ANN based Selective Harmonic Elimination for Cascaded H-Brdige Multilevel Inverter

The paper represents a new approach for the elimination of undesired lower order harmonics (selective harmonic elimination) in the output voltage waveform by using Artificial neural network (ANN) for cascaded multilevel inverters. In this work the network was trained in the offline process and the required converging angles are taken from the classic harmonic elimination method for any values of the modulation index. After training the ANN system a large look up table is prepared from the neural network to generate optimum switching angles to have lower THD value from the given modulation index. This method can be applied to any levels of multilevel inverter. In the paper the work has been done on 5-level and 7-level inverter. It has been trained and the global optimum angles are obtained for the given modulation index values, to eliminate the undesired lower order harmonics and reduce the total THD value. The SHE-elimination are validated in MATLAB simulation, By using ANN method the high performance in the system development and lower THD has been achieved.