Optimized CPML for Emulating Electromagnetic Absorbers in Simulations and Its Optimization Utilizing Artificial Neural Network

To avoid the complex and resource-intensive electromagnetic absorber (EMA) modeling and simulation, this letter proposes to use the convolutional perfectly matched layer (CPML) with the complex frequency-shifted (CFS) factor instead of the EMA in broadband full-wave simulations by optimizing the CPML so that its frequency-domain reflection coefficient (FDRC) approaches that of the EMA. Six parameters determining the distribution of the space-stretched variable, the conductivity, and the CFS factor in the perfectly matched layer (PML) regions and, thus, the absorbing performance of the PML are used as the optimized parameters. An efficient optimization method based on the artificial neural network is proposed to optimize these six parameters by minimizing the mean square error between the FDRCs of the CPML and the EMA. An example is provided to demonstrate the effectiveness and efficiency of our methods.