Anomaly Detection on Multivariate Time Series from CNC Machining using Machine Learning techniques

Gabriel Copolecchia Carvalhal

doi:10.34626/9gdz-vr58

Please use this identifier to cite or link to this item: https://hdl.handle.net/10216/153047

Full metadata record

DC Field	Value	Language
dc.creator	Gabriel Copolecchia Carvalhal
dc.date.accessioned	2025-11-07T23:56:58Z	-
dc.date.available	2025-11-07T23:56:58Z	-
dc.date.issued	2023-09-26
dc.date.submitted	2023-09-29
dc.identifier.other	sigarra:644538
dc.identifier.uri	https://hdl.handle.net/10216/153047	-
dc.description	The research makes a significant contribution by applying a Machine Learning method capable of interpreting anomalous behaviors in unlabeled multivariable time series data from Jasil's CNC machines. The project primarily focuses on identifying machine cycles, where each interval corresponds to the manufacturing of a product, allowing for a better understanding of the temporal patterns and dependencies within the data. To enable more effective and in-depth analysis of anomalous behaviors within each cycle, feature extraction techniques are employed to capture relevant information from the dataset. This process helps identify features that hold higher representability, thereby enhancing the accuracy, precision and recall of the detection process. Furthermore, the research utilizes a range of Machine Learning techniques to identify anomalous points, harnessing the power of algorithms to detect patterns and outliers. By employing different Machine Learning techniques, the research aims to enhance the identification of anomalies, leveraging the strengths of each method and creating a more robust anomaly detection system for the industry.
dc.description.abstract	The industry 4.0 paradigm paves the way that manufacturing factories develop their real-time monitoring capabilities. In this context, this study investigates machine learning applications in real-world data from CNC machining sensors in Jasil, Portugal, for an anomaly detection task as industrial quality control. In order to improve the availability of systems, reduce maintenance costs, increase operational performance, and support decision-making.
dc.language.iso	eng
dc.rights	embargoedAccess
dc.subject	Outras ciências da engenharia e tecnologias
dc.subject	Other engineering and technologies
dc.title	Anomaly Detection on Multivariate Time Series from CNC Machining using Machine Learning techniques
dc.type	Dissertação
dc.date.embargo	2026-09-25
dc.contributor.uporto	Faculdade de Engenharia
dc.identifier.doi	10.34626/9gdz-vr58
dc.identifier.tid	203423518
dc.subject.fos	Ciências da engenharia e tecnologias::Outras ciências da engenharia e tecnologias
dc.subject.fos	Engineering and technology::Other engineering and technologies
thesis.degree.discipline	Mestrado em Engenharia e Ciência de Dados
thesis.degree.grantor	Faculdade de Engenharia
thesis.degree.grantor	Universidade do Porto
thesis.degree.level	1
rcaap.embargofct	Privacidade e segurança de dados de empresa terceira. Política e proteção de dados.
Appears in Collections:	FEUP - Dissertação

Files in This Item:

File	Description	Size	Format
644538.pdf Restricted Access	Anomaly Detection on Multivariate Time Series from CNC Machining using Machine Learning techniques	4.13 MB	Adobe PDF	View/Open

Show simple item record Recommend this item Display Statistics