|Koji Adachi||Opt & Electronics Technology Laboratory,
Research & Technology Group
|Norikazu Yamada||same as above|
|Koki Uwatoko||same as above|
|Kaoru Yasukawa||same as above|
To provide efficient maintenance services for electrophotographic multifunction devices and reduce downtime, we have been developing a fault diagnostic system using Bayesian networks that can identify faults in matching the diagnostic level of skilled service engineers, and which can be implemented in multifunction devices. A Bayesian network is a causal network consisting of directed acyclic graphs with conditional probabilities for each node, and AI (Artificial Intelligence) technology that can represent actual problems under uncertainty as a probabilistic graphical model. This paper introduces a compact Bayesian network inference engine to realize an embedded diagnostic system, diagnostic modeling technologies for various faults, diagnostic model development support technologies to simplify development and maintenance, and an automatic diagnostic system for multifunction devices implemented as part of a diagnostic system.