서포트 벡터 회귀를 이용한 24시간 앞의 태양광 발전량 예측

Since photovoltaic(PV) power system is subject to a lot of impacts on weather conditions especially solar insolation, it is difficult to predict photovoltaic power generation. Accurate forecast of 24-hour ahead PV power output makes stable link to Korea Power Exchange and Smart Grid power system and enable efficient energy management. In this study, actual historical power output and meteorological information data from 2013 to 2014 year were collected. First we establish the solar insolation forecasting model using Support Vector Regression(SVR) of machine learning algorithm, and second establish a SVR based model to predict PV power output from the predicted solar insolation. Also, by analyzing the effects of factors that inhibit the PV power prediction, the general performance of the model for predicting 24-hours ahead PV power output was derived.