面向晶圆制造工期预测的可解释深度学习方法

doi:10.3901/JME.2024.22.179

机械工程学报 ›› 2024, Vol. 60 ›› Issue (22): 179-191.doi: 10.3901/JME.2024.22.179

扫码分享

面向晶圆制造工期预测的可解释深度学习方法

高鹏捷^1,2,3, 汪俊亮^2,3, 张洁^2,3

1. 东华大学机械工程学院上海 201620;
2. 东华大学人工智能研究院上海 201620;
3. 上海工业大数据与智能系统工程技术研究中心上海 201620

收稿日期:2023-12-30 修回日期:2024-07-17 出版日期:2024-11-20 发布日期:2025-01-02
作者简介:高鹏捷,男,1998年出生。主要研究方向为工业大数据和计算机视觉。E-mail:pengjiegao@mail.dhu.edu.cn;汪俊亮(通信作者),男,1991年出生,博士,副教授,硕士研究生导师。主要研究方向为智能制造系统和工业大数据。E-mail:junliangwang@dhu.edu.cn;张洁,女,1963年出生,博士,教授,博士研究生导师。主要研究方向为智能制造。E-mail:mezhangjie@dhu.edu.cn
基金资助:
国家自然科学基金(52275478，51905091)和东华大学研究生创新基金(GSIF-DH-M-2022007)资助项目。

Interpretable Deep Learning Method for Wafer Manufacturing Cycle Time Forecasting

GAO Pengjie^1,2,3, WANG Junliang^2,3, ZHANG Jie^2,3

1. College of Mechanical Engineering, Donghua University, Shanghai 201620;
2. Institute of Artificial Intelligence, Donghua University, Shanghai 201620;
3. Shanghai Engineering Research Center of Industrial Big Data and Intelligent System, Shanghai 201620

Received:2023-12-30 Revised:2024-07-17 Online:2024-11-20 Published:2025-01-02
About author:10.3901/JME.2024.22.179

摘要/Abstract

摘要： 晶圆制造工期预测是晶圆制造系统运行优化的核心问题，是保障晶圆产品准时交付的关键。深度学习方法从海量数据中识别规律，在静态环境下的产品工期预测问题中取得了良好的效果。然而，在车间在制品水平等因素动态变化下，当前方法通过构建复杂系统的黑箱模型预测晶圆制造工期，缺乏可解释性，难以阐述模型随系统状态的变动规律。提出一种面向晶圆制造工期预测的可解释深度学习方法(Interpretable deep learning method，IDLM)，具体包括构建脑启发的深度神经网络结构解析模型，从“神经元→神经环路→神经网络”三个层面为预测网络解析提供结构基础；设计工期预测网络的关键神经元识别算法，利用信息熵权规则滤取工期预测网络中的关键神经元；提出工期预测网络关键神经环路搜索算法，快速搜索相似神经元优化组合以得到关键预测环路。试验结果表明，IDLM可在保持工期预测精度的同时提取出预测神经网络的关键神经环路，为动态环境下工期预测网络的自适应优化提供基础。

关键词: 深度神经网络, 可解释学习, 网络解析, 工期预测, 半导体生产

Abstract: Wafer manufacturing cycle time forecasting is the core problem of semiconductor wafer fabrication system operation optimization, which is the key to guaranteeing the on-time delivery of wafer products. Deep learning methods learn the data fluctuation laws from massive data, construct black-box prediction models of complex systems, and achieve impressive prediction accuracy in static environments. However, under dynamic system state fluctuation, such as workshop work-in-process levels, current methods cannot stay accurate prediction due to the lack of interpretability to explain the changing rules of the forecasting model with system states. Therefore, an interpretable deep learning method(IDLM) for wafer manufacturing cycle time forecasting is proposed to clarify the organization rules of forecasting neural networks under different system states. First, a brain-inspired interpretable structural model of the wafer manufacturing cycle time forecasting neural network is constructed to provide a structural basis for the analysis of the network in the organization form of "neurons-neural circuits-neural network". Second, a key neuron recognition method of cycle time forecasting network is proposed to filter important neurons from the network with information entropy weighted rules constraint. Finally, a key neural circuit search algorithm is designed to quickly search for the optimal combination of similar neurons to obtain the key forecasting circuits. The experimental results show that IPM can extract the key neural circuits of the forecasting network while maintaining the accuracy, which provides a key structural basis for the network self-assembly under dynamic environments.

Key words: deep neural network, interpretable learning, network parsing, cycle time forecasting, semiconductor manufacturing

中图分类号:

TP183

高鹏捷, 汪俊亮, 张洁. 面向晶圆制造工期预测的可解释深度学习方法[J]. 机械工程学报, 2024, 60(22): 179-191.

