Data-Driven Ensemble Prediction Models for Design of Microwave Filter for Sensing Applications

This research introduces the design and optimization of a microwave ring filter tailored for sensing applications. The filter incorporates log-periodic feedlines (LPF) with a hollow ring resonator, enhancing multifunctionality and performance and operating as a band pass filter (BPF) between 0.3 and 1 GHz, with resonating frequencies at 0.4, 0.6, and 0.9 GHz. This study details the filter's design, simulation results, and experimental validation. Additionally, various ensemble models were employed for filter optimization and trained using scattering parameters, with the CatBoost model giving the best results. The proposed optimization scheme's feasibility and accuracy are confirmed through comparisons with predictions.