머신러닝을 이용한 태양광 발전량 예측 모델 비교

Abstract

The amount of new renewable energy generation is increasing worldwide every year. Among many new renewable energy sources, solar energy generation using solar energy accounts for the highest proportion of new renewable energy generation. There is a variation in power production because solar power generation is more affected by climate conditions compared to power generation using crude oil or oil. In order to accurately predict solar energy generation dependent on climate variables, this study compares the performance of machine learning-based solar power generation prediction models using weather forecast data from the current forecast technology, Numeric Weather Prediction (NWP). In this study, we experimented on two NWP types, and 7 machine learning models depending on 21 photovoltaic(pv) power stations. Based on results, we select the model with the lowest statistical indicators nMAE(%) by region as the optimal model for the region. Finally, experimental results show that the 7-Block ANN model devised in this study is better than conventional machine learning models.