自然计算在求解网络优化部署问题中的比较  

Comparison of natural computations in solving network optimization deployment

作　　者：程子光[1] 冯亚军[1] 尹康银[1] 王海林[1] CHENG Ziguang;FENG Yajun;YIN Kangyin;WANG Hailin(Air Force EarlyWarning Academy,Wuhan 430019,China)

机构地区：[1]空军预警学院

出　　处：《空军预警学院学报》2019年第5期367-371,379,共6页Journal of Air Force Early Warning Academy

基　　金：空军军事理论研究资助项目(17KJ3C1-0089R)

摘　　要：针对现有的网络部署模型大多以区域覆盖率或网络寿命为优化目标,并假设监视区域为单一类型,不适用于求解多类型分区区域的网络部署问题,首先构建了与区域可部署度、区域重要度、区域协同度等因素相关的栅格化网络部署模型;然后运用遗传算法、蚁群算法、粒子群算法3种典型自然计算对该模型进行求解;最后分析了3种算法的收敛情况与解的特点,并探讨了关键参数变化对结果部署方案的影响.所得结论为自然计算的改进或融合提供了参考,为网络部署时的算法选取提供了依据.Considering that most of the existing network deployment models aim at the optimization goal of area coverage or network life,and assuming that the surveillance area is of a single type unsuitable for solving the network deployment of multi-type partition area,this paper firstly constructs a rasterized network deployment model related to such factors as regional deployment degree,regional importance degree,regional synergy degree and so on.And then the paper uses three typical natural calculations like genetic algorithm(GA),ant colony algorithm(ACO)and particle swarm optimization(PSO)to solve the model.Finally,the paper analyzes the convergence of the three algorithms and the characteristics of the solutions,and also discusses the effect of key parameter changes on the result deployment scheme.The conclusion provides a reference for the improvement or fusion of natural computation,and comes up with a basis for the selection of algorithm in network deployment.

关 键 词：自然计算 优化部署 遗传算法 蚁群算法 粒子群算法 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]