Development of generic AI models to predict the movement of vehicles on bridges

Abstract

For civil, mechanical, and aerospace structures to extend operation times and to remain in service, structural health monitoring (SHM) is vital. SHM is a method to examining and monitoring the dynamic behavior of essential constructions. Because of its versatility in detecting unfavorable structural changes and enhancing structural dependability and life cycle management, it has been extensively used in many engineering domains, especially in civil bridges. Due to the recent technical developments in sensors, high-speed internet, and cloud computing, data-driven approaches to structural health monitoring are gaining appeal. Since artificial intelligence (AI), especially in SHM, was introduced into civil engineering, these modern and promising methods have attracted significant research attention. In this work, a large dataset of acceleration time series using digital sensors was collected by installing a structural health monitoring (SHM) system on Nibelungen Bridge located in Worms, Germany. In this paper, a deep learning model is developed for accurate classification of different types of vehicle movement on the bridge from the data obtained from accelerometers. The neural network is trained with key features extracted from the acceleration dataset and classification accuracy of 98 % was achieved.