Photovoltaic power forecasting techniques: Application in Artificial Neural Networks

Because photovoltaic energy is heavily reliant on weather and geographic regions, resulting in unpredictable fluctuations, integrating it into electric grids is complex. These discrepancies may lead to system imbalances, spikes, poor planning, and monetary losses. While forecasting methods can mitigate these problems, a number of factors must be taken into consideration, such as the forecasting horizon, input variable correlation analysis, input data processing, weather classification, network optimization, uncertainty quantification, and performance evaluation. In this article, we highlight the main prediction techniques applied to photovoltaic production. We focus on artificial neural networks (ANN), AutoRegressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM) and Support Vector Machine (SVM) regarding their implementation for predicting photovoltaic energy production. Also, we cover in this study all known forecasting horizons: very short-term forecasts, short-term forecasts, medium-term forecasts, and long-term forecasts. In addition, some performance evaluation metrics such as Mean Square Error (MSE), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination R2 are explained. At last, we demonstrate the usefulness of ANN in energy forecasting by applying it to photovoltaic power data.