Smart sensing with edge intelligence for autonomous structural health monitoring

In the pursuit of Singapore becoming a Smart Nation, the integration of advanced technologies, such as sensors in structural health monitoring systems, has become paramount. This paper delves into the significance of sensor reliability in these monitoring systems. Sensor malfunctions, largely attrib...

Full description

Saved in:
Bibliographic Details
Main Author: Chen, John Yong Han
Other Authors: Fu Yuguang
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/172429
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:In the pursuit of Singapore becoming a Smart Nation, the integration of advanced technologies, such as sensors in structural health monitoring systems, has become paramount. This paper delves into the significance of sensor reliability in these monitoring systems. Sensor malfunctions, largely attributed to external environmental factors and internal hardware issues, compromise the reliability of these systems. With this in mind, this study aims to develop an Artificial Intelligence (AI) framework that integrates edge computing, distributed learning, and structural health monitoring with sensor fault diagnosis. Given the increasing emphasis on sustainability, this paper also explores various compact AI models suitable for edge devices, their deep compression techniques, and the deployment of these models using federated machine learning. These models are evaluated based on Green AI criteria, such as training time, floating-point operations, and model size to identify the most optimally performing model. The study also discusses data pre-processing techniques, model compression methods, and deployment strategies on edge devices like the Raspberry Pi. The results from this study highlight the potential of using AI in structural health monitoring systems, offering potential solutions that are not only effective but also environmentally sustainable. Through the development and deployment of these compact models, the study sets the stage for the future of reliable and sustainable infrastructural health monitoring.