Deep Learning-Enabled Rapid Optimization for Microwave Filter Design

Conducting microwave simulations with traditional electronic design automation (EDA) software could be inefficient and time-consuming when the electromagnetic (EM) effects are complex. Deep learning (DL) models have been widely used to improve the efficiency of microwave filter behavior prediction; however, its simulation results cannot guarantee the same accuracy as physical-equation-based EDA software without a large amount of high-quality training data. Many optimizer algorithms have been utilized to improve the DL model performance. The paper introduces a DL-enabled particle swarm optimization (PSO) algorithm to solve the trade-off between efficiency and accuracy of microwave filter behavior prediction. Additionally, the DL-enabled PSO algorithm combines the CST simulation, further optimizing the microwave circuit design. An example of a hairpin band-pass filter is discussed in the paper to verify the performance of the proposed optimization algorithm. The results show that the proposed DL-enabled PSO accelerates optimization by over 500 times in an iteration and decreases return loss by 10 dB across the pass-band of the designed filter.