The-Optimum-Training-Data-Structure-For-Modelling-Of-Microwave-Transistors-Using-Artificial-Neural

In this work, the optimum amount of training data for modelling of microwave transistors using Multilayer Perceptron (MLP) is studied. For this purpose, BPF 640 having wide operation frequency range within the large bias voltage and current range is chosen and optimum training data amount is determined for its accurate and rapid modelling. According to its Manufacturer’s Data Sheets, BPF 640 has the operation frequency from 10 MHz up to 10 GHz within the region of 1V<VDS<4V and 1mA<IDS<20mA. Interpolation is chosen for the MLP for the generalization process since MLP is generalizing successfully in interpolation mode. In congress, all the details of models and comparisons will be presented.