Abstract
In the wind-resistant design of cladding, existing estimation methods approximate to space-averaged peak loading by temporally filtering the time series measured from a pressure tap. This study applied the machine learning method to reconstruct the pressure distribution on a panel with non-linearity and high spatial resolution (i.e., super-resolution field), which allows the estimation of peak pressure by direct space averaging. The top corner tile of high-rise buildings was focused on, and the super- resolution pressure fields were reconstructed from low-resolution measurements under both windward and leeward conditions. The reproduced time-averaged pressures agree well with experiments. The super-resolution method has higher precision in the prediction of extreme pressure than traditional single-point time-filtered method, and has strong robustness which is independent of selection of individual measurement tap.
