Improvement for Forecasting of Photovoltaic Power Output using Real Time Weather Data based on Machine Learning
Jin-Hwa Jeong, Young-Tae Chae · 한국생활환경학회지 · 2018
This study investigated a selection of machine learning model to forecast electric power output from photovoltaic arrays based on forecasted weather data and historic solar radiation data. It tested two approaches to improve forecasting accuracy of power output with three typical machine learning algorithms such as Random Forest(RF), Artificial Neural Network(ANN), and Support Vector Machine(SVM). A forecasting power output was conducted with conventional weather forecasting data from national weather service which does not include solar radiation. The other approach has two steps, forecasting solar radiation with weather forecasting data and historic solar radiation data then it forecasts the electric power output of photovoltaic arrays. It has been studied the importance variables incorporated with the power output forecasting. The results show that the forecasting accuracy of the power output improves by using forecasted solar radiation data and Random Forest outperforms on this power output forecasting problem among other machine learning algorithms.