STUDY ON PHOTOVOLTAIC POWER PREDICTION OF VMD-FE-LSTM CONSIDERING SIMILAR DAYS
In order to solve the low prediction-accuracy issue caused by strong randomness and volatility of photovoltaic output,in this study,a combined photovoltaic electricity prediction model composed of similar day theory,variational mode decomposition(VMD)method,fuzzy entropy(FE)and deep learning algorithm was established innovatively.The grey relation analysis(GRA)method was firstly used to identify the critical meteorological factors affecting photovoltaic output;Secondly,the historical similar day of predictive day was selected by aid of the comprehensive similar distance method;Then,the photovoltaic output sequence decomposed by VMD method was reorganized based on FE calculation result,leading to several new sequences with strong regularity.Next,the long-short term memory(LSTM)neural network prediction model was formulated for each sequence;Finally,the predicted result was obtained through summing up the predicted value of each sub-sequence.The applied results of this combined model in the photovoltaic plant of Yunnan province demonstrated that,compared with other models,the proposed model had the high prediction accuracy and good prospect.
NETL
NSTL
万方数据
