Rossi, Emanuele (2025) Statistical Methods for Sustainable Mobility. [Tesi di dottorato]

	Documento PDF Emanuele_Rossi_38_COMPLETO.pdf Visibile a [TBR] Amministratori dell'archivio Download (10MB) | Richiedi una copia
	Documento PDF Emanuele_Rossi_38_PARZIALE.pdf Visibile a [TBR] Amministratori dell'archivio Download (2MB) | Richiedi una copia

Abstract

In the Industry 4.0 era, the generation of immense volumes of data across diverse fields has fundamentally evolved the statistician's role. The challenge has shifted from mere data collection to learning from data—extracting actionable patterns to understand real-world contexts. This data-rich environment has necessitated the development of innovative statistical methodologies capable of handling massive, complex datasets to drive industrial value. This dissertation focuses on the application of such advanced statistical methods for the real-time control, monitoring, and design optimization of complex systems, specifically within the naval and railway sectors. The research presented herein is divided into two primary application domains. The first domain addresses the preliminary design phase in naval architecture. A supervised statistical learning approach, specifically Relative Least-Squares (RLS) estimation, is applied to model and optimize the Gross Tonnage (GT) of cruise ships. This application provides naval architects with a reliable, data-driven formula to estimate vessel carrying capacity, addressing the growing industry need to balance passenger capacity with onboard comfort. The core of the dissertation, co-funded by Hitachi Rail STS S.p.A., focuses on the railway sector, specifically the condition monitoring of Nickel-Cadmium (Ni-Cd) battery systems powering auxiliary systems on Electric Multiple Unit (EMU) trains. A novel Statistical Process Monitoring (SPM) framework based on Compositional Data Analysis (CoDa) is introduced to detect performance degradation. The research demonstrates that a MEWMA-CoDa control chart is highly effective for detecting gradual degradation, while a CoDa T2 chart is superior for identifying sudden deviations. The combined application of these charts enables predictive maintenance, reducing service interruptions and improving operational efficiency. In this dissertation, as regards the battery system case study, in addition to the application of SPM methodologies, as part of the PhD programme, the author collaborated with the System and Data Science team at the Hitachi Rail STS sites in Naples, Italy, to develop Python scripts for analyzing the status of the parallel-connected battery systems used in certain types of trains. Additionally, during an six-month international research period at Perpetuum ONBOARD, the research extended to the monitoring of mechanical assemblies comprising the train wheels, the corresponding bearings, and the sensor brackets. In this context, classic SPM methodologies, including T2 and MEWMA control charts, were utilized to detect anomalous vibrations, further enhancing operational reliability and safety. Aligned with the innovative Ph.D. programme established by the Ministry of University and Research (MUR) and funded by the PNRR, this thesis bridges the gap between academic research and industrial application. By providing reliable, innovative statistical tools for anomaly detection and design optimization, this work contributes to enhanced decision-making accuracy, energy efficiency, and environmental sustainability in modern industrial contexts.

Downloads

Actions (login required)

Modifica documento