Wind load simulation of large-span structure based on image recognition and neural network

image recognition technology;  neural networks;  Wind Lotus Planting Simulation;  Image information;  Calculate efficiency

Combining image recognition technology and neural network (ANN) system, a simulation method for wind load of large-span structure is given. Two image recognition technologies, recursive graph and feature face recognition, are used to convert the wind speed time numerical series into image information. Then, a large number of wind speed time series are converted into vector Ω with fewer dimensions, and then combined with the multi-layer ANN system to obtain a simplified neural network model to predict the wind speed time history of each point in space. Finally, it is applied to the wind load simulation of a large-span structure, and the results not only conform well with the target value, but also reduce the number of layers of the neural network, and greatly improve the operation speed and the credibility of the results.