基于多目标进化算法的侦察星座优化方法研究

Research on optimization method of reconnaissance constellation based on multi-objective evolutionary algorithm

为进一步解决侦察星座优化问题,加强天基侦察体系的建设,对当前侦察星座优化方法进行研究与综述。侦察星座的设计与优化具有多参数、多目标、非线性、不连续等特点,是一种典型的多目标优化问题。多目标进化算法是一类以种群为基础的、根据预定启发式规则在决策空间内进行概率搜索的生物智能算法,无需问题满足连续性、可微性等条件,能够在有限的搜索次数内得到一组靠近Pareto前沿的解集,能够有效解决多目标优化问题,可用于侦察星座的优化。论述了侦察星座优化模型的构建、多目标进化算法的分类及优缺点,并对多目标进化算法在侦察星座优化中的应用进行了分析,给出了基于多目标进化算法的侦察星座优化的发展方向。

In order to further solve the optimization problem of reconnaissance constellation and strengthen the construction of space-based reconnaissance system,the current optimization methods of reconnaissance constellation are studied and summarized.The design and optimization of reconnaissance constellation has the characteristics of multi-parameter,multi-objective,nonlinear and discontinuous,which is a typical multi-objective optimization problem.Multi-objective evolutionary algorithm is a kind of biological intelligence algorithm based on population,which carries out probabilistic search in decision space according to predetermined heuristic rules.It does not need continuity,differentiability and other conditions of the problem,and can get a group of solution sets close to Pareto frontier within a limited number of search times.It can effectively solve the multi-objective optimization problem and has been widely used in the multi-objective optimization of reconnaissance constellation.This paper discusses the construction of the reconnaissance constellation optimization model and the classification,advantages and disadvantages of the multi-objective evolutionary algorithm,discusses the application of the multi-objective evolutionary algorithm in the reconnaissance constellation optimization,and points out the development direction of the reconnaissance constellation optimization based on the multi-objective evolutionary algorithm.