CNN Simulation Data Preprocessing Technique for Development of Damage Detecting Method for Bridges Based on Convolutional Neural Network

摘要

The present invention relates to a method for processing simulation data for development of a damage detection technique for a bridge structure using a CNN, in which the technique can be used even under various load loading situation related to the bridge structure. The simulation data processing method comprises: a first step of conducting simulation of the bridge structure to generate structural analysis simulation data on load of the bridge structure; a second step of converting the structural analysis simulation data into 3D matrix-shaped data; and a third step of normalizing the structural analysis simulation data converted into 3D matrix-shaped data in order to be used in CNN learning. Therefore, the simulation data processing method can remarkably reduce costs and time required for installation and testing of a system for acquiring measurement data by removing the need of acquisition of actual measurement data on the bridge structure.