Abstract:Capacitive coupling wireless power transfer (CCWPT) is one of the pervasive methods to transfer power in the reactive near-field zone. In this paper, a flexible design methodology based on Binary Particle Swarm Optimization (BPSO) algorithm is proposed for a pixelated microstrip structure. The pixel configuration of each parallel plate (43x43 pixels) determines the frequency response of the system (S-parameters) and by changing this configuration, we can achieve the dedicated operating frequency (resonance frequency) and its related |S21| value. Due to the large number of pixels, iterative optimization algorithm (BPSO) is the solution for designing a CCWPT system. However, the output of each iteration should be simulated in electromagnetic simulators (e.g., CST, HFSS, etc.), hence, the whole optimization process is time-consuming. This paper develops a rapid, agile and efficient method for designing two parallel pixelated microstrip plates of a CCWPT system based on deep neural networks. In the proposed method, CST-based BPSO algorithm is replaced with an AI-based method using ResNet-18. Advantages of the AI-based iterative method are automatic design process, more efficient, less time-consuming, less computational resource-consuming and less background EM knowledge requirements compared to the conventional techniques. Finally, the prototype of the proposed simulated structure is fabricated and measured. The simulation and measurement results validate the design procedure accuracy, using AI-based BPSO algorithm. The MAE (Mean Absolute Error) of prediction for the main resonance frequency and related |S21| are 110 MHz and 0.18 dB, respectively and according to the simulation results, the whole design process is 3629 times faster than the CST-based BPSO algorithm.

Rapid FSS Design Method Based on Fourier Transform

Fourier Subspace-Based Deep Learning Method for Inverse Design of Frequency Selective Surface

Design of Curved FSS with A Fast Accurate Simulation Method Based on Transmission Line Theory

Dual-Band FSS Inverse Design Using ANN with Cognition-Driven Sampling

Conjugate Adjoint Gradient-Based Inverse Design Method for Aperiodic Frequency-Selective Surface with Weighted Loss

A Conformal Frequency Selective Rasorber with Intelligent Band-Transfer Abilities Via Jointly Learning Network

Fabrication-Constrained Conjugate Adjoint Gradient Inverse Design for Large-Scale Aperiodic Frequency Selective Surface

Design of Miniaturized-Element Frequency Selective Surface Using Neural Networks

Frequency Selective Surface Analysis Based on Multimodal Network Approach

A Response Surface Methodology Based On Improved Compactly Supported Radial Basis Function And Its Application To Rapid Optimizations Of Electromagnetic Devices

Multiform Frequency Selective Surfaces Optimal Design Based on Topology Optimization

Inverse Design of Multifunctional Frequency-Selective Surface Under Inhomogeneous Background

Rapid Surrogate Modeling of Electromagnetic Data in Frequency Domain Using Neural Operator.

Multiphysics Inverse Design of Frequency Selective Surface by Data-Physics Driven Deep Neural Network

Fast And Automatic Rf Design Based On Matlab-Hfss Control Applied On Magnetic Absorber With Metasurface

Efficient Numerical Model to Analyze the Conformal Frequency-Selective Surfaces

Efficient Frequency Selective Surface Analysis via End-to-End Model-Based Learning

Intelligent Design of Arbitrary Bandstop FSS Through Deep Learning and Genetic Algorithm

An Ultra-high-speed Fully-parallel Fast Fourier Transform Design

Pre-Processing-based Fast Design of Multiple EM Structures with One Deep Neural Network

Ultra-Fast and Efficient Design Method Using Deep Learning for Capacitive Coupling WPT System