| This paper proposes a photovoltaic clustering method for large-scale distributed PV user output prediction.Firstly,the impact of meteorology on PV output is divided into two categories:large climate and small climate:The large climate is mainly affected by sunshine or four types of weather types,which are divided by the average value of photovoltaic sunrise power,so that the historical data period is divided into four types of weather type sample groups;the microclimate is considered to be the influence of generalized microclimate such as photovoltaic installation elevation,temperature,humidity and surrounding geographical environment,and spatially related to the historical four types of weather type sample groups.Cluster analysis,the user PV area division is obtained;the number of user PV points and the sub-area meteorological consistency of the ungrouped users in the integrated block determine the optimal geographical block plan as the user PV grouping strategy,and the minimum deployment or multi-user for the meteorological site Space-time related power prediction provides a basis.Based on the proposed large-scale regional photovoltaic clustering method,the PV power plant with spatial correlation with the power station to be predicted in the cluster power station is further screened,and the ARMA model is established to predict the photovoltaic power generation output.Selecting the photovoltaic power station with correlation with the power station to be predicted as the correlation slave station,introducing the ARMA model for prediction and analyzing the prediction accuracy,indicating that the model has certain predictive ability;finally,all distributed PV users in the entire region The prediction and error analysis show that the method has universal applicability to wide-area distributed PV clusters. |
