Abstract: To address the issue of high single satellite response latency and difficulty in meeting high timeliness task requirements for reusable orbital platforms, adopting multi platform collaborative deployment can effectively shorten response time. However, this type of constellation system presents significant challenges in terms of coverage mechanism, spatiotemporal constraints, and high-dimensional parameter optimization. To this end, a collaborative optimization design method for rapid-response constellations is proposed. This method deviates from the uniformly distributed orbital planes of traditional Walker constellations by establishing a flexible parametric model for hybrid constellations. The optimization dimensionality is reduced by refining orbital inclination, right ascension of ascending node spacing, and true anomaly spacing. By defining the global weighted average maximum revisit time as the core performance metric, the constellation parameter optimization problem is formulated as a Markov decision process. A proximal policy optimization algorithm integrated with a curiosity-driven exploration mechanism is designed to solve it. The results demonstrate that the proposed approach effectively enhances the uniformity of global coverage and responsiveness of the constellation, significantly reduces the system's maximum revisit time, and provides reliable technical support for the engineering design of returnable reusable orbital platform constellations.

Key words: Returnable orbital platform, Low earth orbit constellation, Constellation design, Proximal Policy Optimization (PPO), Curiosity exploration mechanism