Research On Forecasting Method Of Photovoltaic Power Output

With the vigorous economic development,environmental problems have become increasingly prominent.In this context,photovoltaic power generation has been vigorously developed.However,photovoltaic output power has problems such as randomness and volatility,when a large-scale photovoltaic grid is connected to the grid,it will damage the stability of the grid and affect the quality of power supply.Therefore,it is necessary to predict the photovoltaic output power.Short-term power prediction can not only reduce the harm to the grid,but also help the power dispatching department to arrange power generation plans,improve the grid’s ability to absorb photovoltaics,and benefit the development of the photovoltaic industry.This paper analyzes the main factors affecting the photovoltaic output power,and proposes two different prediction methods based on VMD-LSTM-RVM short-term photovoltaic power prediction and the combination prediction of GWO-SVM and random forest to predict photovoltaic power.Through simulation experiments,the effectiveness of the proposed model is verified.In view of the randomness and obvious volatility of photovoltaic power,a method combining variational modal decomposition(VMD),long short-term memory network(LSTM)and correlation vector machine(RVM)is proposed to realize short-term prediction of photovoltaic power.First,the VMD decomposition technology is used to decompose the historical photovoltaic power sequence into different modes to reduce the non-stationarity of the data;then,the LSTM prediction model is established for each mode,and the predicted value of each mode is reconstructed to obtain the power prediction At the same time,in order to further improve the prediction accuracy of the model,RVM is used to model and predict the error sequence;the predicted power value and the predicted error value are superimposed to obtain the final prediction result.Finally,the historical data from the Australian DKA Solar Centre site was selected for testing,and the simulation results verified that the method effectively improved the accuracy of photovoltaic power prediction.A single prediction model has its own limitations.Therefore,this paper proposes a combined prediction model based on gray wolf pack algorithm(GWO)optimization support vector machine(SVM)and random forest(RF).First,establish two single prediction models,GWO-SVM and RF,respectively,and then use the nonlinear mapping ability of random forest to adjust the weight coefficients to determine the weight of a single model,and combine the GWO-SVM and RF models to perform Forecast,get the predicted value of photovoltaic power.Finally,the historical data of the Australian DKA Solar Centre site is also used for testing.The experimental results show that the proposed model has a better prediction effect than a single prediction model.The various forecasting models and methods mentioned in this article are not only suitable for photovoltaic short-term power generation forecasting research,but also provide useful reference ideas for similar forecasting problems in other scientific and engineering fields.