改进非支配排序遗传算法求解多目标Agent联盟

Multi-objective Agent Coalition Formation Based on Improved NSGA-Ⅱ Algorithm

采用典型多目标进化算法-NSGA-II对从任务角度进行抽象建模所得到的Agent联盟模型进行生成优化,并针对Agent联盟生成存在的特点,将Pareto最优概念与多目标优化相结合对NSGA-II算法进行改进,从而实现兼顾联盟收益、开销、时间约束等多个目标。仿真对比实验结果表明,算法运行一次可以获得多个Pareto最优解,为各个目标之间权衡分析提供了有效的工具,在满足性能要求下,可为联盟生成提供满足多个设计目标的全局优化方案,对联盟实际应用具有借鉴与应用价值。对联盟实际应用具有借鉴与应用价值。

A typical multi-objective evolutionary algorithm-NSGA-II was proposed for Agent Coalition model obtained by abstract model of the perspective of tasks. The NSGA-II algorithm was improved based on Agent Coalition Generation existing characteristics and combination of the concept of Pareto optimal and multi-objective optimization in order to achieve revenue, cost, time constraints and other goals balance Comparative simulation results show that the algorithm can be run one time to get more than one Pareto optimal solution, providing effective tools for the analysis of the various trade-offs between the objectives. It can be generated for Coalition to provide multiple design objectives of the global optimization program, and also has the practical application of the reference and application.