This paper presents an approach to troubleshooting in Manufacturing Execution Systems (MES) to improve their reliability using passive monitoring. It is a result of cooperation between a semiconductor company and Dresden University of Technology. A network analyzer at fab message bus is used for monitoring, which subscribes transactions and their replies using publish/subscribe paradigm. The goal of evaluation of observation results is a cognitive á posteriori diagnosis. That means detection of redundancies, bottlenecks, error-symptoms, not yet classified situations and retrieval of á posteriori model of the observed normal behavior with intensive user interaction. The diagnosis tool “Extrakt” supports the troubleshooter with numerous analysis and visualization options. The event-based diagnostic rules can be derived automatically or be entered by diagnostician during iterative cognitive diagnosis. They are based on assumptions about production, software and use case scenarios.
[1]
John A. Self,et al.
Model-based cognitive diagnosis
,
1993,
User Modeling and User-Adapted Interaction.
[2]
Dieter Haban.
The distributed test methodology DTM
,
1988
.
[3]
Jeffrey J. P. Tsai,et al.
Distributed Real-Time Systems: Monitoring, Visualization, Debugging, and Analysis
,
1996
.
[4]
Volodymyr Vasyutynskyy,et al.
A Monitor Tool for CIM Systems
,
2001
.
[5]
Johan de Kleer,et al.
Fundamentals of model-based diagnosis
,
2003
.
[6]
James Cannady,et al.
Artificial Neural Networks for Misuse Detection
,
1998
.
[7]
Volodymyr Vasyutynskyy,et al.
Architecture and Data Model for Monitoring of Distributed Automation Systems
,
2004
.
[8]
Chris Price.
Computer-Based Diagnostic Systems: Computer-Based Troubleshooting
,
1999
.
[9]
Konrad Slind,et al.
Monitoring distributed systems
,
1987,
TOCS.