-of-the-art review suggests to what extend ML has been utilized in each of these topic areas within wind engineering and provides a comprehensive summary to improve understanding how learning algorithms work and when these schemes succeed or fail. Moreover, critical challenges and prospects of ML applications in wind engineering are identified to facilitate future research efforts.

The consideration of train-bridge system under winds has been attracted extensive attentions, but the winds are typically treated as stationary. This stationary assumption clearly presents a departure from the field observations during extreme storms (e.g., tropical cyclones, thunderstorms and tornadoes). The assurance of structural safety and reliability requires accurate modeling of the non-stationary features in the coupled high-speed train-bridge vibration system. In this study, an efficient analysis framework for high-speed train-bridge coupled vibrations under non-stationary winds based on the pseudo excitation method (PEM) has been developed, in which the non-stationary winds were transformed into a series of pseudoharmonic excitation vectors. The high simulation fidelity and computational efficiency of the established analysis framework were verified based on a case study where the train run over a seven-span continuous girder high-speed railway bridge. A number of transient durations and maximum wind speeds associated with various extreme events are investigated to comprehensively examine the non-stationary effects on the high-speed train-bridge coupled vibrations. It has been demonstrated that the non-stationary characteristics of winds presented important contributions to the dynamic performance of the coupled train-bridge interaction system.