Optimum control of Multilevel Inverters Using Artificial Neural Networks

Estimation of optimum switching angles for multilevel converters so as to produce the required fundamental voltage while at the same time eliminate specified lower order harmonics through Artificial Neural Networks (ANNs) is presented. The proposed technique uses the best performance solutions (lowest THD solutions) as the training data set for proposed ANN system. This technique can be applied to multilevel inverters with any number of levels. As an example, in this paper estimation results of optimum switching angles are given for seven levels inverter to eliminate the 5th and 7th harmonics. After training the proposed ANN system, a large and memory-demanding look-up table replaced with trained neural network to generate the optimum switching angles with lowest THD for a given modulation index. The simulation results are presented in PSCAD/EMTDC software package to validate the accuracy of estimation results by proposed ANN system.