Abstract:With the increasing demand for energy efficiency and the pohtovoltaic data that is increasing coustantly,traditional PV forecasting methods have gradually lost their advantages.In order to make more accurate PV predictions,a deep learning framework is adopted.Due to the powerful processing power of time series,the most important structure in the recurrent neural network (RNN)-long-short-term memory network (LSTM) is used for intelligent algorithm modeling.Because LSTM has the functions of "forgetting" and "updating",it solves the problem of long-order dependence,which makes the PV prediction have a qualitative change in accuracy,and the prediction speed is also significantly improved.