Ph.D. Proposal: Esma Karagoz

Fri Mar 12 2021 02:00 PM to 03:00 PM
"A Decision Advisor Framework for Model-Based Systems Engineering"

You're invited to attend


Ph.D. Proposal




Esma Karagoz

(Advisor: Prof. Mavris)


"A Decision Advisor Framework for
Model-Based Systems Engineering"


Friday, March 12 at 2 p.m.


Design and development of complex engineered systems in the aerospace industry have been challenged by (1) reduced development times and budgets to fulfill economic and strategic goals, (2) more demanding quality and performance requirements and higher degrees of maturity to meet environmental, industrial and regulatory standards, (3) increased interdependencies between system elements, and (4) unexpected failures (often due to emergent behaviors) causing additional efforts. Due to these challenges, it is often difficult to accurately predict the behavior of the system or to guarantee that the outcomes of a project align with its initial intent. To help mitigate these issues, informed decisions need to be made regularly throughout the system development. This in turn requires that the necessary knowledge be available in the early phases of design, which is usually needed in later phases.

Model-Based Systems Engineering provides the means to manage the challenges of system complexity through the use of models. It also standardizes systems engineering knowledge to manage significant, unpredictable changes throughout the life-cycle of a product and to facilitate collaboration within multidisciplinary teams. It provides a ``single source of truth'' for the workflow orchestration, which requires cross-domain data interoperability. This has been achieved through the formal knowledge representation techniques such as ontologies and semantic web technologies. Current applications use this knowledge as an information endpoint with consistency and completeness checks enabled by formal reasoning. However, this knowledge base can also be used to make predictions about the system. Hence, this work aims to investigate how the evolving knowledge base is managed, and how the available knowledge can be leveraged for better-informed decisions.

A methodology is proposed to (1) identify a decision support system allowing formal knowledge representation and computational techniques, (2) develop a framework utilizing available structured domain knowledge to aid systems engineers when making decisions, (3) identify missing information in the knowledge base for more effective decision-making, and (4) leverage the design knowledge for more accurate and explainable decisions. In this methodology, an initial knowledge base is built by utilizing OpenMBEE, ontologies and semantic web technologies. Then, deep graph neural networks are used to identify whether there is any missing information in the knowledge graphs. After securing a complete knowledge base, reinforcement learning is used for knowledge graph reasoning, which provides recommendations along with the explanation paths behind those, to aid systems engineers when making decisions.

The expected contributions of this research to the aerospace industry are (1) aiding systems engineers when making design decisions in an MBSE environment, (2) providing rationale behind design decisions, (3) improving early identification of unexpected failures causing additional efforts, (4) enabling systems engineers to capture previous knowledge and experiences, and (5) supporting the efficient training of novice engineers through the established knowledge base.



  • Prof. Dimitri Mavris – School of Aerospace Engineering (advisor)
  • Prof. Daniel Schrage – School of Aerospace Engineering
  • Prof. Chao Zhang – School of Computational Science and Engineering
  •  Dr. Olivia Fischer – School of Aerospace Engineering