Parameter-based representation for modeling complex systems. 2

For pt. 1 see Friedenthal, proc. NCOSE (1995). This paper describes a concept for a parameter-based representation (PER) to model complex systems. The representation combines features from both process and object-oriented representations, including IDEFO, data and control flow diagrams, entity relationship attribute (ERA) diagrams, and parameter dependency diagrams. PER applies and extends many concepts that are widely used in software engineering to system-level design. The parameter-based approach provides enhanced capability over existing representations because it represents the multiple attributes of a system-including the performance, physical, design, and process parameters-and explicitly defines the relationships between them. As part of the system design, the parameters are logically partitioned into input, output, and mechanism objects. The relationships between the parameters are captured by behavioral, structural, and class representations. Because PER can represent explicit mathematical relationships between system parameters, it can also be used to develop an executable model of the system. The PER approach assists the systems designer in performing sensitivity and trade-off analysis, assessing the impact of design decisions and changes on system performance, and maintaining design traceability.