融合双重注意力机制与并行门控循环单元的晶圆加工周期预测方法

doi:10.3969/j.issn.1004-132X.2023.14.001

中国机械工程 ›› 2023, Vol. 34 ›› Issue (14): 1640-1646.DOI: 10.3969/j.issn.1004-132X.2023.14.001

融合双重注意力机制与并行门控循环单元的晶圆加工周期预测方法

戴佳斌1，2;张洁1;吴立辉3

1.东华大学人工智能研究院，上海，201620
2.东华大学信息科学与技术学院，上海，201620
3.上海应用技术大学机械工程学院，上海，201418

出版日期:2023-07-25 发布日期:2023-07-28
通讯作者: 吴立辉（通信作者），男，1981年生，副教授。研究方向为复杂制造系统调度、制造大数据分析、AGV调度。发表论文20余篇。E-mail：wulihui@sit.edu.cn。
作者简介:戴佳斌，男，1998年生，硕士研究生。研究方向为大数据。发表论文1篇。E-mail：2211967@mail.dhu.edu.cn。
基金资助:
国家重点研发计划（2022YFB3305003）；国家自然科学基金（U1704156）

A Wafer Cycle Processing Time Prediction Method Incorporating Double Attention Mechanism and Parallel GRU

DAI Jiabin1,2;ZHANG Jie1;WU Lihui3

1.Institute of Artificial Intelligence,Donghua University,Shanghai,201620
2.School of Information Science and Technology,Donghua University,Shanghai,201620
3.School of Mechanical Engineering,Shanghai Institute of Technology,Shanghai,201418

Online:2023-07-25 Published:2023-07-28

摘要/Abstract

摘要： 晶圆制造过程中的生产特征数据大规模、特征间复杂关联和特征样本强相关特性导致晶圆加工周期预测效率低与预测精度不高，因此提出了一种融合双重注意力机制与并行门控循环单元的晶圆加工周期预测方法。首先，利用Relief-F算法对生产特征数据进行降维处理；然后，基于模糊C均值算法对数据样本进行工艺相似性聚类，并设计并行门控循环单元网络来挖掘晶圆lot特征样本间的强相关特性；最后，设计了双重注意力机制来学习关键特征内部及特征与加工周期的复杂关联信息。实验结果表明，所提出方法能有效缩短预测训练时间，提高预测精度

关键词: 晶圆制造, 预测, 并行门控循环单元, 注意力机制

Abstract: Low efficiency and low prediction accuracy were caused by the large scale of production feature data, complex correlation among features, and strong correlation of feature samples in wafer fabrication processes, so a wafer processing cycle prediction method integrating double attention mechanism and parallel GRU was proposed. Firstly, Relief-F algorithm was used to reduce the dimensionality of production feature data. Secondly, a fuzzy C-mean algorithm was used to cluster the process similarity of data samples and design a parallel GRU network to explore the strong correlation among wafer feature samples. Finally, a double attention mechanism was designed to learn the complex correlation information within key features and among features and processing cycle. The experimental results show that the proposed method may effectively reduce the prediction training time and improve the prediction accuracy.

Key words: wafer fabrication, prediction, parallel gated recurrent unit (GRU), attention mechanism

中图分类号:

TP391

戴佳斌, 张洁, 吴立辉. 融合双重注意力机制与并行门控循环单元的晶圆加工周期预测方法[J]. 中国机械工程, 2023, 34(14): 1640-1646.

DAI Jiabin, ZHANG Jie, WU Lihui. A Wafer Cycle Processing Time Prediction Method Incorporating Double Attention Mechanism and Parallel GRU[J]. China Mechanical Engineering, 2023, 34(14): 1640-1646.

