Model-transformation-based computational design synthesis for mission architecture optimization

• Published in: 2017 IEEE Aerospace Conference pp. 1 - 15
• Main Authors: Herzig, Sebastian J. I., Mandutianu, Sanda, Hongman Kim, Hernandez, Sonia, Imken, Travis
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.03.2017
• Subjects: Analytical models; Computer architecture; Mathematical model; Space missions; Space vehicles
• DOI: 10.1109/AERO.2017.7943953

In this paper, a model-based approach to exploring the trade space of multi-spacecraft mission architectures is introduced. Missions involving multiple spacecraft are inherently more complex to design than traditional single spacecraft missions. This is particularly true for fractionated mission concepts, where spacecraft have diverse roles and distributed responsibilities, and fulfilling mission goals requires communication and collaboration. In practice, this complexity and the lack of computational models and tools limit design teams to consider small design spaces and force them to quickly fixate on a single mission design. However, design fixation at such early stages often leads to sub-optimal designs. Towards overcoming this limitation, we propose an automated approach to exploring and visualizing the trade space of multi-spacecraft mission architectures that aids users in decision making, and provides a basis for identifying solutions that are Pareto-optimal with respect to user-defined science requirements, technical and resource constraints, and mission objectives. Central to our approach is the automated synthesis and analysis of mission architecture alternatives from a set of user-provided functional requirements and mission goals, as well as a library of spacecraft components and analysis models. Design rules (synthesis knowledge) are provided in the form of model-transformation rules. Sequences of endogenous model transformations are applied in-place to produce sets of candidate solutions, thereby effectively searching the design space. The search process is guided by the specified objectives, and is implemented using evolutionary algorithms. We demonstrate our approach to architectural optimization using a simplified radio interferometry mission design as a case study. We conclude that using the proposed approach, the number and diversity of candidate mission architectures available for consideration can be significantly increased. Furthermore, the automated synthesis and evaluation of mission architectures can lead to emergent and non-intuitive solutions to be discovered.