Effectiveness of Convolutional Perfectly Matched Layer in Time-domain Numerical Analysis of Low-frequency Wireless Power Transfer

抄録

In this study, the absorption boundary conditions (ABCs) in electromagnetic field numerical analysis were investigated to improve the estimation accuracy of the magnetic coupling coefficient (k) for wireless power transfer (WPT). When using Berenger's perfectly matched layer (B-PML), the estimation accuracy of the k value was higher than that obtained using Mur's ABC. However, in the low-frequency band, at around 400Hz, the magnetic field was not sufficiently attenuated in the absorption layer. There was also a problem with numerical stability. To solve these problems, we introduce the convolutional perfectly matched layer (C-PML). In this study, we propose a simple Debye distribution function as the matrix coefficient of the stretch coordinate metric. As a result, it was possible to evolve from B-PML to C-PML simply by adding a proportional coefficient γ. With the introduction of C-PML, the magnetic field was sufficiently attenuated in the absorption layer, and the numerical stability dramatically improved. C-PML is thus useful for low-frequency WPT simulation.