CAJ | 학술논문

针对以往对算法邻域结构的设计过于注重局部搜索能力而忽略全局探索能力，不利于算法快速收敛的问题，基于传统方位思想设计一种基于方位邻域的随机动态信息互享差分进化算法。设计两种邻域：分别是东、西、南、北方向的固定邻域和东南、东北、西南、西北方向带偏角的随机动态邻域，前者侧重于局部搜索，后者侧重于不同局部间的信息互享，有利于全局探索，主要目的是平衡算法的局部全局搜索能力以提高其性能。结合该算法对虚拟网络映射问题进行研究。在通用标准函数上的仿真结果表明，改进算法具有很好的寻优性能。
침대이왕대산법린역결구적설계과우주중국부수색능력이홀략전국탐색능력，불리우산법쾌속수렴적문제，기우전통방위사상설계일충기우방위린역적수궤동태신식호향차분진화산법。설계량충린역：분별시동、서、남、북방향적고정린역화동남、동북、서남、서북방향대편각적수궤동태린역，전자측중우국부수색，후자측중우불동국부간적신식호향，유리우전국탐색，주요목적시평형산법적국부전국수색능력이제고기성능。결합해산법대허의망락영사문제진행연구。재통용표준함수상적방진결과표명，개진산법구유흔호적심우성능。
The existing algorithm neighborhood structure was designed to focus on the ability of local search more ,but neglecting the global exploring ability ,which is not conducive to rapid convergence algorithm .To solve the problem ,a kind of stochastic dynamic information sharing differential evolution algorithm based on range neighborhood was proposed .Two kinds of neighbor‐hoods were designed including stochastic dynamic neighborhood with angle East ,West ,South ,north ,and fixed neighborhood with southeast ,southwest ,northwest ,northeast .The former focused on the local search and the latter focused on the different local information sharing ,which was advantageous to global exploration .The main purpose of the design was balancing the ca‐pacity of local search and that of global search in the algorithm to improve its performance .The proposed algorithm was used in the research of virtual network mapping .The simulation on the standard function show that the improved algorithm has good op‐timization performance .