Building wind field ground roughness identification method based on multi-scale input neural network

• Main Authors: XIAO DANLING, QIU JIANLEI, LI QINGXIANG, XU WEI
• Format: Patent
• Language: Chinese; English
• Published: 17.05.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a building wind field ground roughness identification method based on a multi-scale input neural network, and the method comprises the following steps: S1, dividing at least two scales according to the size of the surrounding range of a building, and collecting a satellite image; s2, using an ESDU method to calibrate the ground roughness of the image; s3, constructing a multi-scale input convolutional neural network model, and respectively training the input sector diagram of each scale by adopting mutually independent convolutional neural networks until a SoftMax layer outputs a result; s4, fusing training output results of the fan-shaped diagrams with various scales on a SoftMax layer, and calculating a Loss function; s5, comparing Loss function value convergence errors to obtain a classification model; and S6, inputting the to-be-detected image into the classification model for identification and classification to obtain a classification result. The method is used for improving the