基于遗传模拟退火算法的空战编队优化研究

发展了一种用遗传算法优化大规模空战编队战术的方法，并用模拟退火算法对其进行了改进，使得收敛性能大大提高。该方法是借鉴层级编队思想，对多机编队实行层级编码，在一定的空战模型的基础上，用遗传算法优化大规模空战编队战术，最终得到一个对给定编队作战的最优编队，并用两个16机编队作战的算例验证了该方法的有效性。

On Optimizing Large-Scale Air-Combat Formation with Simulated-Annealing GA (Genetic Algorithm)

By Large scale air combat formation we mean a formation of eight fighters or more. The stochastic GA proposed by Krishnakumar et al 3] cannot solve the problem of premature convergence when applied to optimizing large scale air combat formation 1,2] . We propose simulated annealing GA to overcome such premature convergence. Section 1 briefly reviews essentials of modeling of large scale air combat formation and includes three topics: (1) the method and tactics of air combat formation (subsection 1.1); (2) the distribution of targets (subsection 1.2); (3) killing probability (subsection 1.3). Section 2 is most important in this paper. Subsection 2.1 explains the hierarchical method for encoding formation. Subsection 2.2 gives the key idea for evaluating formation: taking the air combat formation as the design variable, we employ fighter casualty statistics to evaluate BVR (beyond visual range) combat outcome and the effectiveness of large scale air combat formation. Section 3 gives two numerical examples. The results of these two examples, shown in Figs.4, 5, and 6, and listed in Tables 2, 3, and 4, show preliminarily the effectiveness of our simulated annealing GA.