Open AccessOne‐day‐ahead hourly forecasting for photovoltaic power generation using an intelligent method with weather‐based forecasting models
Author(s) -
Huang ChaoMing,
Chen ShinJu,
Yang SungPei,
Kuo ChungJen
Publication year - 2015
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2015.0175
Subject(s) - photovoltaic system , artificial neural network , computer science , radial basis function , weather forecasting , fuzzy logic , meteorology , electric power system , artificial intelligence , machine learning , data mining , power (physics) , engineering , geography , physics , quantum mechanics , electrical engineering
An intelligent method is proposed in this study to predict one‐day‐ahead hourly photovoltaic (PV) power generation. The proposed method comprises data classification, training, forecasting and forecasting updating stages. In the first stage, a fuzzy k ‐means algorithm is used to classify the historical data of daily PV power generation into various weather types. In the second stage, five training models are established, according to the verbal weather forecast of the Taiwan Central Weather Bureau (TCWB), in terms such as the sunny, sunny and cloudy, cloudy, cloudy and rainy and rainy. Each training model is constructed using a radial basis function neural network (RBFNN), for which the parameters of each RBFNN, including the position of the radial basis function (RBF) centres, the width of the RBFs and the weights between the hidden and the output layers, are optimised using a harmony search algorithm (HSA). In the forecasting stage, fuzzy inference is used to select an adequate forecasting model from the trained models. To cope with the possible fluctuation of PV power generation, the forecasts are updated every 3 h, according to the updated weather forecasts of the TCWB. The proposed approach is tested on a practical PV power generation system. The results show that the proposed method provides better forecasting results than the existing methods over 1‐year testing data.