多主体目标优化的动态合作博弈方法

Dynamic-cooperative-game approach to multi-agent objective optimization

通过多目标优化和动态合作博弈理论,定义了联盟中多主体目标优化问题,提出了能够适应动态环境的基于合作博弈的多主体目标优化模型。该模型的组成一方面能够利用主体的协作能力,另一方面又能够充分考虑动态联盟的特征,适合大规模网络中多主体协作,避免模型中主体理性和团体理性的冲突。基于所提出的多主体目标优化模型,设计了一种联盟效用分配算法。仿真实验表明,联盟效用分配算法能够使多主体根据最优共识原则,分配各方的合作效用,从而达到多赢的帕累托最优局面。

With multi-objective optimization technology and dynamic cooperative game theory, this paper introduced a muhiagent objective optimization model, which could adapt to dynamic environments. The model could make use of the cooperative ability of the multi-agent well and could consider dynamic coalition characteristic fully. This model was suit for the large scale complex task agent cooperation and could avoid the conflict between individual object and group object. Designed a coalition utility allocation algorithm based on the multi-agent objective optimization problem. The results of emulation test show that the coalition utility allocation algorithm can achieve a multi-win Pareto-optimal outcome, which make the coalition tending to be more stable.