基于模糊神经网络的半导体生产线重调度策略优化  

FUZZY NEURAL NETWORKS BASED RESCHEDULING STRATEGIES OPTIMIZATION OF SEMICONDUCTOR FABRICATION LINE

作　　者：张洁[1] 翟文彬[1] 严隽琪[1] 马登哲[1] 

机构地区：[1]上海交通大学计算机集成制造研究所,上海200030

出　　处：《机械工程学报》2005年第10期75-79,共5页Journal of Mechanical Engineering

基　　金：国家自然科学基金资助项目(50375098)。

摘　　要：针对缺乏半导体生产线重调度策略优化方法的研究现状,提出了基于模糊神经网络的半导体生产线重调度策略优化技术。将重调度策略划分为半导体生产线、设备组和设备重调度层次,利用仿真评价确定优化的重调度策略,并获得样本数据。通过对模糊神经网络训练,建立干扰和半导体生产线状态等输入参数与优化的重调度策略输出之间的映射关系。以上海某125mm晶圆生产线为例,结果表明了该重调度策略优化方法的有效性。For the lack of rescheduling optimization approach for semiconductor fabrication line status in quo, a novel fuzzy neural networks based algorithm for rescheduling strategies optimization of semiconductor fabrication line is proposed. The rescheduling strategies are divided into line layer, machine group layer and machine layer. The samples for FNN are setup by simulation evaluation. The relation between the input of FNN, such as disturbance, system state parameters, and output of FNN, optimal rescheduling strategy, is built by FNN. With an illustrative example of a 152 mm semiconductor fabrication line in shanghai, the lineal regression between FNN output and targets demonstrates the effectiveness of the approach.

关 键 词：半导体生产线 重调度 模糊神经网络 

分 类 号：TH166[机械工程—机械制造及自动化]