Short Term Prediction Of Photovoltaic Power Generation Based On Improved Neural Network

With the continuous progress of global industrialization,the strategic plan for the development of the global energy industry has also undergone earth-shaking changes.After wind power generation,the photovoltaic(photovoltaic,PV)power generation technology has gained rapid development in recent years due to its clean,non-polluting,convenient installation,low maintenance costs,and high efficiency of use.PVs installed capacity and installed ratio show burst state growth.At the same time,PV output power has obvious randomness and uncertainty.When it is connected to the power grid on a large scale,its fluctuation characteristics are more prominent,which brings a huge impact on the power grid and reduces the reliability of the power grid operation,adding to the power grid dispatching operation management.Therefore,it is of great practical significance to predict the power of PV power system reasonably to improve the utilization rate of PV power station and the safe and stable operation level of power grid.In order to improve the precision of PV power prediction,the paper firstly summarizes the modeling and control methods of PV grid-connected system in detail,and deeply understands the energy conversion process of PV grid-connected process.Secondly,based on the historical running data,the internal mechanism of various factors(daily type,radiation intensity,humidity and temperature)that affect the power output of PV is deeply explored.Thirdly,the advantages and disadvantages of PV power prediction model and prediction algorithm are evaluated,and a combined prediction method based on(particle swarm optimization,PSO)algorithm analysis and neural network algorithm is proposed to predict PV power in the short term.Finally,the simulation results and the actual engineering power prediction calculations show that the proposed PV power prediction model can fully take into account the interaction of multiple types of PV power influence factors.The prediction results of the prediction algorithm are in good agreement with the actual operating power of PV and have good engineering generalization value.