Capezza, Christian (2020) Statistical Methods for Industrial Process Monitoring Based on Functional Data Analysis. [Tesi di dottorato]

[thumbnail of Capezza_PhD_thesis.pdf]
Preview
Text
Capezza_PhD_thesis.pdf

Download (16MB) | Preview
Item Type: Tesi di dottorato
Resource language: English
Title: Statistical Methods for Industrial Process Monitoring Based on Functional Data Analysis
Creators:
Creators
Email
Capezza, Christian
christian.capezza@unina.it
Date: 13 March 2020
Number of Pages: 163
Institution: Università degli Studi di Napoli Federico II
Department: Ingegneria Industriale
Dottorato: Ingegneria industriale
Ciclo di dottorato: 32
Coordinatore del Corso di dottorato:
nome
email
Grassi, Michele
michele.grassi@unina.it
Tutor:
nome
email
Palumbo, Biagio
UNSPECIFIED
Lepore, Antonio
UNSPECIFIED
Date: 13 March 2020
Number of Pages: 163
Keywords: Profile Monitoring; Scalar-on-Function Regression; Ship CO2 Emissions
Settori scientifico-disciplinari del MIUR: Area 13 - Scienze economiche e statistiche > SECS-S/02 - Statistica per la ricerca sperimentale e tecnologica
Date Deposited: 02 Apr 2020 08:35
Last Modified: 31 Oct 2021 21:32
URI: http://www.fedoa.unina.it/id/eprint/13242

Collection description

In this thesis, statistical methods for industrial process monitoring are proposed. The industrial scenario that motivates the research work is the monitoring and prediction of fuel consumption and CO2 emissions from maritime transportation. The two main objectives are, on the one hand, the prediction of fuel consumption (and/or CO2 emissions) on the basis of covariates describing the ship operating conditions at each voyage by means of advanced regression methods, and, on the other hand, the statistical process monitoring of the ship operating conditions and the fuel consumption (and/or CO2 emissions) based on control charts. The proposed methodologies can be arranged in three groups on the basis of how they treat the data for each voyage of a ship. The first group uses multivariate techniques, which, for each voyage, consider each individual observation of the variables as scalar quantities. Typically, the mean value of each variable over a voyage is considered. The second group considers the data for each voyage as profiles, from which several features are extracted in order to describe them in the best possible way. The third group considers the data for each voyage as functions, i.e. as complex, unique objects that have to be treated using functional data analysis techniques. The common ground of all the proposed methodologies is the need to provide tools to industrial practitioners that are easily interpretable and give clear indications of anomalies by identifying the related causes, possibly in real-time, and the use of real-data examples to demonstrate their predictive and monitoring abilities.

Downloads

Downloads per month over past year

Actions (login required)

View Item View Item