A Fast Surrogate Model-Based Algorithm Using Multilayer Perceptron Neural Networks for Microwave Circuit Design

This paper introduces a novel algorithm for designing low-pass filter (LPF) and microstrip Wilkinson power divider (WPD) using neural network surrogate model. The proposed is applicable to various microwave devices, enhancing their performance frequency response. Desirable output parameters can be achieved the designed LPF WPD by algorithm. artificial (ANN) model employed calculate dimensions of WPD, resulting in efficient design. designs incorporate open stubs, stepped impedances, triangular-shaped resonators, meandered lines achieve optimal performance. compact occupies size only 0.15 λg × 0.081 λg, exhibits sharp response within transmission band, with sharpness parameter approximately 185 dB/GHz. operating at 1.5 GHz, outstanding harmonics suppression from 2 GHz 20 attenuation levels exceeding dB. successfully suppresses 12 unwanted (2nd 13th). obtained results demonstrate that design effectively accomplishes designs, exhibiting desirable such as high-frequency suppression. demonstrates low insertion loss 0.1 dB (S21 = dB), input return losses 30 (S11 −35 dB, S22 −30 an ports isolation more than 32 (S23 −32 making it suitable integration into modern communication systems.