Frequency domain methods for analysis VSI-PWM converters, using neural networks

This paper describes the development of a pulse-width modulation voltage-source inverter (VSI-PWM) model using neural networks. Two kinds of neural networks have been tested: a multilayer perceptron with backpropagation; and a radial basis function network. Both have been considered and the models have been compared with a frequency domain analytical model and with time domain models. Results have been tested with measurements. This paper presents a novel technique, based on neural networks, for the analysis of power converter harmonics