Interdigital Dual Band Filter Design Based on Surrogate Modeling and Optimization

Abstract In recent years, dual‐band filters have become critically important in satellite and space communication systems. This study presents an optimized digital twin‐based modeling strategy for a small‐sized microstrip interdigital dual‐band filter. The main innovation of the study is the achievement of high‐accuracy S 11 (dB) estimation in the 1–6 GHz frequency range through the combination of reduced data usage via Latin Hypercube sampling and optimization methods. Models created using eight input parameters (including the frequency dimension) were evaluated using MSE, MAE, and R 2 error metrics. According to the results, the proposed approach achieved MSE of 1.871, MAE of 0.486, and R 2 of 0.810 on the test data, providing a significant performance improvement, especially compared to the SVM‐based model (Test MSE: 5.130, R 2 : 0.329). Furthermore, lower error values and more stable generalization performance were obtained compared to the Ensemble method. The results clearly demonstrate the frequency‐dependent variation of S‐parameters through target/prediction comparisons. The proposed approach offers an effective and generalizable solution for digital twin‐supported modeling in high‐dimensional and computationally expensive electromagnetic problems.