多智能体布谷鸟算法的网络计划资源均衡优化  

Multi-agent cuckoo search algorithm for resource leveling problem of network planning

作　　者：宋玉坚 叶春明[1] 黄佐钘[1,2] 

机构地区：[1]上海理工大学管理学院,上海200093 [2]上海期货交易所,上海200122

出　　处：《计算机工程与应用》2015年第15期56-61,共6页Computer Engineering and Applications

基　　金：国家自然科学基金(No.71271138);上海市教育委员会科研创新项目(No.12ZS133);上海市一流学科建设项目(No.S1201YLXK)

摘　　要：网络计划资源均衡属于组合优化问题,为了能快速有效地求解此类问题,提出了一种多智能体布谷鸟算法。针对标准布谷鸟算法缺乏信息共享的缺陷,将多智能体系统引入布谷鸟算法中。多智能体的邻域竞争合作算子实现智能体间信息的交流,加快算法收敛速度;变异算子扩大搜索范围增加种群多样性;自学习算子提高局部寻优的能力;布谷鸟算法的Levy飞行进化机制能有效地跳出局部最优实现全局收敛。实例仿真结果证实了,与其他算法相比多智能体布谷鸟算法能更有效地求解网络计划资源均衡优化问题。Resource leveling problem of network planning is a combinatorial optimization problem. For the purpose of solving it efficiently and effectively, this paper proposes a multi-agent cuckoo search algorithm. The multi-agent system is embedded to basic cuckoo search algorithm to counteract the lack of information exchange. The competition cooperation operation enhances agents＇ communication and boosts convergence speed. Mutation operation can search widely and maintain the diversity of the population. Self-study operation can exploit high quality solutions. Meanwhile, the Levy flight evolution mechanism can avoid trapping to local optima. The case study also indicates that the modified algorithm can solve the resource leveling problems more efficiently and effectively when compared with other algorithms.

关 键 词：资源均衡 多智能体布谷鸟算法 竞争合作算子 变异算子 自学习算子 Levy进化机制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]