Low Voltage Ride Through Estimation in Microgrid using Deep Neural Network

Abstract

One of the vital needs for the distribution systems is the Low-Voltage-Ride-through (LVRT) capability which has to meet the grid code standards. The capability of the distribution system to stay connected even during voltage sag issues is termed as LVRT. A solar-wind-battery based hybrid renewable energy system (HRES) for microgrid applications is considered in this work which enables the use of renewable energy resources effectively, each and every system of HRES is controlled exclusively. The output of PV is boosted with the aid of a LUO converter which is controlled by a closed loop control based on Crow Search Algorithm. The wind energy conversion system utilizes doubly-fed-induction generator (DFIG), the output of which is converted to DC by a PWM rectifier and this is controlled by a PI controller. The battery system uses a bidirectional Buck-Boost converter and the state of charge (SOC) of the battery is monitored by artificial neural network (ANN). The key aspect of this work is the estimation of LVRT and this is accomplished by Signal processing approach based Deep Neural Network (DNN). Notch filter is used for pre-processing by which the noises are removed, Hilbert transform is used for segmentation and SIFT for feature extraction. The trained and test data are classified with DNN classifier from which the LVRT is estimated. The proposed strategy is implemented in MATLAB and the results were attained. The grid current THD is observed as 4.72% and the LVRT is estimated at 2.6sec.