Short-term photovoltaic power combination prediction based on DBO-VMD and CPO-LSTM

Abstract Regarding how to promote the accuracy of photovoltaic power generation prediction, a photovoltaic power prediction model based on Dung Beetle Optimizer (DBO), Variational Mode Decomposition (VMD), Crested Porcupine Optimizer (CPO), and Long Short-Term Memory Network (LSTM) is established. First, the DBO is used to optimize the VMD parameters. Then, the power data is decomposed by using VMD. Secondly, an LSTM prediction model is established for each decomposed component, and the hyperparameters of the LSTM model are optimized by using CPO. Last, each component model is trained and tested, and the final power prediction result is obtained by superimposing the prediction results of each component. The experiment shows that the model can achieve good prediction performance in different seasonal types and compares it with other models to verify the effectiveness of this model.