Multi-Agent Case-Based Diagnosis in the
Aircraft Domain
Pascal Reuss
12
, Klaus-Dieter Althoff
12
, Alexander Hundt
1
, Wolfram Henkel
3
,
and Matthias Pfeiffer
3
1
German Research Center for Artificial Intelligence
Kaiserslautern, Germany
http://www.dfki.de
2
Institute of Computer Science, Intelligent Information Systems Lab
University of Hildesheim, Hildesheim, Germany
http://www.uni-hildesheim.de
3
Airbus
Kreetslag 10 21129 Hamburg, Germany
Abstract. Aircraft diagnosis is a highly complex topic. Many knowl-
edge sources are required and have to be integrated into a diagnosis
system. This paper describes the instantiation of a multi-agent system
for case-based aircraft diagnosis based on the SEASALT architecture.
This system will extend a existing rule-based diagnosis system, to make
use of the experience of occurred faults and their solutions. We describe
the agents within our diagnosis system, the planned diagnosis workflow
and the current status of the implementation. For the case-based agents,
we give an overview of the initial case structures and similarity measures.
In addition, we describe some challenges we have during the development
of the multi-agent system, especially during the knowledge modeling.
1 Introduction
An aircraft is a complex mechanism, consisting of many subsystems. Occurring
faults cannot be easily tracked to their root cause. A fault can be caused by
one system, by the interaction of several systems or by the communication line
between systems. Finding the root cause can be very time and resource con-
suming. Therefore the use of experience from successfully found and solved root
causes can be very helpful for aircraft diagnosis. This paper describes the in-
stantiation of a multi-agent system (MAS) based on the SEASALT architecture.
The MAS contains several Case-Based Reasoning (CBR) systems to store the
experience and provide this knowledge for diagnosis. In the next section, we
give an overview of the OMAHA project and the SEASALT architecture. Sec-
tion 2 contains related work with comparing our approach to other diagnosis
and multi-agent approaches. In Section 3 we describe the instantiation of the
SEASALT components for our MAS and describe the case-bases agents with the
case structure and similarity measures of the underlying CBR systems in more
detail. Section 4 gives a short summary of the paper and an outlook on future
work.
Copyright © 2015 for this paper by its authors. Copying permitted for private and
academic purposes. In Proceedings of the ICCBR 2015 Workshops. Frankfurt, Germany.
43