Machine-Learning-Based PML for the FDTD Method

In this letter, a novel absorbing boundary condition (ABC) computation method for finite-difference time-domain (FDTD) is proposed based on the machine learning approach. The hyperbolic tangent basis function (HTBF) neural network is introduced to replace traditional perfectly matched layer (PML) ABC during the FDTD solving process. The field data on the interface of conventional PML are employed to train HTBF-based PML model. Compared to the conventional approach, the novel method greatly decreases the size of a computation domain and the computation complexity of FDTD because the new model only involves the one-cell boundary layer. Numerical examples are provided to benchmark the performance of the proposed method. The results demonstrate that the newly proposed method could replace conventional PML and could be integrated into FDTD solving process with satisfactory accuracy and compatibility to FDTD. According to our knowledge, this proposed model combined artificial neural network (ANN) model is an unreported new approach based on a machine learning based for FDTD.

[1]  A. Majda,et al.  Absorbing boundary conditions for the numerical simulation of waves , 1977 .

[2]  He Ming Yao,et al.  Embedding the Behavior Macromodel Into TDIE for Transient Field-Circuit Simulations , 2016, IEEE Transactions on Antennas and Propagation.

[3]  G. Mur.  Absorbing Boundary Conditions for the Finite-Difference Approximation of the Time-Domain Electromagnetic-Field Equations , 1981, IEEE Transactions on Electromagnetic Compatibility.

[4]  Berkman Sahiner,et al.  Classification of mass and normal breast tissue: a convolution neural network classifier with spatial domain and texture images , 1996, IEEE Trans. Medical Imaging.

[5]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[6]  K. Yee.  Numerical solution of initial boundary value problems involving maxwell's equations in isotropic media , 1966 .

[7]  W. Sha,et al.  The FDTD Method: Essences, Evolutions and Applications to Nano-Optics and Quantum Physics , 2017 .

[8]  Stephen D. Gedney,et al.  Convolution PML (CPML): An efficient FDTD implementation of the CFS–PML for arbitrary media , 2000 .

[9]  Jian-Ming Jin,et al.  On the development of a higher-order PML , 2005, SBMO/IEEE MTT-S International Conference on Microwave and Optoelectronics, 2005..

[10]  I.S. Stievano,et al.  M/spl pi/log, macromodeling via parametric identification of logic gates , 2004, IEEE Transactions on Advanced Packaging.

[11]  Yong Wang,et al.  FDTD Analysis of EW Wave Circulating by a Magnetized Ferrite Body in Free Space , 1998 .

[12]  Phil Kim,et al.  MATLAB Deep Learning , 2017, Apress.

[13]  Radford M. Neal.  Pattern Recognition and Machine Learning , 2007, Technometrics.

[14]  He Ming Yao,et al.  Machine Learning Based Neural Network Solving Methods for the FDTD Method , 2018, 2018 IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting.

[15]  Qi-Jun Zhang,et al.  A New Training Approach for Robust Recurrent Neural-Network Modeling of Nonlinear Circuits , 2009, IEEE Transactions on Microwave Theory and Techniques.

[16]  M.W. Chevalier,et al.  A PML using a convolutional curl operator and a numerical reflection coefficient for general linear media , 2004, IEEE Transactions on Antennas and Propagation.

[17]  Ji Wu,et al.  Study on a Poisson's equation solver based on deep learning technique , 2017, 2017 IEEE Electrical Design of Advanced Packaging and Systems Symposium (EDAPS).

[18]  D. Katz,et al.  Validation and extension to three dimensions of the Berenger PML absorbing boundary condition for FD-TD meshes , 1994, IEEE Microwave and Guided Wave Letters.

[19]  Simon Haykin,et al.  Neural Networks and Learning Machines , 2010 .

[20]  Jean-Pierre Berenger,et al.  Improved PML for the FDTD solution of wave-structure interaction problems , 1997 .

[21]  He Ming Yao,et al.  Machine learning based method of moments (ML-MoM) , 2017, 2017 IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting.

[22]  Wei E. I. Sha,et al.  Waveguide Simulation Using the High-Order Symplectic Finite-Difference Time-Domain Scheme , 2010 .

[23]  S. Cummer,et al.  A simple, nearly perfectly matched layer for general electromagnetic media , 2003, IEEE Microwave and Wireless Components Letters.

[24]  S. Gedney.  An anisotropic perfectly matched layer-absorbing medium for the truncation of FDTD lattices , 1996 .

[25]  Xiaoou Tang,et al.  Image Super-Resolution Using Deep Convolutional Networks , 2014, IEEE Transactions on Pattern Analysis and Machine Intelligence.