Calculation of Dyadic Green's Function in the RBFNN-Based Layered Media

Abstract

The Dyadic Green's function, in general, is comprised of spectral Dyadic Green's functions and Sommerfeld integral. Since the integral featuring high-frequency oscillation and slow decay may lead to a time-consuming calculation process. With optimal approximation and global optimum characteristics that cannot be found in other feedforward neural networks, the RBF neural network is adopted in this paper to solve the dyadic Green function. By doing so, the efficiency of solving Dyadic Green's functions can be enhanced thanks to its simple structure and fast training speed. According to the numerical results, the calculation scheme proposed in this paper is proved to be feasible.