Partitioning large diagnostic problems

Abstract

A basic approach to modeling large diagnostic problems, such as Space Station Freedom or the ATF avionics, is presented. For this approach, the authors present three techniques for developing submodels to create an interacting network of submodels of reasonable size. Specifically, they discuss: partitioning by articulation points; partitioning by logical cut points; and partitioning at arbitrary points. These techniques have been developed as part of a total integrated diagnostics package that includes testability analysis, design for testability, and a full range of diagnostic capabilities, including built-in test, embedded diagnostics, automatic tests and portable maintenance aids.<>