Increasing the Efficiency and Value of Tradespace Analysis Through Machine Learning Methods

For years, systems engineers have used tradespace exploration as an analytical method for exploring the potential solution space of design and architecture alternatives. Methods such as Multi-Attribute Tradespace Exploration, Epoch Era Analysis, and Technology Progression-Informed Tradespace Analysis continue to expand the scope and robustness of the systems engineering process to make more informed decisions. However, as these processes increase in capability, so does their complexity and computational times. This also leads to crowded solution spaces that may be hard to comprehend by visual inspection or without substantial further analysis and filtering. To address these issues, this paper shows how Machine Learning techniques can quickly identify patterns in a complex tradespace with fewer design points which can help systems engineers find the driving factors to any architectural or design decision to best meet stakeholders' needs.