基于改进合同网协议的半导体封装测试工厂启发式产能平衡优化

Ameliorated CNP Based Heuristic Capacity Equilibrium Optimization in Semiconductor Assembly and Test Factory

针对半导体封装测试工厂生产制造过程中生产计划分配的交互性问题,运用分布式多Agent技术,对生产分配过程进行模块化封装。在传统合同网协议基础上进行改进,嵌入"产能注册矩阵"的模式构建了多Agent协商策略模型,提出了智能体间招标—投标—中标的交互机制。对投标过程的产能平衡问题,运用启发式算法进行了优化。通过S企业生产计划实例分析和传统MRPⅡ方法投标结果对比,验证了改进的合同网协商模型和启发式算法对生产任务分配和产能平衡具有相对优良的投标完成率和相关系数矩阵。

To solve the interactive issues during semiconductor assembly and test production planning, the multi-agent technology was applied to separate the semiconductor assembly and test production planning into several modules. The traditional contract net protocol （CNP） was improved with ＂capacity registered matrix＂ to build the multi-agent negotiation model, and the announce-bid-award process between agents interaction was discussed. Subsequently, the heuristic algorithm was employed to optimize the capacity equilib- rium during bidding process. The presented method was implemented into S company production planning and compared with the tradi- tional MRP Ⅱ method. The results showed that the integrated ameliorated CNP and heuristic algorithm has the better response close rate and correlation coefficient matrix for the 13roduetion arranning arrangement and eanaeitv eauilibrium.