基于滑动平均订单预测的半导体生产库存模型  

Production and inventory control model of semiconductor manufacturing based on moving average method

作　　者：刘飏[1] 戴敏[1] 张志胜[1] 史金飞[1] 

机构地区：[1]东南大学机械工程学院,江苏南京211189

出　　处：《计算机集成制造系统》2010年第2期324-330,共7页Computer Integrated Manufacturing Systems

基　　金：机械系统与振动国家重点实验室开放基金资助项目( MSV200908);江苏省科技支撑计划资助项目(BE2008081);江苏省"青蓝工程"资助项目(2008-30-1)~~

摘　　要：为满足变化莫测的市场需求与半导体制造企业复杂的制造过程,针对半导体生产企业的生产时间不确定性和订单到达不确定性,分别使用滑动平均法与指数平滑法建立客户订单预测模型,提出了基于这两种客户需求预测模型的离散反馈状态生产库存控制系统模型;讨论了这两种模型系统有界输入有界输出的稳定性,并给出了基于这两种模型的系统达到有界输入有界输出稳定的参数限制条件。使用某半导体制造企业的订单生产数据,对两种预测模型进行仿真运算,比较了这两种模型下系统生产库存控制的具体效果。仿真结果显示,两种模型都可以满足客户订单的需求。通过检验表明,滑动平均法预测模型能使半导体企业生产库存控制获得小于3%的牛鞭效应,优于指数平滑法预测模型的结果。To satisfy changing market demands and complicated manufacturing process in semiconductor manufacturing enterprises,aiming at the uncertainties of the semiconductor manufacturing process and order arrival,Moving Average method （MA） and Exponential Smoothing （ES） were used to construct customer order prediction model respectively. Based on these two models,a production inventory control system model under discrete feedback status was put forward. Furthermore,the stability of the two different prediction models were analyzed,and the general stability criterion of the order policy parameters for a system,which was Bound Input,Bound Output （BIBO）,with any lead time was given. Finally,based on the practical data taken from a semiconductor manufacturing factory,the effectiveness of these two prediction models were verified. Results showed that the MA prediction model could achieve less than 3% bullwhip effect in this factory,which was better than ES prediction model.

关 键 词：库存控制 滑动平均法 指数平滑法 预测模型 半导体生产 

分 类 号：TP278[自动化与计算机技术—检测技术与自动化装置]