Research On Large-scale Photovoltaic Output Power Characteristic And Its Forecasting Method Based On Measured Data

Nowadays the main approach to explore solar energy is to connect large-scale photovoltaic(PV)into the power grid in centralized method.With PV power plant capacity increasing,high permeability of PV power fluctuation will cause a series of adverse effects to the power grid.A comprehensive analysis of the operating characteristics and accurate prediction of output power of PV power plants will be an effective way to solve this problem.This paper is based on the measured data of large-scale PV power generation in Qinghai Province.The power characteristics of the PV power plant and the power characteristics of the power station group are analyzed.The PV power station's sunrise characteristics and the influence of weather and seasonal changes on the output power are studied.Construct the fluctuation characteristic index and analyze the fluctuation characteristics of photovoltaic power at different time scales and different installed capacity.From the perspective of power correlation between PV power plants,the clustering effect of PV power plants is revealed.The concept of clustering coefficient is proposed to measure the clustering effect of PV power plant group.Point out the application direction of clustering effect and put forward the forecasting method of PV power plant group using correlation correction.This paper introduces the prediction principle of gray neural network model applied to PV prediction and its applicability to PV prediction is analyzed.The original power sequence is smoothed based on the analysis results to improve the gray model and the defects of BP neural network are optimized by particle swarm optimization.Constructing improved grey neural network combined model for power forecasting in one day in advanced.The numerical results show that the prediction accuracy of the improved model is obviously improved compared with the gray neural network model.Based on the analysis of the correlation of PV power plants,a short-term power forecasting method for regional PV power plants considered the clustering effect is proposed.The method selects the base PV power plant and predicts it according to the correlation calculation result.The predicted value is linearly amplified as an estimate of the PV power plant group.Finally,according to the correlation coefficient between the photovoltaic power plants to correct the estimated value to achieve the short-term power plant photovoltaic group forecast.The results show that the prediction results of this method are closer to the actual value and the prediction accuracy is obviously improved compared with the conventional superposition method and the prediction accuracy of the regional PV power plant group is higher than that of the single photovoltaic power plant.