Selective Harmonic Minimization method For Cascaded Multilevel Inverters using ANNs

The dc sources that feed the multilevel inverter are varying in time, and the switching angles should adapt to the dc source variations. Any variations in the input will result in harmonics at the output. To mitigate the harmonics in the inverter output, GA is used. A genetic algorithm optimization technique is applied to determine the switching angles for a cascaded multilevel inverter. Then, artificial neural networks are used to determine the switching angles for real time application. This eliminates specified order harmonics while maintaining the required fundamental output constant