Roof Wind Pressure Reconstruction Method based on NMF and LSF

For the problems of high cost, large amount of data and poor real time of roof health monitoring, this paper proposes a method of roof wind pressure reconstruction based on non-negative matrix factroization and least squares fit. The method combines the field measurement method with the numerical wind tunnel simulation method to obtain the real-time wind pressure distribution of the whole roof by a small amount of sensor data. The proposed method is also validated using Fluent software. The reconstructed error of the obtained results is less than 5%. It is demonstrated that this method can reconstruct the roof wind pressure quickly and accurately, reduce the measurement cost, and facilitate the wind pressure monitoring for more roofs.