Volterra series and neural networks to model an electronic device nonlinear behavior

Electronic devices that have a dynamic nonlinear behavior are generally modeled by the so-called Volterra series model. The construction of such analytical model, however, is a complex and time-consuming task. In this work we propose a new approach to help in the building of a Volterra series model for modeling an electronic device, using a very simple neural network model. As an initial case of study, the proposed approach is applied to the modeling of a diode and the simulation results are presented. Conclusions are drawn from the comparison of the results obtained with different kernels approximations and network topologies.