Super Short-term Power Prediction Of Distributed Photovoltaic Power Generation Systemmethod Study

In recent years,the state has vigorously promoted solar energy technology to cope with the increasingly serious energy and environmental crisis.Among them,photovoltaic power generation technology has become the main direction of current research.However,due to the limitation of natural conditions,the power of photovoltaic generation is stochastic,intermittent and fluctuating.The existence of these characteristics will damage the stability of the power grid and have great influence on the power quality.Therefore,studying the power characteristics of photovoltaic power generation and reasonably and effectively predicting the power of photovoltaic generation are of great significance for safe and stable operation of power grid and high quality power supply.In this paper,the main problems of existing photovoltaic power prediction methods are analyzed,and the photovoltaic power prediction method based on BP neural network and Elman neural network algorithm is studied,and the super short-term prediction method of photovoltaic power generation by improving the Elman neural network algorithm is proposed.First,the basic situation of the photovoltaic power station test platform is introduced,and the relevant data related to the power generation power of the photovoltaic power station are obtained through the monitoring platform,and the correlation analysis of these data is carried out.The correlation coefficient of the power generation power of the photovoltaic power station and its influence factors is obtained,which lays a foundation for the subsequent research.Set data support.Secondly,most of the weather types are obtained through the weather forecast at home and abroad,but the weather forecast is often aimed at large range weather types,and the weather types in the area where the photovoltaic power station is located belong to the small range weather type.In view of this situation,the current real-time environment data is monitored,and a super short-term forecast model of weather type based on support vector machine is established,and the super short-term weather type in the area of the photovoltaic power station is predicted by the model.Thirdly,the traditional BP neural network and Elman neural network are analyzed,and the super short-term prediction model of photovoltaic power generation power based on BP neural network and Elman neural network is established respectively.The two neural networks are obtained by comparison and analysis of the prediction results.Finally,aiming at the shortcomings of traditional neural network,an improved Elman neural network super short-term power prediction algorithm is proposed.First,the weather type correlation coefficient is used to initialize the input of the predicted data,and the malat algorithm is used to decompose the initialized data.Then the decomposed data is processed by Elman neural network.Finally,the malat algorithm is used to reconstruct the data and get the final prediction results.The effectiveness and feasibility of the proposed algorithm is verified by modeling and forecasting the PV power of an actual PV power station.