Journal article
Diagnosability Study of Multistage Manufacturing Processes Based on Linear Mixed-Effects Models
Technometrics, Vol.45(4), pp.312-325
11/01/2003
DOI: 10.1198/004017003000000131
Abstract
Automatic in-process data collection techniques have been widely used in complicated manufacturing processes in recent years. The huge amounts of product measurement data have created great opportunities for process monitoring and diagnosis. Given such product quality measurements, this article examines the diagnosability of the process faults in a multistage manufacturing process using a linear mixed-effects model. Fault diagnosability is defined in a general way that does not depend on specific diagnosis algorithms. The concept of a minimal diagnosable class is proposed to expose the "aliasing" structure among process faults in a partially diagnosable system. The algorithms and procedures needed to obtain the minimal diagnosable class and to evaluate the system-level diagnosability are presented. The methodology, which can be used for any general linear input-output system, is illustrated using a panel assembly process and an engine head machining process.
Details
- Title: Subtitle
- Diagnosability Study of Multistage Manufacturing Processes Based on Linear Mixed-Effects Models
- Creators
- Shiyu ZhouYu DingYong ChenJianjun Shi
- Resource Type
- Journal article
- Publication Details
- Technometrics, Vol.45(4), pp.312-325
- Publisher
- Taylor & Francis
- DOI
- 10.1198/004017003000000131
- ISSN
- 0040-1706
- eISSN
- 1537-2723
- Language
- English
- Date published
- 11/01/2003
- Academic Unit
- Industrial and Systems Engineering
- Record Identifier
- 9984064583802771
Metrics
18 Record Views