GAO Pengjie, WANG Junliang, ZHANG Jie. Interpretable Deep Learning Method for Wafer Manufacturing Cycle Time Forecasting[J]. Journal of Mechanical Engineering, 2024, 60(22): 179-191.

参考文献

[1] UZSOY R，FOWLER J，MOENCH L. A survey of semiconductor supply chain models Part II：Demand planning，inventory management，and capacity planning[J]. International Journal of Production Research，2018，56(13)：4546-4564.
[2] CHEN T，WANG Y. A nonlinearly normalized back propagation network and cloud computing approach for determining cycle time allowance during wafer fabrication[J]. Robotics and Computer-Integrated Manufacturing，2017，45：144-156.
[3] 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.
[4] 刘子文，刘检华，程益，等. 基于文本挖掘与神经网络的复杂产品装配工时估算方法[J]. 机械工程学报，2021，57(15)：199-210. LIU Ziwen，LIU Jianhua，CHENG Yi，et al. Assembly time estimation method for complex products based on text mining and neural network[J]. Journal of Mechanical Engineering，2021，57(15)：199-210.
[5] 刘娟，刘检华，庄存波，等. 基于特征聚类的结构件数控加工工时预测方法[J]. 机械工程学报，2023，59(15)：232-246. LIU Juan，LIU Jianhua，ZHUANG Cunbo，et al. Working-hours prediction method of CNC Machining of structural parts based on feature clustering[J]. Journal of Mechanical Engineering，2023，59(15)：232-246.
[6] WANG Junliang，YANG Jungang，ZHANG Jie，et al. Big data driven cycle time parallel prediction for production planning in wafer manufacturing[J]. Enterprise Information Systems，2018，12(6)：714-732.
[7] ZHANG Chuwen，BARD J，CHACON R. Controlling work in process during semiconductor assembly and test operations[J]. International Journal of Production Research，2017，55(24)：7251-7275.
[8] 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.
[9] SHA D，STORCH R，LIU C. Development of a regression-based method with case-based tuning to solve the due date assignment problem[J]. International Journal of Production Research，2007，45(1)：65-82.
[10] XU Ke，LI Yingguang，LIU Changqing，et al. Advanced data collection and analysis in data-driven manufacturing process[J]. Chinese Journal of Mechanical Engineering，2020，33(1)：1-21.
[11] SCHELASIN R. Using static capacity modeling and queuing theory equations to predict factory cycle time performance in semiconductor manufacturing[C]// Proceedings of the 2011 Winter Simulation Conference (WSC). Phoenix，AZ，USA：IEEE，2011：2040-2049.
[12] GUO Zhishan，BARUAH S. A neurodynamic approach for real-time scheduling via maximizing piecewise linear utility[J]. IEEE Transactions on Neural Networks and Learning Systems，2016，27(2)：238-248.
[13] CHU Zhenzhong，ZHU Daqi，YANG S. Observer-based adaptive neural network trajectory tracking control for remotely operated vehicle[J]. IEEE Transactions on Neural Networks and Learning Systems，2017，28(7)：1633-1645.
[14] BACKUS P，JANAKIRAM M，MOWZOON S，et al. Factory cycle-time prediction with a data-mining approach[J]. IEEE Transactions on Semiconductor Manufacturing，2006，19(2)：252-258.
[15] TIRKEL I. Forecasting flow time in semiconductor manufacturing using knowledge discovery in databases[J]. International Journal of Production Research，2013，51(18)：5536-5548.
[16] CHEN T，WANG Y. Incorporating the FCM–BPN approach with nonlinear programming for internal due date assignment in a wafer fabrication plant[J]. Robotics and Computer-Integrated Manufacturing，2010，26(1)：83-91.
[17] ZHAO Baoxuan，CHENG Changming，TU Guowei，et al. An interpretable denoising layer for neural networks based on reproducing kernel hilbert space and its application in machine fault diagnosis[J]. Chinese Journal of Mechanical Engineering，2021，34：44.
[18] WANG C，JIANG P. Deep neural networks based order completion time prediction by using real-time job shop RFID data[J]. Journal of Intelligent Manufacturing，2019，30(3)：1303-1318.
[19] ULLMAN S. Using neuroscience to develop artificial intelligence[J]. Science，2019，363(6428)：692-693.
[20] 汪俊亮，高鹏捷，张洁，等. 制造大数据分析综述：内涵、方法、应用和趋势[J]. 机械工程学报，2023，59(12)：1-16. WANG Junliang，GAO Pengjie，ZHANG Jie，et al. A review of manufacturing big data：Connotation，methodology，application and trends[J]. Journal of Mechanical Engineering，2023，59(12)：1-16.
[21] BAU D，ZHU Junyan，STROBELT H，et al. Understanding the role of individual units in a deep neural network[J]. Proceedings of the National Academy of Sciences，2020，117(48)：30071-30078.
[22] O’NEILL D，XUE Bing，ZHANG Mengjie. Evolutionary neural architecture search for high-dimensional skip-connection structures on densenet style networks[J]. IEEE Transactions on Evolutionary Computation，2021，25(6)：1118-1132.
[23] MOLCHANOV P，TYREE S，KARRAS T，et al. Pruning convolutional neural networks for resource efficient inference[C]// 5th International Conference on Learning Representations (ICLR 2017). Toulon，France，2017：1-17.
[24] LIU Zhuang，LI Jianguo，SHEN Zhiqiang，et al. Learning efficient convolutional networks through network slimming[C]// 2017 IEEE International Conference on Computer Vision (ICCV 2017). Venice，ITALY：IEEE，2017：2755-2763.
[25] DELLAFERRERA G，WOŹNIAK S，INDIVERI G，et al. Introducing principles of synaptic integration in the optimization of deep neural networks[J]. Nature Communications，2022，13(1)：1-14.
[26] ELSKEN T，METZEN J，HUTTER F. Neural architecture search：A survey[J]. Journal of Machine Learning Research，2019，20：55.
[27] ZOPH B，LE Q. Neural architecture search with reinforcement learning[C]// 5th International Conference on Learning Representations (ICLR 2017). Toulon，France，2017：1-16.
[28] ELSKEN T，METZEN J，HUTTER F. Efficient multi-objective neural architecture search via lamarckian evolution[C]// 7th International Conference on Learning Representations (ICLR 2019). New Orleans，Louisiana，USA，2019：1-23.
[29] CUI Yuwei，LIU L，KHAWAJA F，et al. Diverse suppressive influences in area MT and selectivity to complex motion features[J]. Journal of Neuroscience，2013，33(42)：16715-16728.
[30] 汪俊亮，张洁. 大数据驱动的晶圆工期预测关键参数识别方法[J]. 机械工程学报，2018，54(23)：185-191. WANG Junliang，ZHANG Jie. Big data driven key factor identification for cycle-time forecasting of wafer lots in semiconductor wafer fabrication system[J]. Journal of Mechanical Engineering，2018，54(23)：185-191.
[31] QIN Wei，ZHANG Jie，SUN Yinbin. Multiple-objective scheduling for interbay AMHS by using genetic-programming-based composite dispatching rules generator[J]. Computers in Industry，2013，64(6)：694-707.

