Simultaneous Prediction for Multiple Key Performance Indicators in Semiconductor Wafer Fabrication  

作　　者：CAO Zhengcai LIU Xuelian HAO Jinghua LIU Min 

机构地区：[1]College of Information Science and Technology,Beijing University of Chemical Technology [2]Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education,Jilin University [3]Department of Automation,Tsinghua University

出　　处：《Chinese Journal of Electronics》2016年第6期1159-1165,共7页电子学报（英文版）

基　　金：supported by the National Natural Science Foundation of China(No.61104172,No.51375038);the Doctoral Fund of Ministry of Education of China(No.20130010110009);Beijing Municipal Natural Science Foundation(No.4162046);the Open Research Project from SKLMCCS(No.20120104);the Open Project Program of Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education,Jilin University(No.93K172014K05)

摘　　要：The prediction and key factors identification for lot Cycle time(CT) and Equipment utilization(EU) which remain the Key performance indicators(KPI)are vital for multi-objective optimization in semiconductor manufacturing industry. This paper proposes a prediction methodology which predicts CT and EU simultaneously and identifies their key factors. Bayesian neural network(BNN) is used to establish the simultaneous prediction model for Multiple key performance indicators(MKPI),and Bayes theorem is key solution in model complexity controlling. The closed-loop structure is built to keep the stability of MKPI prediction model and the weight analysis method is the basis of identifying the key factors for CT and EU. Compared with Artificial neural network(ANN)and Selective naive Bayesian classifier(SNBC), the simulation results of the prediction method of BNN are proved to be more feasible and effective. The prediction accuracy of BNN has been obviously improved than ANN and SNBC.The prediction and key factors identification for lot Cycle time（CT） and Equipment utilization（EU） which remain the Key performance indicators（KPI）are vital for multi-objective optimization in semiconductor manufacturing industry. This paper proposes a prediction methodology which predicts CT and EU simultaneously and identifies their key factors. Bayesian neural network（BNN） is used to establish the simultaneous prediction model for Multiple key performance indicators（MKPI）,and Bayes theorem is key solution in model complexity controlling. The closed-loop structure is built to keep the stability of MKPI prediction model and the weight analysis method is the basis of identifying the key factors for CT and EU. Compared with Artificial neural network（ANN）and Selective naive Bayesian classifier（SNBC）, the simulation results of the prediction method of BNN are proved to be more feasible and effective. The prediction accuracy of BNN has been obviously improved than ANN and SNBC.

关 键 词：Prediction for MKPI Cycle time Equipment utilization BNN Key factors identification 

分 类 号：TN305[电子电信—物理电子学]