Multiple Linear and Polynomial Regression Models for the Design of Ultra-Wideband Antennas with Band-Notch Characteristics

We have developed a supervised learning model using artificial intelligence to synthesize electromagnetic bandgap structure-loaded ultra-wideband antennas.The effectiveness of the proposed model was validated by creating a modified U-slot microstrip patch antenna.Multiple linear and polynomial regression models were used to predict the cutoff and notch frequencies, and the error performance matrix.This innovative approach makes proposed model ideal for efficient machine-tomachine communications in the S-band and C-band for space applications.Polynomial regression performs better than multiple linear regression for notch frequency, and it achieved correlation coefficients of 0.989 and 0.961 for training and testing data, respectively.