Connectionists: CFP: Neural Networks and Data Mining for Fault Diagnosis and Failure Prognosis of Engineering Systems

[yilu.zhang@gm.com]

Call for papers:

               Neural Networks and Data Mining 
for Fault Diagnosis and Failure Prognosis of Engineering Systems

A Special Session of International Joint Conference on Neural Networks
               August 12-17, Orlando, Florida
                  http://www.ijcnn2007.org/
               (Submission deadline: 1/31/2007)

The increasing complexity of engineering systems and the increasing 
interactions among components boost the number of system states 
exponentially, which poses a big challenge to the diagnosis and 
prognosis of such systems. When designing a complex system, it is 
difficult to exhaust and understand all the operation states for 
failure mode and effects analysis (FMEA). The difficulties are 
compounded by the demand of short time-to-market for new systems. 
When maintaining a complex system, it is difficult to isolate and 
resolve all the failure modes. As a result, many cases are reported 
in service bays as ?no trouble found? or ?customer concern not 
duplicated.? It becomes very challenging to execute the growing 
condition-based maintenance (CBM) strategy when system prognosis 
and predictive maintenance are required. Effective and efficient 
diagnosis and prognosis technologies are called for across industries, 
especially for those safety-critical systems that require reliable 
and uninterrupted operations.

Among other technologies, data mining offers unique and promising 
potential to complex system diagnosis. Data mining is the process of 
automatically searching large volumes of data for patterns. In the 
system diagnosis and prognosis context, the patterns can be anything 
from hidden relationship, root cause, to degradation trend. In the 
design stage, data mining can gain insight into the hardware/software 
interactions. In the validation stage, data mining can spot the 
weakest link in the system. In the maintenance stage, data mining 
can identify shared symptoms for fault isolation and shared failure 
precursors for predictive maintenance. Overall, data mining serves 
as a cost/time effective way to reduce massive engineering data 
into diagnostic and prognostic knowledge. Its application spans over 
the whole system lifecycle from development, deployment, to service.

Data mining is not a new concept to the neural networks community. 
Neural networks is one of the widely used modeling and learning 
techniques for data mining. At the same time, data mining for 
complex system diagnosis and prognosis has its unique challenges 
in enterprise-level data collection, diagnostic knowledge discovery 
and management, and field deployment. This special session invites 
research papers from both academia and industry to identify the 
challenges, present successful practices, share lessons learned, 
and define roadmaps for further advance in the area of neural 
networks and data mining for complex system diagnosis and prognosis. 
Topics of interest include, but are not limited to

- Neural networks and data mining algorithms for system monitoring 
  and diagnosis
- Neural networks and data mining algorithms for system prognostics 
  and predictive maintenance
- Categorical and parametric data hybrid analysis for anomaly 
  detection and fault isolation
- Diagnostic data collection, cleaning, and management over system 
  lifecycle
- Fusion of diagnostic knowledge extracted from data and from domain 
  experts
- Evaluation of diagnostic knowledge derived from data
- Diagnostic data mining practice in system development and validation
- Diagnostic data mining practice in system deployment and maintenance
- Interactive data mining practice in diagnosis and prognosis algorithm 
  development
- Other enabling technologies for complex engineering system diagnosis 
  and prognosis

Detailed instructions for paper submission are available at 
http://www.ijcnn2007.org/. The deadline is January 31, 2007. Please 
select ?Neural Networks and Data Mining for Fault Diagnosis and Failure 
Prognosis of Engineering Systems? as the ?main research topic? in order 
to be considered as a submission to this special session. 

Special session organizers:

George Vachtsevanos, Professor
School of Electrical and Computer Engineering
Georgia Institute of Technology
Atlanta, Georgia
Email: gjv at ece.gatech.edu

Yilu Zhang, Senior Researcher
Electrical & Controls Integration Lab
GM R&D Center
Warren, Michigan
Email: yilu.zhang at gm.com

Feng (Fred) Xue, Information Scientist
Industrial Artificial Intelligence Lab
GE Global Research Center
Niskayuna, New York
Email: xue at research.ge.com

Weizhong Yan, Research Engineer 
Industrial Artificial Intelligence Lab
GE Global Research Center 
Niskayuna, New York
Email: yan at crd.ge.com


-----------
Yilu Zhang, Ph.D.
Senior Researcher
Electrical & Controls Integration Lab

GM R&D and Planning
General Motors Corporation
MC: 480-106-390
30500 Mound Road
Warren, MI 48090

Voice: (586) 986-4717
Fax: (586) 986-3003
yilu.zhang at gm.com