Abstract: To solve the problem of task allocation in networked manufacturing process,a semi-autonomous multi-agent task allocation method was proposed by combining the mental model with the extended contract net,which included six mental parameters in two categories.On this basis,a three-step process of extended contract net was proposed,which included the bidder optimization,the tender offer and the multi-Agent task allocation.The Agent set of optimized bidding was calculated with tender fitness,and the tender price was obtained with task cost calculation.Through comprehensive assessment of price,collaboration and distance,a task allocation optimization model to minimize the total cost and its solving algorithm were presented.The effectiveness of the proposed method was verified by the case of raw materials purchasing optimization.The result showed that the proposed method could eliminate the defects of high communication costs,reflect the subjective desire of task candidate and provide effective support for task allocation in networked manufacturing.

Key words: multi-Agent, networked manufacturing, task allocation, mental model, extended contract net

摘要： 为解决网络化制造背景下的任务优化分配难题,提出一种将心智模型与扩展合同网机制结合的半自治多Agent任务分配方法。所构建的心智模型包括两类共六种心智参数。以此为基础,扩展了合同网机制包括发标优选、竞标报价与多Agent任务分配三个过程,通过发标适合度计算优化竞标Agent集;通过任务开销计算得到竞标报价;通过价格、协作、距离三个方面的成本评估,以总成本最小为目标建立任务分配优化模型与求解算法。采用企业原材料采购优化案例验证了该任务分配方法的有效性。结果表明,该方法消除了传统任务分配通信量大的缺陷,同时反映了任务候选者的主观意愿,为网络化制造条件下的任务分配提供了有效支持。

关键词: 多智能体, 网络化制造, 任务分配, 心智模型, 扩展合同网