With the development of multi service wireless communication technology,the differential band-pass filter with common mode rejection capability has attracted more and more attention.However,the traditional method of designing differential microstrip filters is complex and long time consumption.To solve these problems,combined with the eXtreme Gradient Boosting(XGBoost)algorithm with nonlinear weighting,the filter is designed by training and learning the relationship between the filter structure parameters and the corresponding frequency response S parameters.A dumbbell-shaped differential band-pass filter based on the improved step impedance resonator(SIR)is designed.The central frequency of the filter is 2.6 GHz,the relative bandwidth is 7.3%,the insertion loss in the passband is better than 0.95 dB,the return loss is better than 20 dB,the common mode rejection in the passband is better than 42 dB,and the stopband is wide.The consistency between the physical test results and the algorithm prediction results verifies that PCA-XGBoost algorithm(Principal Component Analysis,PCA)can be used to assist in the design of filters.
关键词
哑铃状SIR谐振器/差分微带滤波器/PCA-XGBoost算法
Key words
dumbbell-shaped SIR resonator/differential microstrip filter/PCA-XGBoost algorithm