论文信息 - Fast forecasting of VGF crystal growth process by dynamic neural networks

Fast forecasting of VGF crystal growth process by dynamic neural networks

Abstract Fast forecasting of process variables during the crystal growth is a critical step in a process development, optimization and control. The common approach based on computational fluid dynamics modeling is accurate, but too slow to deliver results in real time. Here we conducted a feasibility study on the application of dynamic artificial neural networks in the forecasting of VGF-GaAs crystal growth cooling program. Particularly, we studied various Nonlinear-AutoRegressive artificial neural networks with eXogenous inputs (NARX) with 2 external inputs and 6 outputs derived from 500 transient data sets. Data were generated by transient 1D CFD simulation. The first encouraging results are presented and the pros and cons of the application of dynamic artificial neural networks for the fast predictions of VGF process parameters are discussed.

[1] Jeffrey J. Derby,et al. On the quasi-steady-state assumption in modeling Czochralski crystal growth , 1988 .

[2] Jürgen Schmidhuber,et al. Long Short-Term Memory , 1997, Neural Computation.

[3] Jürgen Schmidhuber,et al. Learning to Forget: Continual Prediction with LSTM , 2000, Neural Computation.

[4] Philippe Martin,et al. MOTION PLANNING FOR A NONLINEAR STEFAN PROBLEM , 2003 .

[5] M. Friedman. The Use of Ranks to Avoid the Assumption of Normality Implicit in the Analysis of Variance , 1937 .

[6] Stephen A. Billings,et al. Non-linear system identification using neural networks , 1990 .

[7] Madan M. Gupta,et al. Static and Dynamic Neural Networks: From Fundamentals to Advanced Theory , 2003 .

[8] Kumpati S. Narendra,et al. Identification and control of dynamical systems using neural networks , 1990, IEEE Trans. Neural Networks.

[9] Douglas A. Wolfe,et al. Nonparametric Statistical Methods , 1973 .

[10] I. J. Leontaritis,et al. Input-output parametric models for non-linear systems Part II: stochastic non-linear systems , 1985 .

[11] Martin Holena,et al. Optimization of magnetically driven directional solidification of silicon using artificial neural networks and Gaussian process models , 2017 .