Survey on renewable energy forecasting using different techniques

Abstract

Wind and solar are the renewable technologies which are very popular and well known source of energies throughout the world. Fossil fuels are formed by natural processes which contain a high quantity of carbon include coal, natural gas and petroleum which comes under non-renewable energy sources. Wind and solar Energy Forecasting is done to estimate the output power and energy of renewable energy sources. Forecasting is done at regular intervals to balance the supply and demand of energy. Solar and wind power forecasting are completely depends on metrological parameters such as velocity and direction of the wind, temperature, and humidity As solar and wind variability is stochastic, many of statistical models along with linear and non-linear models such as ARIMA, kalman filters, ANN, and support vector machines respectively used to catch the randomness of solar and wind energy. Lot of disadvantages are there for various approaches along with its computation complexity and incapability to alter the time varying time-series systems. This paper provides a comprehensive review of the theoretical forecasting methodologies for both solar and wind energy and also merits and demerits of different methods. The study of time series prediction of solar and wind power generation mainly focus on reviewing the advantage of using Long-Short Term Memory (LSTM) and Recurrent Neural Network (RNN).