A collaborative optimization approach to improve the design and deployment of satellite constellations.

摘要

This thesis introduces a systematic, multivariable, multidisciplinary method for the conceptual design of satellite constellations. The system consisted of three separate, but coupled, contributing analyses. The configuration and orbit design module performed coverage analysis for different orbit parameters and constellation patterns. The spacecraft design tool estimated mass, power, and costs for the payload and spacecraft bus that satisfy the resolution and sensitivity requirements. The launch manifest model found the minimum launch cost strategy, to deploy the given constellation system to the specified orbit.; Collaborative Optimization (CO) has been previously implemented successfully as a design architecture for large-scale, highly-constrained multidisciplinary optimization problems related to aircraft and space vehicle studies. It is a distributed design architecture that allows its subsystems flexibility with regards to computing platforms and programming environment and, as its name suggests, many opportunities for collaboration. It is thus well suited to a team-oriented design environment, such as found in the constellation design process, and was implemented in this research.; Two problems were solved using the CO method related to the design and deployment of a space-based infrared system to provide early missile warning. Successful convergence of these problems proved the feasibility of the CO architecture for solving the satellite constellation design problem. Verification of the results was accomplished by also implementing a large All-at-Once (AAO) optimization.; This study further demonstrated several advantages of this approach over the standard practice used for designing satellite constellation systems. The CO method explored the design space more systematically and more extensively, improved subsystem flexibility, and its formulation was more scalable to growth in problem complexity. However, the intensive computational requirement of this method, even with automation and parallel processing of the subsystem tasks, reduced its competitiveness versus the current practice, given today's computing limitations.; This thesis also contributed to the current knowledge of the collaborative optimization. To date, CO has been used exclusively with gradient-based optimization scheme, specifically Sequential Quadratic Programming (SQP). This research demonstrated the feasibility of zero-order methods as both system and subsystem optimizers. Finally, with integer variables involved in the problem, CO's flexibility for handling mixed-discrete nonlinear problems was demonstrated.