参考文献

［1］吕佑龙, 许鸿伟, 郑城，等. 基于混合式特征选择模型的晶圆允收测试关键参数识别方法［J］.中国机械工程, 2020, 31(16):1978-1984.
LYU Youlong, XU Hongwei, ZHENG Cheng, et al. Wafer Acceptance Test key Parameter Identification Based on Hybrid Feature Selection Method［J］. China Mechanical Engineering, 2020, 31(16):1978-1984.
［2］朱雪初, 乔非. 基于工业大数据的晶圆制造系统加工周期预测方法［J］. 计算机集成制造系统, 2017,23(10):2172-2179.
ZHU Xuechu, QIAO Fei. Process Cycle Prediction Method for Wafer Manufacturing System Based on Industrial Big Data［J］. Computer Integrated Manufacturing System, 2017, 23(10):2172-2179.
［3］WANG Junliang, YANG Jungang, ZHANG Jie, et al. Big Data Driven Cycle Time Parallel Prediction for Production Planning in Wafer Manufacturing［J］. Enterprise Information System, 2018, 12(6):714-732.
［4］TAI Y, PEARN W, LEE J. Cycle Time Estimation for Semiconductor Final Testing Processes with Weibull-distributed Waiting Time［J］. International Journal of Production Research, 2012, 50(2):581-592.
［5］YANG F, ANKENMAN B. Estimating Cycle Time Percentile Curves for Manufacturing Systems via Simulation［J］. INFORMS Journal on Computing, 2008, 20(4):628-643.
［6］YANG F, ANKENMAN B, NELSON L. Efficient Generation of Cycle Time-throughput Curves through Simulation and Metamodeling［J］. Naval Research Logistics, 2007, 54(1):78-93.
［7］HSIEH L, CHANG K. Efficient Development of Cycle Time Response Surfaces Using Progressive Simulation Metamodeling［J］. International Journal of Production Research, 2014, 52(10):3097-3109.
［8］SCHELASIN R. Using Static Capacity Modeling and Queuing Theory Equation to Predict Factory Cycle Time Performance in Semiconductor［C］∥IEEE Proceedings of the 2011 Winter Simulation Conference (WSC). Phoenix, 2011:2040-2049.
［9］CHUANG S, HUANG H. Cycle Time Estimation for Wafer Fab with Engineering Lots［J］. IIE Transactions, 2022, 34(2):105-118.
［10］WANG Junliang, ZHANG Jie, WANG Xiaoxi. Bilateral LSTM:a Two-dimensional Long Short-term Memory Model with Multiply Memory Units for Short-term Cycle Time Forecasting in Re-entrant Manufacturing Systems［J］. IEEE Transactions on Industrial Informatics. 2018, 14(2):748-758.
［11］CHEN Toly, WANG Yichi. An Iterative Procedure for Optimizing the Performance of the Fuzzy-Neural Job Cycle Time Estimation Approach in a Wafer Fabrication Factory［J］. Mathematical Problems in Engineering, 2013，2013:740478.
［12］CHIEN C, HSU C, HSIAO C. Manufacturing Intelligence to Forecast and Reduce Semiconductor Cycle Time［J］. Journal of Intelligent Manufacturing, 2012, 23(6):2281-2294.
［13］TIRKEL L. Forecasting Flow Time in Semiconductor Manufacturing Using Knowledge Discovery in Database［J］. International Journal of Production Research, 2013, 51(18):5536-5548.
［14］WANG Junliang, ZHENG Peng, ZHANG Jie. Big Data Analytics for Cycle Time Related Feature Selection in the Semiconductor Wafer Fabrication System［J］. Computers ＆ Industrial Engineering, 2014, 6(2):409-426.
［15］PALMA M, RODRIGUEZ D. Distributed Relief F-based Feature Selection in Spark［J］. Knowledge and Information Systems, 2018, 57(1):1-20.
［16］WANG C, PEDRYCZ W, YANG J, et al. Wavelet Frame-based Fuzzy C-means Clustering for Segmenting Images on Graphs［J］. IEEE Transactions on Cybernetics, 2020, 50(9):3938-3949.
［17］GAO Shuai, HUANG Yuefei, ZHANG Shuo, et al. Short-term Runoff Prediction with GRU and LSTM Networks without Requiring Time Step Optimization during Sample Generation［J］. Journal of Hydrology, 2020, 589:125188.
［18］YAIR M, BOAZ L,GAD R, et al. Cycle-time Key Factor Identification and Prediction in Semiconductor Manufacturing Using Machine Learning and Data Mining［J］. IEEE Transactions on Semiconductor Manufacturing, 2011, 24(2):237-248.

