Optimization Design of Reconnaissance Constellation Based on TOPSIS-MOPSO
刘亚丽 1熊伟 1韩驰 1熊明晖 1刘正1
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 航天工程大学复杂电子系统仿真实验室,北京 101416
折叠
摘要
侦察星座优化是天基信息体系建设的关键问题.为弥补以往研究大多只采用少量性能指标进行侦察星座优化的不足,提出了一种综合考虑5项性能指标的侦察星座优化模型.在解算优化模型过程中,为解决传统基于Pareto支配的进化算法出现的选择压力与多样性不足的问题,提出了TOPSIS-MOPSO(Technique for Order Preference by Similarity to an Ideal Solution-Multi-Objective Particle Swarm Optimization)算法,将多属性决策领域的TOPSIS引入进化算法中,并与SPD(Strengthened Pareto Dominate)相结合,得到一种能够同时增强种群收敛性与多样性的环境选择策略.提出了基于Harmonic距离的全局最优粒子选择策略,加快种群收敛速度,保护种群多样性;提出了自适应进化算子选择策略,帮助算法摆脱局部最优解.将TOPSIS-MOPSO算法应用在侦察星座优化问题上,并与MOPSO、DGEA、AR-MOEA 3种经典方法进行实验对比分析,实验结果显示,所提算法比其他3种算法在Δ*、IGD和HV上的最优指标值分别提升了19.76％、89.07％和28.2％.
Abstract
The optimization of reconnaissance constellation is a key issue in the construction of space-based information system. In order to make up for the shortcomings of existing studies in which only a few performance indicators are used to optimize the reconnaissance constellation,an optimization model which comprehensively considers five performance indicators is presented.TOPSIS-MOPSO algorithm is proposed to solve the problem of insufficient selection pressure and diversity in the traditional Pareto-dominated evolutionary algorithm.The Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) in the field of multi-attribute decision making is introduced into evolutionary algorithm and combined with Strengthened Pareto Dominate (SPD). An environmental selection strategy that can enhance both convergence and diversity of population is obtained. A global optimal particle selection strategy based on Harmonic distance is proposed to accelerate population convergence and protect the population diversity. The adaptive evolutionary operator selection strategy is proposed to help the algorithm get rid of the local optimal solution. Finally,TOPSIS-MOPSO algorithm is applied to the reconnaissance constellation optimization problem and is compared and analyzed with MOPSO,DGEA and AR-MOEA through experiment.The results show that,compared with thoses of other three algorithms,the optimal index values of Δ*,IGD and HV are improved by 19.76％,89.07％ and 28.2％ respectively.
维普
万方数据