不确定条件下基于Multi-agent的模糊预测优化

Fuzzy Forecasting Optimization Based on Multi-agent under Uncertainty

为解决多层分销链中预测订货忽略当前库存状况和决策者参与的问题,运用Multi-agent和模糊理论构建了不确定条件下的协同预测订货模型,采用模糊数对分销链中的不确定性需求和单位成本参数进行了表述并建立了模糊规则库用于分销链的协商谈判。应用遗传算法和解模糊运算求解分销链中整体或局部优化条件下的目标函数,并给出了公司间冲突的模糊协商算法。仿真实验表明,不确定条件下的分销链预测订货量没有出现信息放大,模糊协商谈判获取了符合双方利益的预测订货成本。与确定性条件下的计算相比,运用模糊理论解决分销链中基于Multi-agent的预测订货是合理可行的。

To deal with current inventory variation and decision makers＇ participation which was often ignored in forecasting of multi-echelon distribution chain, a forecasting model was constructed based on multi-agent and fuzzy theory under uncertainty. Uncertainty demand and unit cost parameter was described by using fuzzy number in distribution chain, and then fuzzy rule sets were proposed and used for negotiation. The fuzzy forecasting model about whole or local optimization cost was solved by genetic algorithm and anti-fiazzy calculation method, and then fuzzy negotiation algorithm was proposed to resolve some conflict between firms. The results show that the forecasting quantities are not enlarged in distribution chain, and the forecasting cost obtained by negotiation is in accordance with two parties＇ benefit. So the agent-based forecasting model by fuzzy theory solving is reasonable than one＇s under certainty.