[1]	阴艳超, 施成娟, 邹朝普, 刘孝保. 基于深度时间卷积神经网络与迁移学习的流程制造工艺过程质量时序关联预测[J]. 中国机械工程, 2023, 34(14): 1659-1671.
[2]	陈昱, 项薇, 龚川. 基于数据挖掘的注塑产品质量在线故障检测及预测[J]. 中国机械工程, 2023, 34(14): 1749-1755.
[3]	向柔潼, 王时龙, 周杰, 易力力. 辐照环境下新型气动位置传感器压力响应特性研究[J]. 中国机械工程, 2023, 34(11): 1364-1371.
[4]	高新, 叶楠, 周坤, 陶友瑞, . 考虑测量误差的两阶段Wiener建模与可靠性分析[J]. 中国机械工程, 2023, 34(09): 1067-1076.
[5]	石晴晴, 张润锋, 张连洪, 兰世泉, . 基于强化学习算法的水下滑翔机路径跟踪研究[J]. 中国机械工程, 2023, 34(09): 1100-1110.
[6]	郝澜宇, 周笛, 李艳婷, 潘尔顺. 考虑二元Copula统计量的晶圆制造叠加误差监测[J]. 中国机械工程, 2023, 34(03): 369-377.
[7]	庹军波, 彭秋媛, 张贤明, 李聪波. 工业机器人能耗预测研究[J]. 中国机械工程, 2022, 33(22): 2727-2732,2740.
[8]	陈帅, 郭飞燕, 孟月梅, 王明阳, 侯志霞. 融合实测数据的航空结构件修配量迭代寻优及评价方法[J]. 中国机械工程, 2022, 33(17): 2061-2070，2078.
[9]	张瑾, 韩福柱, . 电弧铣削加工间隙电压预测模型的建立与研究[J]. 中国机械工程, 2022, 33(16): 1891-1896.
[10]	屈聪, 孟智娟, 赵亮, 陈耀, 马立东. 基于变弹性模量的Ti-6Al-4V板材五点弯曲回弹预测[J]. 中国机械工程, 2022, 33(16): 1991-1999.
[11]	张阳, 高曙, 何伟, 蔡菁. 基于多模型融合的内河船舶航行轨迹预测方法[J]. 中国机械工程, 2022, 33(10): 1142-1152.
[12]	陈冬梅, 赵思恒, 魏承印, 陈亚杰. 船舶柴油机状态监测及预测性维护研究及应用[J]. 中国机械工程, 2022, 33(10): 1162-1168.
[13]	刘永兴, 唐小琦, 钟靖龙, 钟震宇, 周向东. 基于轻量级梯度提升机的非对称风险注塑成形产品尺寸预测模型[J]. 中国机械工程, 2022, 33(08): 965-969.
[14]	艾青波, 张洁, 程辉, 吕佑龙, 左丽玲, 胡蓝. 基于PointCPP-LSF方法的航天筒段薄壁件滚弯成形质量分析[J]. 中国机械工程, 2022, 33(08): 977-985.
[15]	何彦, 肖圳, 李育锋, 吴鹏程, 刘德高, 杜江. 使用CNN-SVR的汽车组合仪表组装质量预测方法[J]. 中国机械工程, 2022, 33(07): 825-833.

融合双重注意力机制与并行门控循环单元的晶圆加工周期预测方法

A Wafer Cycle Processing Time Prediction Method Incorporating Double Attention Mechanism and Parallel GRU

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics