多代理信息融合算法

Multi-agent Information Fusion Algorithm

提出了一种基于模糊数的多代理信息融合算法。多代理系统中,自身代理为实现自身利益,有时提供非真实的诱导信息影响决策。通过分析信息融合算子对诱导信息的响应,扩充了简洁OWA(n-OWA)为信息融合算子以消除诱导信息的影响。同时,引进后续惩罚因子,降低该代理在后续阶段的作用来惩罚提供诱导信息代理。

Based on fuzzy set theory, this paper presents a multi-agent information fusion algorithm. In multi-agent systems, agents may provide beguiling information to affect groups decision. By analyzing commonly used information fusion operators, this paper extends the neat OWA and shows the extended neat OWA discourages beguiling information. A penalty factor is introduced to punish those agents in immediate successive phase for providing beguiling information. By lowering agents effect in current and immediate successive phases, extended neat OWA discourages agents to provide beguiling information.