Control Parameter Extraction Model for Reconfigurable Bandpass Filter Based on Bayesian Optimized Multiple-Output XGBoost

This letter proposes a control parameter extraction (CPE) model for the reconfigurable bandpass filter (RBPF) to achieve fast reconfiguration of center frequency and broadband. The CPE model consists of an automatic hyperparameter generation (AHG) module and a multioutput XGBoost (MO-XGB) module. AHG, based on Bayesian optimization, is used to generate the MO-XGB module hyperparameters. Through experiments on the fabricated circuit, the model extracted the control parameters of RBPFs in 15 ms with root mean square error (RMSE) of 0.0098 and 0.0129 for center frequency and bandwidth, respectively. Therefore, the model’s effectiveness has been proven.