Automated System Architecture Generation

DoD and its industrial base are moving from a document centric system engineering approach to Model Based System Engineering (MBSE) Extensive effort is required to manually generate architecture models for legacy systems defined in disparate sources. An efficient, cost-effective, automated approach is needed to generate system architecture models for legacy systems from all its disparate sources of information (e.g. specifications, spreadsheets, and automated tools). Trident proposes to develop a novel Automated System Architecture Generation (ASAG) tool that uses machine learning (ML)/artificial intelligence (AI) models to automate the development of a system architecture models, in SysML format from disparate sources, integrates SysML models together to provide traceability across a mission, and identifies the impacts of a baseline change across the models. A multilayer, time independent ML model will be used to automate the capture of dissimilarly formatted source data, normalize the data into a common format, resolve conflicts between sources, and fill in missing data to ensure a SysML-compliant model after all sources are integrated. In the Phase I, Trident performed research and analysis to design the ASAG system architecture. Leveraging experience in developing two related tool suites, Trident demonstrated the feasibility of the ASAG tool within a mature SysML analysis framework with baseline impact analysis capability. In the Phase II, Trident will develop an initial ASAG prototype tool for evaluation by MDA’s digital engineering team. The ASAG results will be evaluated against a manually developed system architecture for both accuracy and time saved. Feedback and lessons learned in Phase II will be used to develop a final ASAG prototype tool in Phase III for assimilation into MDA’s digital engineering environment. Approved for Public Release | 22-MDA-11340 (16 Dec 22)