Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.689314
Title: Non-destructive evaluation and condition monitoring of tool wear
Author: Seemuang, Nopparat
ISNI:       0000 0004 5918 6866
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
As tool wear is unavoidable, a tool condition monitoring system is an essential system to prevent machine tool downtime due to unnecessary tool replacement, tool breakage caused by using worn tools, and to reduce part rejections. Currently, multiple indirect sensing signals are commonly fused and used to detect tool wear to enhance the system reliability and generalisation. Many of the recently developed systems uses expensive sensing methods which are seemingly not suitable for real machining. Almost all of these monitoring systems were also developed in a laboratory control environment, resulting in lower performance if they are utilised in real machining operations. This thesis takes these discrepancies as motivation to investigate and develop low-cost based tool condition monitoring systems. The cost effective tool monitoring systems were developed from a non-destructive evaluation (NDE) method and a multiple sensor fusion approach in order to monitor tool wear of common machining processes (turning, drilling, and gear hobbing). First, Barkhausen noise technique, an off-line NDT method commonly used in the hardened case-depth evaluation, was used to evaluate the coating thickness of TiN and CrN layers on HSS cutting tools. The results confirm that this proposed measurement system can be successfully used to indicate between different coating thicknesses. Secondly, an on-line tool condition monitoring system based on multiple sensor fusion using a combination of inexpensive sensors (AE, microphone, and power monitoring system) was developed. Sensory features extracted from those sensors were trained by neural networks to obtain the tool wear prediction and tool wear state classification models. The system was successfully used to predict flank wear width and classify the tool wear states during a turning operation. Furthermore, a novel sensing feature extracted from cutting sound, named 'spindle noise', was first introduced in this study as this feature can successfully detect the excessive tool wear in turning and drilling or any other machining process which has a rotary spindle. The cost-effective systems proposed in this study can be utilised in small and medium sized manufacturing companies and will improve productivity and add more value to the manufacturing processes.
Supervisor: Slatter, Tom ; Marshall, Matt Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.689314  DOI: Not available
Share: