Training a Neural Network to learn the behavior and characteristics of Microstrip. The extent of deviation of predicted output from the actual output was measured in terms of maximum error percentage and average error percentage. The goal is to train Neural Networks, to predict characteristics such as impedance or resonant frequency against design parameters of 6 types of transmission lines. Using formulas and equations that define characteristics of Transmission lines, training data was generated. Training of the Neural Network was done using back-propagation algorithm. We also used other regression algorithms for comparison of performance. The extent of deviation of predicted output from the actual output was measured in terms of maximum error percentage and average error percentage. This helps determine how well an algorithm worked for a particular transmission line. Secondary objective was to determine which algorithm is best suited for practical applications.
Algorithms used: Backpropagation, Linear Regression, Logistic Regression, MLP Classifier, SGD Regression,
Transmission Lines considered: Microstrip, Slotline, Stripline, Co-Planar Waveguide(CPW), Co-Planar Strip(CPS), Microstrip Patch Antenna.