Caputo, Enzo (2023) Innovative IoT-based Methods for remote control, monitoring and measurements. [Tesi di dottorato]

Anteprima

Testo Caputo_Enzo_36.pdf Download (7MB) | Anteprima

Abstract

The impact of the SARS-Cov-2 pandemic has prompted a reevaluation of human activities and interactions, significantly affecting educational practices and industrial processes. This thesis comprises two interconnected sections, each addressing distinct domains within the technological landscape. In Section I, the focus is on transforming didactic approaches, particularly in laboratory-based education, in response to pandemic-induced restrictions. Leveraging the Internet of Things (IoT) and the fourth industrial revolution, a solution is proposed that utilizes augmented reality (AR) on consumer devices for remotely controlling laboratory instruments. This extends to Cyber-Physical Systems (CPS) and digital twins, where AR serves as a Digital Twin for measurement instruments, demonstrating practicality through a digital storage oscilloscope application. Safety concerns in laboratory experiences, especially in STEM disciplines, are addressed through a mixed-reality solution, combining augmented reality on the student side and actual instrumentation on the laboratory side. The approach ensures a secure and enriching educational experience, as evidenced by experiments involving remote control of measurement instruments and a case study on gamma radiation measurements. Section II shifts focus to innovations in industrial systems, exploring Bluetooth Low-Energy (BLE), Micro-Electro-Mechanical Systems (MEMS), and predictive maintenance algorithms. In the automotive field has been investigated the use of smartphones and BLE Fingerprinting for a Passive Entry Passive Start (PEPS) system, overcoming spatial orientation challenges and showcasing the method's reliability in determining the driver's position within or outside the vehicle. In structural health monitoring (SHM) has been introduced a method combining redundant MEMS accelerometers with a Kalman Filter approach, compensating for inherent errors and establishing a real-time monitoring sensor node using LoRaWAN and NFC protocols. The final part explores the application of Long Short-Term Memory (LSTM) deep learning algorithms for predictive maintenance in the railway domain. Conducted within an academic-industrial partnership, the study showcases the efficacy of the methodology in forecasting failures of railway rolling stock equipment with remarkable accuracy.

Downloads

Actions (login required)

Modifica documento