Abstract: An automated diagnostic system uses Bayesian networks to diagnose a system. Knowledge acquisition is performed in preparation to diagnose the system. An issue to diagnose is identified. Causes of the issue are identified. Subcauses of the causes are identified. Diagnostic steps are identified. Diagnostic steps are matched to causes and subcauses. Probabilities for the causes and the subcauses identified are estimated. Probabilities for actions and questions set are estimated. Costs for actions and questions are estimated.
Type:
Grant
Filed:
January 11, 2001
Date of Patent:
April 12, 2005
Assignee:
Hewlett-Packard Development Compant, LP.
Inventors:
Claus Skaanning, Finn V. Jensen, Uffe Kjærulff, Paul A. Pelletier, Lasse Rostrup Jensen, Marilyn A. Parker, Janice L. Bogorad