Mathematical Modeling and Numerical Solving of the Optimal Two-Impulse Rendezvous Problem
Zhi-kun SHE,Bai XUE,Yuan-liang CONG,Tie-gang LIU,Zhi-ming ZHENG
DOI: https://doi.org/10.3873/j.issn.1000-1328.2010.01.025
2010-01-01
Abstract:Technology of space rendezvous is one of the key technologies of space operations. Based on the Universal Variable's method, in this paper, we study the optimal two-impulse rendezvous problem of two coplanar orbits. Specifically, we firstly review the Lambert two-impulse rendezvous problem, which can be solved by the classical Gauss method based on the Universal Variable's method, discuss the optimal Lambert two-impulse rendezvous problem with the assumption that the rendezvous time and the beginning and ending positions are all fixed, and then build a corresponding two-dimensional optimal mathematical model. Secondly, for further energy saving, we extend this optimal mathematical model by introducing two additional time variables to divide the given fixed time period into three distinct stages (i.e., initial coasting stage, transfer orbit period and finial coasting stage) for two specific problems: one is the optimal two-impulse rendezvous problem of coplanar circular orbits and the other is the optimal two-impulse rendezvous problem of coplanar elliptical orbits with the same arch line, and build general four-dimensional optimal mathematical models for these two problems, respectively. Because of the existence of singular points in the optimal mathematical models, by taking advantage of the continuity of the object functions within the neighborhood of singular points and by introducing a sufficiently small positive constant to approximate the zero, we add a new constraint condition to remove singular points and obtain approximate optimal mathematical models. Lastly, we solve these approximate optimal mathematical models using Lingo, which is a mathematical software tool for solving non-linear programming, and get the global optimal numerical solution to the two-impulse rendezvous problems. The numerical results of four examples show that our approximate optimal mathematical models are feasible and that we can get a better solution by dividing the given time period.