A Novel Modified Space Vector Pulse Width Modulator Based on Artificial Neural Network for Z-Source Inverter


This paper proposes a novel modified space vector pulse width modulator based on Artificial Neural Network (ANN).This method overcomes the drawbacks of conventional DSP-based modulator in the z-source inverter. This generates timing signals without computations delay time, which increases the capability of using high switching frequency. Moreover, the proposed method overcomes the nonlinearity relationship between modulation index and the boosting factor during buck and boost regions.The simulation of the proposed system is fully evaluated for different operation condition. The schemes and training of the proposed neural networks are presented and examined with respect to conventional DSP-based method. Simulation results show excellent performance for both AC-side and DC-side of the Z-source inverter for different output voltages. In addition, the proposed ANN-based modulator operates very well in both buck and boost mode and overcomes the discontinuity of modulation index and the boosting factor. The voltage stress across the devices in the proposed method is less than of the conventional DSP-based method especially for high gain operation condition.