【摘要】 针对三维卫星舱布局优化问题(three-dimensional satellite module layout optimization problem,3DSMLOP),本文提出了一种具有双邻域的改进人工蜂群算法,并将其与多阶段求解策略结合,形成多阶段双邻域人工蜂群算法(multi-stage dual neighborhood artificial bee colony algorithm,MS-DABC).3DSMLOP是一个复杂的多约束耦合问题,其解空间是非连续、非线性、多模态的.MS-DABC将3DSMLOP分解为多个子系统,并根据优化目标在各个子系统之间的耦合关系,将优化过程分为两个阶段.第一阶段,针对无耦合的优化目标,各个子系统利用具有双邻域结构的改进人工蜂群算法独立进行优化.在第一阶段求得的最优解的基础上,第二阶段采用一般人工蜂群算法来优化子系统的旋转角度,利用各个子系统之间的协同旋转来解决耦合的优化目标.仿真实验结果表明,该算法求解复杂的卫星舱布局问题非常有效,在3DSMLOP算例求解上性能突出.更多还原

【Abstract】 This paper presents an improved artificial bee colony algorithm with dual neighborhood, and combines it with a multi-stage solving strategy to form a multi-stage dual neighborhood artificial bee colony algorithm(MS-DABC) for the three-dimensional satellite module layout optimization problem(3DSMLOP). 3DSMLOP is a complex multi-constrained coupling problem whose solution space is non-continuous, non-linear, multi-modal.MS-DABC decomposes 3DSMLOP into several subsystems, and divides the optimization process into two stages according to the coupling relationship on optimization goals between subsystems. In the first stage, each subsystem uses the dual neighborhood artificial bee colony algorithm(DABC) to optimize no-coupling optimization goals independently. Based on the optimal solution obtained in the first stage, the coupling optimization goals are solved by the collaborative rotation between the subsystems in the second stage. The rotation angles are optimized by the artificial bee colony algorithm. Numerical experiment results show that MS-DABC is very effective and has outstanding performance for solving the 3DSMLOP example.更多还原