|InterJournal Complex Systems, 1053
|Manuscript Number: |
Submission Date: 2004
|Architecting Systems Under Uncertainty with Object-Process Networks|
Architects of complex systems and products, such as an aircraft power system must routinely make tradeoff decisions given limited resources, incomplete information, and evolving stakeholder needs. Their decisions ultimately require a certain amount of subjective judgment, as a result of which system architecting involves not just science and engineering, but also art. By embedding Bayesian belief propagation algorithms in Object-Process Methodology (OPM), we show that architects of large-scale complex systems can better describe, quantify and communicate system attribute tradeoffs under uncertainty. This new framework, Object-Process Network (OPN), is instrumental in formulating and preserving the rationale behind tradeoff decisions, thereby offering a systematic approach to system-related decision making. Object Process Networks achieve the following objectives: 1. Graphically compose component-level knowledge to help visualize complex interrelationships among variables in a system, 2. Model the space of architectural options constrained by deterministic rules or conditional probability functions between system variables, and 3. Compare and select preferred architectural options under uncertainty. To demonstrate the use of OPN in complex system architecting, we have applied this framework to an architectural tradeoff case study of an aircraft power system. A simplified architectural OPN model and its model construction process are presented and discussed in this paper.
|Submit referee report/comment|