基于数据驱动的晶体直径模型辨识方法研究

摘要/Abstract

摘要： 直拉法硅单晶生长是一个多场多相耦合、物理变化复杂,且具有大滞后和非线性现象的过程,基于单晶硅生长系统内部机理所构建的机理模型由于存在诸多假设而无法应用于工程实际。因此,本文以现有CL120-97单晶炉拉晶车间的长期、海量晶体生长数据为基础,忽略炉内复杂的晶体生长环境,对影响晶体直径的拉晶参数进行关联性分析及特征量化,探寻拉晶数据中所蕴藏的规律信息,进而建立基于数据驱动的BP神经网络晶体直径预测模型,并针对现有BP神经网络易陷入局部极小值的问题,采用遗传算法对BP神经网络的阈值和权值进行优化,以提高晶体直径预测的准确性。通过实际拉晶数据对模型预测结果进行验证,结果表明,对任意选取的8组拉晶数据进行直径预测,预测的平均相对百分比误差为0.095 71%,证明该模型对于等径阶段晶体直径的预测是可行的。

关键词: 晶体直径预测, BP神经网络, 遗传算法, 单晶炉, 数值模拟

Abstract: Czochralski silicon single crystal growth is a dynamic time-varying process with multi-field and multi-phase coupling, complex physical changes, nonlinearity and large hysteresis. However, mechanism models based on a large number of assumptions are difficult to apply in practice. Therefore, this article is based on long-term and massive crystal growth data from the CL120-97 single crystal furnace of crystal pulling workshop, which ignores the complex crystal growth environment in the furnace, analyzes the correlation of the crystal pulling parameters that affect the crystal diameter, mines the regular information contained in the data, and further builds a crystal diameter prediction model based on the BP neural network. Aiming at the problem that the existing BP neural network model is easy to fall into the local minimum, the genetic algorithm is used to optimize the weight and threshold of the BP neural network to improve the accuracy of the crystal diameter prediction of the algorithm. The model prediction results are verified by actual crystal pulling data. The results show that the average relative percentage error of prediction is 0.095 71% for diameter prediction of 8 groups of randomly selected crystal pulling data. It is proved that the model is feasible for the prediction of crystal diameter in the equal-diameter stage.

Key words: crystal diameter prediction, BP neural network, genetic algorithm, single crystal furnace, numerical simulation

中图分类号:

O78
TQ127.2

张西亚, 高德东, 王珊, 林光伟, 高俊伟. 基于数据驱动的晶体直径模型辨识方法研究[J]. 人工晶体学报, 2021, 50(8): 1552-1561.

ZHANG Xiya, GAO Dedong, WANG Shan, LIN Guangwei, GAO Junwei. Research on Identification Method of Crystal Diameter Model Based on Data Driven[J]. JOURNAL OF SYNTHETIC CRYSTALS, 2021, 50(8): 1552-1561.

