Reconstructing Health Monitoring Data of Railway Truss Bridges using One-dimensional Convolutional Neural Networks

Abstract

Structural Health Monitoring (SHM) system uses sensors to collect information and evaluate the structure, aiming for early damage detection. For many reasons, data from sensors can be corrupted, affecting the assessment results. Reconstructing lost or corrupted data helps complete it, improves structural assessments, and ensures structural safety. Artificial Intelligence (AI) has emerged in recent years as a solution to data problems. This study proposes the use of a One-Dimensional Convolutional Neural Network (1DCNN) to reconstruct lost vibration data during SHM. A complete dataset was used to train the 1DCNN network. After completing the training, the 1DCNN network received incomplete data to return erroneous data. The results of the study show that the proposed method is able to reconstruct vibration sensor data.