CAJ | 학술논문

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。
망락계화자원균형속우조합우화문제，위료능쾌속유효지구해차류문제，제출료일충다지능체포곡조산법。침대표준포곡조산법결핍신식공향적결함，장다지능체계통인입포곡조산법중。다지능체적린역경쟁합작산자실현지능체간신식적교류，가쾌산법수렴속도；변이산자확대수색범위증가충군다양성；자학습산자제고국부심우적능력；포곡조산법적Levy비행진화궤제능유효지도출국부최우실현전국수렴。실례방진결과증실료，여기타산법상비다지능체포곡조산법능경유효지구해망락계화자원균형우화문제。
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 & coopera-tion 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.