AUTHOR=Wang Shuaijie , Liu Shu , Guan Xin 

TITLE=Ultra-Short-Term Power Prediction of a Photovoltaic Power Station Based on the VMD-CEEMDAN-LSTM Model

JOURNAL=Frontiers in Energy Research

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2022.945327

DOI=10.3389/fenrg.2022.945327

ISSN=2296-598X

ABSTRACT=The prediction of photovoltaic power generation is helpful to the overall allocation of power planning departments and improve the utilization rate of photovoltaic power generation. Therefore, this paper puts forward an ultra-short-term power forecasting model of photovoltaic power station based on modal decomposition and deep learning. Methodology involved taking the data of a 50MW photovoltaic power generation system in Inner Mongolia Autonomous Region as a sample. Further, the weather conditions were classified, and the historical power data was decomposed into multiple VMF subcomponents and residual terms by VMD method. Then, the residual term  was decomposed twice by CEEMDAN method. All sub-components were sent to LSTM network for prediction, and the predicted value of photovoltaic power station   was obtained by superimposing the sub-component prediction results. ARIMA, SVM, LSTM, VMD－LSTM models  were built to compare the accuracy with the proposed models. The results  revealed that the prediction accuracy of non-combination forecasting model  was limited when the weather suddenly changed. The VMD method  was used to decompose the residual term twice, which  could fully extract the complex data information in the residual term, and compared with VMD－LSTM model, the eRMSE, eMAPE and eTIC of VMD-CEEMDAN-LSTM model  were reduced by 0.104, 16.596 and 0.038 respectively. The second decomposition technology has obvious prediction advantages. The proposed quadratic modal decomposition model effectively improvess the precision of ultra-short-term prediction of photovoltaic power plants.