System architecture optimization: an example application to space mission planning

Abstract Space mission planning involves coupled architecture decisions that non-trivially influence system-level performance metrics such as weight, power usage, and scientific value. We present an exemplary space mission planning problem involving several mission-level and spacecraft-level choices, and optimizing for the conflicting objectives of system mass and scientific value. The problem is solved using System Architecture Optimization (SAO): a technique where numerical optimization algorithms are used to explore an architecture design space and find a Pareto front of optimal architectures. The architecture design space is modeled using the Architecture Design Space Graph (ADSG) implemented in the ADORE editing and optimization tool. The design space model includes function-component allocation choices, component-level design variables, and system-level objectives and constraints to optimize for. Evaluation code is implemented in Python and linked to the design space model using class factories. The design space is explored using NSGA-II, a multi-objective evolutionary algorithm, resulting in a Pareto front trading-off system mass and total experiment duration.