A Machine Learning Based Approach for Estimating Renewable Energy Production

With the every increasing severity of global warming, climate change and depleting fossil fuel reserves globally, it is imminent that the world would need to migrate towards renewable energy rapidly. As the population of the world increases, the energy consumption would also increase in accordance. Hence, it is necessary to estimate the amount of renewable energy would we can produce in future so as to estimate the magnitude increase in renewable energy to meet total energy demands of the future. This in turn necessitates the pattern recognition of the renewables to total primary energy supply (TPES), through regression analysis. While conventional statistical algorithms have been employed thus far for the purpose, with the emergence of machine learning and deep learning algorithms, better forecasting accuracies have been achieved. This paper evaluates the renewable energy scenario and its salient features and presents a machine learning based approach to forecast future renewable generation so as to aid meeting energy demands of the future through renewable energy sources. Different renewable energy sources have been analyzed such as wind, solar, biomass and hydro. The forecasting MAPE has been computed as the performance metric.