面向晶圆制造工期预测的可解释深度学习方法

Interpretable Deep Learning Method for Wafer Manufacturing Cycle Time Forecasting

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 11

编辑推荐

Metrics

本文评价

[1]	常建涛, 乔子萱, 孔宪光, 杨胜康, 罗才文. 多维非线性特征重构与融合的复杂产品工期预测方法[J]. 机械工程学报, 2023, 59(6): 295-309.
[2]	陆晨浩, 陈耀, 何若琪, 范维莹, 冯健. 基于深度神经网络的四折痕锥形折纸结构设计[J]. 机械工程学报, 2023, 59(21): 167-176.
[3]	林京, 焦金阳. 可解释机械智能诊断技术的研究进展与挑战[J]. 机械工程学报, 2023, 59(20): 215-224.
[4]	张立, 陈江涛, 熊芬芬, 任成坤, 李超. 基于元学习的多可信度深度神经网络代理模型[J]. 机械工程学报, 2022, 58(1): 190-200.
[5]	喻曦, 赵欢, 李祥飞, 丁汉. 基于双深度神经网络的轮廓误差补偿策略研究[J]. 机械工程学报, 2019, 55(3): 130-137.
[6]	侯文擎, 叶鸣, 李巍华. 基于改进堆叠降噪自编码的滚动轴承故障分类[J]. 机械工程学报, 2018, 54(7): 87-96.
[7]	汪俊亮, 张洁. 大数据驱动的晶圆工期预测关键参数识别方法[J]. 机械工程学报, 2018, 54(23): 185-191.
[8]	孙文珺, 邵思羽, 严如强. 基于稀疏自动编码深度神经网络的感应电动机故障诊断[J]. 机械工程学报, 2016, 52(9): 65-71.
[9]	翟文彬;张洁;严隽琪;马登哲. 基于ETAEMS/GPGP-CN的半导体生产线动态调度技术研究[J]. , 2005, 41(3): 53-58.
[10]	张洁;翟文彬;严隽琪;马登哲. 基于模糊神经网络的半导体生产线重调度策略优化[J]. , 2005, 41(10): 75-79.
[11]	翟文彬;褚学宁;张洁;马登哲;金烨;严隽琪. 基于组合拍卖的半导体生产线短期调度技术研究[J]. , 2004, 40(9): 95-99.