Development of Smartphone-based AI-SHM Algorithm for Industry 4.0

Abstract

In the last few decades, technological advancements have transformed every industry, ushering in a new technological era known as the “Fourth Industrial Revolution”. In the case of civil engineering, advances in sensor technology, computer vision, and deep learning over the last decade have boosted the importance of innovative structural health monitoring system (SHMs) for the safety assessment of civil infrastructures. However, these SHMs are mainly applied to long span bridges. Moreover, economic problems in installation of various sensors systems and their operations have remained as critical issues resulting in the limited application of SHMs. Therefore, it is necessary to develop a new and cost-effective technique enabled to quickly measure the dynamic characteristics of bridges including small or medium scale. The main purpose of this dissertation is to develop a methodology for reliable and practical monitoring of bridges using smartphone. For this purpose, this study has devised SHM algorithm for dynamic response estimation of a bridge based on smartphone-embedded image sensor. This study utilized a smartphone as mobile sensor and communication interface. In order to evaluate the dynamic response of a bridge, a motion image, a rotation-robust cross-correlation template matching algorithm, and homography were developed. Augmented reality technology was adopted to visualize the estimated data on the smartphone screen. In order to reduce image processing time, this dissertation also proposes a method of automatically setting template image and region-of-interest (RoI). Moreover, to enhance the current marker tracking system, a YOLO model based on convolutional neural network was developed to accurately estimate the dynamic displacements. Based on these innovative approaches, a novel smartphone based-SHM algorithm has been developed capable of estimating dynamic displacement and visualization of data. The techniques introduced in this study can be effectively used for a comprehensive infrastructure safety evaluation in future.