ANN modeling of synthetic cold loads

Preliminary results are presented for artificial neural network (ANN) models of the available output noise temperature of a FET-based synthetic cold load. Two different ANNs were studied for this application: the radial basis function (RBF) and the Levenberg-Marquardt (LM) backpropagation (BP). The best average relative error (ARE) and maximum local relative error (MLRE) results for the model of incident noise temperature versus load impedance were 0.1439% and 1.1544% respectively. The best ARE and MLRE results for the model of incident noise temperature versus load reflection coefficient were 0.1810% and 1.5044% respectively.