基于遗传算法的多Agent工程索赔谈判模型

A GA-Agents model of construction claim negotiation

分析了已有基于贝叶斯学习规则的索赔谈判多agent系统,针对该系统的一些不足,提出基于遗传算法的多agent(以下称为GA-Agents)工程索赔谈判模型。模型采用“谈判门限序列”的概念给出索赔双方的谈判策略,设计了基于承包商Agent,业主Agent和运算Agent的操作流程,并利用遗传算法来不断优化索赔双方的谈判门限序列,使得索赔谈判能够在短时间内达成协议,为解决索赔谈判中存在的低效率、费时费力等问题提出了一个新思路。

By analyzing the existing multi - agent system of construction claim negotiation based on Bayes Rule and pointing out some insufficient aspects of that system, this paper puts forward a multi - agent model of construction claim negotiation based on genetic algorithm （GA- Agents）. In the meantime, it shows the negotiation strategy by using the concept of “sequence of negotiation offer threshold”, designs the working flow chart based on the contractor＇ s agent, owner＇s agent and calculation agent, and also provides a improving algorithm structure of the genetic operation in the flow chart to optimize the sequence of negotiation offer threshold. Furthermore, it makes the claim negotiation get an agreement in a short time. Thus, through this model, this paper suggests a new way to resolve the problems of low-efficiency, high labour costing and time-consuming, which exists in construction claim negotiation.