Abstract:Detailed wind pressure information plays a critical role in the accurate estimation of wind loads on high‐rise buildings, especially for complex‐shaped supertall buildings. However, owing to the limited internal space of a scaled building model and the capacity of data‐acquisition devices, it is often difficult to acquire the wind pressures at all positions of interest on the entire model in wind tunnel testing. To this end, a novel approach that combines the proper orthogonal decomposition (POD) and machine learning (ML) is presented in this paper for the prediction of wind pressure time series (WPTS) on supertall building models in wind tunnel testing. In this approach, the prediction of WPTS is converted into the estimation of several main eigenmodes and mean wind pressures by combining the POD with ML. This strategy can effectively reduce the computational effort compared to the direct prediction of WPTS. A combined ML model consisting of the Gaussian process regression (GPR), decision tree regression (DTR), and random forest (RF) (i.e., POD‐GPR‐DTR‐RF model) is utilized for the prediction of eigenmodes and mean wind pressures. Wind pressure records from a wind tunnel experiment of a 600‐m‐high building are employed to verify the accuracy and effectiveness of the presented approach. The results show that the combined ML model (i.e., POD‐GPR‐DTR‐RF model) developed based on the proposed approach performs satisfactorily in the prediction of WPTS and outperforms the conventional method that combines POD with backpropagation neural network model (i.e., POD‐BPNN model), demonstrating that the proposed approach is an effective tool for prediction of WPTS on supertall buildings.

POD Technique and Its Application to Prediction of Wind Pressure Fields on Roof

Numerical simulation of wind pressures on low-rise hip roofed buildings

Pressure Modes For Hyperbolic Paraboloid Roofs

Predicting roof-surface wind pressure induced by conical vortex using a BP neural network combined with POD

The prediction on non-Gaussian characteristics of wind pressure for the long-span roof in the mountainous area using proper orthogonal decomposition–deep learning framework

NUMERICAL SIMULATION OF WIND PRESSURES ON CANTILEVER GRANDSTAND ROOFS

Wind Loads on Fixed-Roof Cylindrical Tanks with Very Low Aspect Ratio

Numerical simulation of wind pressures on low-rise building with gable roof

Numerical Simulation Analysis of Multi-types Wind Disturbance on Gabled Roof of Low-rise Building

Simulation and analysis of wind pressure distribution on low-rise buildings

Field measurement of wind characteristics on roof of the national stadium

Wind Tunnel Experiment and Numerical Simulation of Wind Pressure Distribution of Large-Span Membrane Structure

Wind tunnel experiment and numerical simulation of wind pressure distribution of long-span ellipsoidal shell structure

NUMERICAL SIMULATION AND TEST INVESTIGATION OF WIND PRESSURES ON ROOF OF JINGGANGSHAN AIRPORT BUILDING

Extended spectral proper orthogonal decomposition for analysis of correlated surrounding flow structures and wind load components of a building

NUMERICAL SIMULATION OF WIND PRESSURES ON A LOW-RISE BUILDING COMPLEX WITH GABLE ROOFS

DETERMINATION OF WIND LOADS ON CANTILEVERED ROOFS OF STADIUM GRANDSTAND

Prediction of wind pressures on supertall buildings based on proper orthogonal decomposition and machine learning

Wind-induced Dynamic Response Analysis of Long-Span Roof Structure Based on Selection of Dominant Vibration-Modes with POD Method

POD-LSTM model for predicting pressure time series on structures

Higher order dynamic mode decomposition of wind pressures on square buildings