Short-Term PV Power Prediction Considering Weather-Coupled Similar Days

ObjectivesTo make full use of historical information, maximize the optimization of model effect and improve the accuracy of photovoltaic (PV) power prediction, a short-term PV power prediction method considering weather-coupled similar days is proposed.MethodsFirstly, fuzzy C-means clustering is used to divide the dataset into different weather types, and the correlation weight factors are calculated according to the affiliation and features selection of the day to be predicted for each weather type, and the similarity of the historical days is calculated by grey relation analysis, and similar days with weather coupling are selected. The similar days are decomposed into modal components with different frequencies by variational mode decomposition to realize further denoising. Secondly, in order to fully utilize the nonlinear fitting ability of the model, the red-tailed hawk algorithm (RTHA) is used to optimize the hyperparameters of the bidirectional long short-term memory (BiLSTM) network model, and the RTHA-BiLSTM model is constructed to predict the modal components. Finally, taking the actual data of a power plant in Jiangsu Province as an example, the simulation experiment is carried out to verify the effectiveness of the proposed method.ResultsIn sunny, cloudy and rainy scenarios, compared with the method without similar days, the proposed method reduces the root mean square error by 9.1%, 6.1%, 2.9% and 11.1%, 6.5%, 13.9% in the single model and the combined model, respectively.ConclusionsThe proposed method can effectively improve the prediction accuracy of PV power, has good robustness and strong prediction ability, and can better cope with the prediction tasks in different scenarios.