A Hybrid topology for PV Forecasting and its application in Energy Management for Microgrid

Abstract Photovoltaic (PV) system is the most promising Renewable technology. PV Forecasting is needed due to the natural variation of meteorological variables such as solar radiations and climatic conditions. Due to this the power produced by a PV system is always non-linear. A hybrid forecasting approach has been presented in this paper. This hybrid approach is a combination of statistical approach, machine learning approach as well as physical approach. The machine learning approach uses a single layer and double layer perception concept based on Artificial Neural Network (ANN) whereas the statistical and physical approach used data driven formulation concept for forecasting. This data are based on historical analysis as well as they are also helpful in future forecasting. An implementation of a two-layer hierarchical model for Energy Management System (EMS) of islanded solar Microgrid (MG) is presented. The MG control employing forecast module and simulation process has been discussed briefly. The whole forecasting has been done on the basis of real-time data (RTD) of an industry and has been simulated in HOMER student’s version software.