Performance improvement of off-grid hybrid renewable energy system using dynamic voltage restorer

Abstract

This article proposes an Artificial Neural Network (ANN) controller of Dynamic Voltage Restorer (DVR) to improve the performance of a stand-alone hybrid renewable energy system that is feeding a new community located in Egypt. The hybrid system consists of three renewable energy sources, namely, solar PV cells, a wind turbines based-permanent magnet synchronous generator, and fuel cells. These three sources are tied to a common DC link by three boost converters, one for each source. The common DC link is connected to the AC side via a DC/AC inverter. The optimal size of the three proposed renewable sources is calculated using the HOMER software package. The DVR control is attained through regulating the load voltage at different anomalous working conditions. These conditions are three-phase fault, voltage sag/swell, and unbalanced loading. Two ANNs are utilized to adjust the IGBT pulses of the voltage source inverter (VSI) used to control DVR by regulating the D-Q axes voltage signals. These D-Q axes components at any loading condition represent the inputs to the two ANNs. The outputs of the two ANNs represent the IGBT pulses. The input/output data used for training ANNs are obtained by two optimized PI controllers, introduced for regulating the load voltage through DVR-VSI pulses at different abnormal operating conditions, and accordingly convert the static optimized PI controller into adaptive one based ANN. The system performance with the proposed ANN-DVR controller is enhanced through improving the current, voltage, and power waveforms of each generating source. With compensation of the faulty line voltage, the system retains an uninterrupted operation of the three renewable sources during fault events and consequently increases the low voltage ride through (LVRT) capability. Moreover, the total harmonic distortion is reduced.