Balsamo, Vittorio (2017) Development of Cognitive Decision Making Methods for Tool Wear Prediction and Catastrophic Tool Failure Detection. [Tesi di dottorato]

[img] Text
BALSAMO_VITTORIO_30.pdf
Visibile a [TBR] Repository staff only

Download (13MB) | Request a copy
[error in script] [error in script]
Item Type: Tesi di dottorato
Lingua: English
Title: Development of Cognitive Decision Making Methods for Tool Wear Prediction and Catastrophic Tool Failure Detection
Creators:
CreatorsEmail
Balsamo, Vittoriovittorio.balsamo@unina.it
Date: 11 December 2017
Number of Pages: 146
Institution: Università degli Studi di Napoli Federico II
Department: dep08
Dottorato: phd038
Ciclo di dottorato: 30
Coordinatore del Corso di dottorato:
nomeemail
Mensitieri, GiuseppeUNSPECIFIED
Tutor:
nomeemail
Teti, RobertoUNSPECIFIED
Date: 11 December 2017
Number of Pages: 146
Uncontrolled Keywords: Machining, Tool condition monitoring, Cognitive paradigm
Settori scientifico-disciplinari del MIUR: Area 09 - Ingegneria industriale e dell'informazione > ING-IND/16 - Tecnologie e sistemi di lavorazione
Date Deposited: 08 Jan 2018 01:17
Last Modified: 07 Oct 2021 10:33
URI: http://www.fedoa.unina.it/id/eprint/12195

Abstract

The research work developed in this thesis is focused on cognitive decision making systems for tool condition monitoring of machining processes. Multiple sensor signals are acquired during machining and processed to extract relevant sensor signal features that are correlated with a chosen quality parameter representative of the tool condition. The final aim of the thesis work is to create a methodology to predict tool consumption and to detect in real-time a catastrophic tool failure potentially occurred. The research work has been addressed within the framework of the international research project EC FP7 “REALISM – Real Time in Situ Monitoring of Tool Wear in Precision Engineering Applications” and the national MIUR PON Project “CAPRI – Carrello per Atterraggio con Attuazione Intelligente”. The specific goal of REALISM project is to develop a robust ‘smart’ sensor-based tool condition monitoring system to provide in real-time a feedback loop to both the CNC machine and the operator. The aim of the CAPRI project is to develop innovative technologies for manufacturing of the main landing gear components and subsystems of a commercial aircraft in order to improve effectiveness in terms of performance, reliability and maintenance while improving the sustainability of the product and of its manufacturing processes. The two projects consider machining processes based on a green technologies approach. The present thesis work starts with an introductive chapter about sensor monitoring systems and their applications. A comprehensive survey review of the general concept of sensor monitoring of manufacturing processes and of the state of the art of sensor technologies, advanced signal processing techniques, sensor fusion approach, and cognitive decision making strategies for process monitoring is provided. Subsequently, the description of the experimental campaign carried out under various cutting conditions (cutting speed, feed, depth of cut) using a multi-sensor monitoring system is provided. In particular, has been descripted the CNC machine tools, the workpiece and the machining parameters used in the experimental campaign. In the CAPRI project, the experimental campaign on turning process was directly conducted and the signals directly acquired on the industrial site, while in the REALISM project the experimental campaign was performed by a project partner that provided the outcoming sensor signals and the parameters used in the three machining process under study: turning, boring and drilling. First of all, the catastrophic tool failure occurrence case study is discussed and analysed. The developed procedure starts with the analysis of sensor signals acquired and then, using a knowledge based approach, the catastrophic tool failure detection procedure is defined. A specific algorithm for the real-time detection has been implemented using the National Instrument LabVIEW software. The results obtained are moreover explained and discussed. Then, the consumed tool life prediction case study is discussed and analysed. In this case, a cognitive paradigm based on a neural network approach has been used for the identification of correlations between the sensor signal features and the consumed tool wear percentage. The training and testing procedure using signals features of four consumed tools is described. The obtained results are also explained and discussed. Finally, the last part of the thesis reports the concluding remarks and future developments of this work.

Downloads

Downloads per month over past year

Actions (login required)

View Item View Item