动态连续蚁群系统及其在天基预警中的应用  

Dynamic Continuous Ant Colony Optimization and Its Application to Space-based Warning System

作　　者：胡崇海[1] 李一军[1] 姜维[1] 王铁军[1] 

机构地区：[1]哈尔滨工业大学管理学院,黑龙江哈尔滨150001

出　　处：《运筹与管理》2011年第2期89-96,共8页Operations Research and Management Science

基　　金：高等学校博士学科点专项科研基金资助课题(200802131048)

摘　　要：存在监控冲突的天基中段预警传感器调度优化是一个动态、高维、复杂多约束的非线性优化问题,其解空间的高维度与状态复杂性直接制约了智能优化算法的运用。本文以任务分解与任务复合优先权计算为基础,通过二级分离机制将解空间维度与状态复杂性降低至适于连续蚁群(continuous ant-colony optimization,CACO)处理的全局优化形态,构建出相应的优化子路径集.在此基础上,针对监控冲突导致的状态变化特性,从局部搜索递进与募集的角度提出适于传感器调度优化的MG-DCACO(double direction continuous ant-colony optimizationbased mass recruitment and group recruitment)算法,成功将智能优化算法应用于基于低轨星座的天基中段预警中.最后对算法的收敛性进行论证,并通过与已有规则调度算法的对比得出MG-DCACO算法可获得优于规则调度算法的全局最优解。The scheduling method of sensors on space-based warning in middle age is a dynamic,multi-dimensional,complex-constraints nonlinear optimization problem.Considering the monitoring conflict,it is nearly impossible to use intelligent optimization algorithms in this problem.On the basis of task decomposition and task multiplex priority,by means of second-stage separating,this paper reduces the multi-dimensional and complex-constraints to a suitable area.Then,through the angles of monitoring conflict,area searching and collecting,the author puts forward a MG-DCACO（double direction continuous ant-colony optimization based mass recruitment and group recruitment）algorithm which can be used in sensors scheduling.At last,it is proved that,the MG-DCACO is convergence and outperforming the other algorithms of sensors scheduling.

关 键 词：管理科学与工程 蚁群系统 动态优化 任务分解 天基预警 

分 类 号：O221.2[理学—运筹学与控制论] TP181[理学—数学]