Short-term Power Prediction Of Photovoltaic Based On EEMD And Integrated Neural Network

Today,with the rapid development of the world’s industry and economy,the environment is deteriorating and the energy resources are depleting.The clean energy such as new energy represented by wind power generation and photovoltaic power generation is receiving widespread attention from the people of the world.Compared with other renewable energy sources,photovoltaic power generation has outstanding characteristics and advantages.After the establishment of the power generation system,the problem of erecting transmission lines in remote areas has been effectively solved.However,photovoltaic power generation has some disadvantages due to its inherent properties.Photovoltaic power generation itself is also intermittent,and it is very intense due to meteorological factors such as irradiance.It is inherently random and uncertain,and its grid connection affects the stability of the entire power grid.In order to overcome this shortcoming,reduce the impact of photovoltaic grid-connected grid on the grid,and at the same time contribute to the energy-saving dispatching and economic operation of the grid,high-precision prediction of PV output power is needed.It can be seen that it is of great significance to improve the prediction of photovoltaic power.At present,a considerable number of researchers at home and abroad have studied this and achieved certain results,but the prediction accuracy has not reached a particularly perfect level and needs further research.Based on the previous research results and the basic theory of photovoltaic prediction and the analysis of influencing factors,this paper proposes a short-term power rolling prediction model based on empirical modal decomposition algorithm and integrated neural network.Firstly,the clustering algorithm and discriminant algorithm are used to select the similar day according to the daily feature index.Secondly,the power sequence and the influencing factor sequence are decomposed by the empirical mode decomposition algorithm,and the permutation entropy algorithm is used to reconstruct the high frequency.The three sequences of low frequency and trend items;again,the Elman neural network is optimized by a particle swarm optimization algorithm with simple and flexible,fast convergence,good optimization effect and wide application.Thus,the optimization parameters of the Elman neural network are obtained: the feedback gain factor and all the weights are connected for prediction.Finally,for the high frequency,low frequency and trend items,the integrated prediction Elman neural network is used for rolling prediction,and then the predicted values of the three are added to obtain the predicted value of the photovoltaic power.Through the prediction of the proposed model and the comparative analysis of other models,it is concluded that the model highlighted in this paper is very suitable for PV power prediction and has a good prediction effect on the data prediction.