Solar Irradiation Forecasting - Comparative Analysis of Various Methods

The global energy system is transforming with the consideration of a growing population with rising living standards that will need more energy. Simultaneously, the world must find ways to reduce greenhouse gas emissions and provide sustainable energy production. As electricity is the fastest-growing part of the energy system and thus shifting to renewable energy production systems is the need of the hour. IEA reports that the Sun could be one of the largest resources for energy production for the upcoming ‘net zero goals’ of the world. Rapid decrease in the cost of installation and availability of solar radiation are major advantages for adoption, although this mode of generation of energy has a variable output which is influenced by several parameters. Hence solar energy forecasting is crucial to understand and predict the output of production, the demand of the consumers, and optimizing the dispatch of the electricity from grids to users. In this paper, we have analyzed innovative methodologies for forecasting solar energy radiation. Moreover, it provides a review and comparative analysis on various Artificial Neural Networks (ANN), Probabilistic and Statistical, and Time-Series based models used for estimating solar irradiation. Each of these models have been trained and tested in different geographical regions, weather conditions, and corresponding relevant parameters, which would provide an insight on efficient model selection and optimization for predicting short and long-term solar irradiation. Further, we have discussed some challenges while forecasting solar irradiation and future research direction in this domain.