参考文献

[1] 段伟锋.硅单晶等径阶段直径模型辨识与控制研究[D].西安:西安理工大学,2017.
DUAN W F. Model identification and control of silicon single crystal diameter for constant-diameter growth stage[D]. Xi'an: Xi'an University of Technology, 2017(in Chinese).
[2] GEVELBER M A, WARGO M J, STEPHANOPOULOS G. Advanced control design considerations for the Czochralski process[J]. Journal of Crystal Growth, 1987, 85(1/2): 256-263.
[3] BARDSLEY W, HURLE D T J, JOYCE G C. The weighing method of automatic Czochralski crystal growth: i. Basic theory[J]. Journal of Crystal Growth, 1977, 40(1): 13-20.
[4] 刘秀.基于双重预测PI的单晶硅直径控制系统[D].上海:东华大学,2016.
LIU X. Silicon single crystal diameter control systembased on dual predictive PI[D]. Shanghai: Donghua University, 2016(in Chinese).
[5] 王展.恒拉速工艺下硅单晶直径控制方法研究[D].西安:西安理工大学,2020.
WANG Z, Research on diameter control method of silicon single crystal under constant pulling velocity process[J]. Xi'an: Xi'an University of Technology, 2020(in Chinese).
[6] ZHENG Z C, SETO T, KIM S, et al. A first-principle model of 300 mm Czochralski single-crystal Si production process for predicting crystal radius and crystal growth rate[J]. Journal of Crystal Growth, 2018, 492: 105-113.
[7] 王海龙.模型预测控制在复杂工业过程中的应用研究[D].兰州:兰州大学,2013.
WANG H L. Application studies on model predictive control of complex industrial process[D]. Lanzhou: Lanzhou University, 2013(in Chinese).
[8] 梁炎明,刘丁,伍光宇.基于自适应重叠系数的T-S模型在线辨识算法及应用[J].控制与决策,2012,27(9):1425-1428+1432.
LIANG Y M, LIU D, WU G Y. Online T-S model identification algorithm based on adaptive overlap coefficient and its application[J]. Control and Decision, 2012, 27(9): 1425-1428+1432(in Chinese).
[9] RAHMANPOUR P, SÆLID S, HOVD M, et al. Nonlinear model predictive control of the czochralski process[J]. IFAC-PapersOnLine, 2016, 49(20): 120-125.
[10] RAHMANPOUR P, SÆLID S, HOVD M. Run-To-Run control of the Czochralski process[J]. Computers & Chemical Engineering, 2017, 104: 353-365.
[11] WINKLER J, NEUBERT M, RUDOLPH J. Nonlinear model-based control of the Czochralski process I: motivation, modeling and feedback controller design[J]. Journal of Crystal Growth, 2010, 312(7): 1005-1018.
[12] 张诚.基于模型预测控制的硅单晶体提拉效果仿真研究[D].上海:东华大学,2014.
ZHANG C. Research and simulation on effect of czochralski crystal based on the model predictive control[D]. Shanghai: Donghua University, 2014(in Chinese).
[13] 杨曼.直拉硅单晶生长中晶体直径模型辨识方法研究[D].西安:西安理工大学,2019.
YANG M. Study on identification method of crystal diameter model in czochralski silicon single crystal growth[D]. Xi'an: Xi'an University of Technology, 2019(in Chinese).
[14] WANG J Z, HOU R, WANG C, et al. Improved v-Support vector regression model based on variable selection and brain storm optimization for stock price forecasting[J]. Applied Soft Computing, 2016, 49: 164-178.
[15] YU C J, LI Y L, BAO Y L, et al. A novel framework for wind speed prediction based on recurrent neural networks and support vector machine[J]. Energy Conversion and Management, 2018, 178: 137-145.
[16] LIU P L, YAO X D, GE G Y, et al. A dynamic linearization modeling of thermally induced error based on data-driven control for CNC machine tools[J]. International Journal of Precision Engineering and Manufacturing, 2021, 22(2): 241-258.
[17] 梁炎明,苏芳,李琦,等.基于支持向量机回归的T-S模糊模型自组织算法及应用[J].自动化学报,2013,39(12):2143-2149.
LIANG Y M, SU F, LI Q, et al. A self-organizing algorithm for T-S fuzzy model based on support vector machine regression and its application[J]. Acta Automatica Sinica, 2013, 39(12): 2143-2149(in Chinese).
[18] SHAH D, KLEMENZ C F. Delay-based control model for Czochralski growth of high-quality oxides[J]. Journal of Crystal Growth, 2008, 310(7/8/9): 1448-1454.
[19] 严博涛.直拉单晶炉生长系统模型预测控制方法研究[D].西安:西安理工大学,2010.
YAN B T. Czochralski crystal growth furnace system of model predictive control method[D]. Xi'an: Xi'an University of Technology, 2010(in Chinese).
[20] 景坤雷.Cz硅单晶等径阶段热场温度建模与控制方法研究[D].西安:西安理工大学,2018.
JING K L. Research on thermal field temperature modeling and control method of Cz silicon single crystal in constant-diameter growth stage[D]. Xi'an: Xi'an University of Technology, 2018(in Chinese).
[21] 邱勇,杨泽文,周鑫宇,等.基于BP神经网络的直角折线堰过流能力预测[J].水电能源科学,2021,39(3):74-77.
QIU Y, YANG Z W, ZHOU X Y, et al. Prediction of overflow capability of right-angle polyline weir based on BP neural network[J]. Water Resources and Power, 2021, 39(3): 74-77(in Chinese).
[22] 卢兆兴,吕志峰,李婷,等.基于BP神经网络的地磁变化场预测研究[J].大地测量与地球动力学,2021,41(3):229-233.
LU Z X, LÜ Z F, LI T, et al. Forecasting of the variable geomagnetic field based on BP neural network[J]. Journal of Geodesy and Geodynamics, 2021, 41(3): 229-233(in Chinese).
[23] HOU W N, FANG XS, LI M H, et al. Stress fitting and forecast model of dams based on BP neural network[J]. IOP Conference Series: Earth and Environmental Science, 2020, 560(1): 12-28.
[24] 王林军,史宝周,张东,等.GA优化的LQR控制下3-UPS并联机器人稳定性分析[J].计算机集成制造系统,2021,4(2):1-17.
WANG L J, SHI B Z, ZHANG D, et al. Stability analysis of parallel robot under linear quadratic regulator control based on genetic algorithm optimization[J]. Computer Integrated Manufacturing Systems, 2021, 4(2): 1-17(in Chinese).
[25] HU T, WANG J H, LI G B, et al. Prediction of allowable debonding strain of {FRP}-strengthened {RC} beam based on {GA}-{BP} neural network[J]. {IOP} Conference Series: Earth and Environmental Science, 2021, 719(2): 022028.