A Comparative Analysis of Artificial Neural Network Algorithms to Enhance the Power Quality of Photovoltaic Distributed Generation System Based on Metrological Parameters

This paper examines the solar irradiance estimation as well as power quality enhancement of photovoltaic distributed generation system as seen from a metrological perspective. The enhancement of power quality is fundamental considerations. In this paper, the artificial neural network has been trained on historical data for solar forecasting and to anticipate reference current for the controller to integrate the power quality enhancement features. Six algorithms have been used for training. Metrological parameters were used to determine the optimum training of neural network. Training results of some of the algorithms were found approximately same with minor difference in terms mean square error and number of epochs. Scaled conjugate gradient algorithm shows least mean square error 0.01611 among six distinct algorithm. Results of the ANN model show that the comparable models are obtainable from different training algorithms. Using the power quality enhancement measure, a neural network estimates the reference current required for uninterrupted power transfer across systems. With a R = 25 Ω, L = 30 mH load, the harmonic content of the grid current was decreased from 19.69 to 4.20% using power quality enhancement controller based on an artificial neural network. Statistical parameter has been used to plot the error for a visual comparison of system efficacy.