Open AccessDay-ahead solar irradiance forecast using sequence-to-sequence model with attention mechanism
Author(s) -
Sowkarthika Subramanian,
Yasoda Kailasa Gounder,
Sumathi Lingana
Publication year - 2022
Publication title -
indonesian journal of electrical engineering and computer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.241
H-Index - 17
eISSN - 2502-4760
pISSN - 2502-4752
DOI - 10.11591/ijeecs.v25.i2.pp900-909
Subject(s) - computer science , recurrent neural network , sequence (biology) , artificial neural network , encoder , irradiance , solar irradiance , artificial intelligence , deep learning , real time computing , machine learning , meteorology , physics , quantum mechanics , biology , genetics , operating system
The increasing integration of distributed energy resources (DERs) into power grid makes it significant to forecast solar irradiance for power system planning. With the advent of deep learning techniques, it is possible to forecast solar irradiance accurately for a longer time. In this paper, day-ahead solar irradiance is forecasted using encoder-decoder sequence-to-sequence models with attention mechanism. This study formulates the problem as structured multivariate forecasting and comprehensive experiments are made with the data collected from National Solar Radiation Database (NSRDB). Two error metrics are adopted to measure the errors of encoder-decoder sequence-to-sequence model and compared with smart persistence (SP), back propagation neural network (BPNN), recurrent neural network (RNN), long short term memory (LSTM) and encoder-decoder sequence-to-sequence LSTM with attention mechanism (Enc-Dec-LSTM). Compared with SP, BPNN and RNN, Enc-Dec-LSTM is more accurate and has reduced forecast error of 31.1%, 19.3% and 8.5% respectively for day-ahead solar irradiance forecast with 31.07% as forecast skill.