A Dynamical Control in Multilevel Converter Using an Adaptive Learning Algorithm


  • Mohan S. Selokar, Manoj Sankhe


 There is a wide use of multilevel converters as they can be used with high power and high voltage applications. A multilevel converter consists of large number of voltage levels in load  voltage and load current. A multilevel converter has number of various advantages like good quality of output voltage waveform, smaller values of  inductor and capacitor in passive filters. The output consists of less harmonics. Reduction in total harmonic distortion can be obtained  with the help of multilevel converter. Using learning algorithm like neural network, output voltage is controlled. Error in the reference voltage and output voltage is reduced. Neural network replaces the PI controller completely. A reduction in THD in output voltage and output current can be obtained by neural network by large margin as compared to PI controller. This increases the wide application of AC motor as load as it reduces torque pulsation and RF/EMI effect. It increases the efficiency by reducing power losses.