Reference: Larsson, J. E. A Toolbox for Fast Model-Based Diagnosis. Knowledge Systems Laboratory, January, 1995.
Abstract: This paper describes the MFM Toolbox, a program which implements three model-based diagnostic algorithms: measurement validation, alarm analysis, and fault diagnosis. The algorithms are based on explicit modeling of goals and functions using multilelvel flow modeling concepts, and they are well-suited for problems where hard real-time behavior, speed, precise worst-case time information, and integration with conventional software are needed. The program provides the user with data structures, an intelligent graphical editor, and a model data file system using a general ASCII representation. It is written in C and runs on Macintosh systems. The graphical editor and the MFM models make it easy to build a knowledge database, and the simplicity, efficiency, and portability of the algorithms allow the constructed database to be used in conventional monitoring and control systems.
Full paper available as ps.