AI-based Hybrid Approach (RL/GA) used for Calculating the Characteristic Parameters of a Single Surface Microstrip Transmission Line

This contribution presents a hybrid AI method for predicting the characteristic parameters of transmission line networks on PCBs using a simple example (single surface microstrip structure). This work is part of the current developments in the field of Physically Based AI Methods for EDA Applications. It is intended as a proof of concept and helping to better understand the application of a hybrid algorithm (ML-Assisted GA: DDRL + GA) for the purpose of waveform behaviour (shape) and routing-pattern optimization in the solution of physics-based problems with the application to a simple problem (field problem for Q-TEM structures; single surface microstrip). To determine the characteristic parameters for the Q-TEM case, the shape of potential curve describing the solution of the corresponding field problem is determined using Thomson's theorem (energy minimum of the microstrip arrangement under consideration).