Efficient Microstrip Bandpass Filter Design by a Surrogate-Model-Assisted Improved Decomposition-Based Multi-Objective Differential Evolution Algorithm

A one-dimensional convolutional autoencoder (1D-CAE) assisted improved decomposition-based multiobjective differential evolution (IMODEA/D) algorithm is proposed for electromagnetic (EM) optimization in microstrip bandpass filter design. The 1D-CAE is employed as a surrogate model to accelerate the...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:International Conference on Numerical Electromagnetic Modeling and Optimization for RF, Microwave, and Terahertz Applications (Online) Ročník 1; s. 1 - 3
Hlavní autoři: Wang, Zeyu, Zhang, Lu, Li, Kun-Lai, Zhang, Yongliang
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 29.07.2025
Témata:
Autoencoders
Band-pass filters
Convergence
Electromagnetic (EM) optimization
Electromagnetics
microstrip bandpass filter design
Microstrip filters
multi-objective optimization algorithm
Multiphysics
Numerical models
Optimization
surrogate model
ISSN:2575-4769
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:A one-dimensional convolutional autoencoder (1D-CAE) assisted improved decomposition-based multiobjective differential evolution (IMODEA/D) algorithm is proposed for electromagnetic (EM) optimization in microstrip bandpass filter design. The 1D-CAE is employed as a surrogate model to accelerate the design process. In addition, the convergence of the MOEA/D algorithm is improved by adding extreme value optimization and modifying the individual selection, mutation probability, boundary optimization strategy and replacing the crossover operation in the differential evolution (DE) algorithm with the simulated binary crossover (SBX). The efficiency of the proposed method is verified by a planar microstrip bandpass filter.
ISSN:2575-4769
DOI:10.1109/NEMO62710.2025.11215258