Research On Output Characteristics Of Photovoltaic Power Generation Cluster Considering Clustering Effect

Since the use of fossil fuels has decreased,solar energy has become a popular form of renewable energy.Due to the instability of solar power generation and other factors,the output of the power station is unstable in the case of a large-scale photovoltaic cluster connected to the grid,but the variation of the photovoltaic cluster’s overall output is less than that of each individual power plant,this indicates that as scale increases,the output curve of the cluster tends to have a smoother profile,that is,smoothing effect.Therefore,it is of practical engineering value to study the smoothing effect and convergence trend change of photovoltaic power generation clusters to reduce the difficulty of grid connection planning and configuration capacity.Based on the actual project support,this paper carries out the output analysis and application work considering the convergence effect.The specific work of this paper is as follows:First of all,the research method for the PV cluster convergence effect based on time-frequency analysis is designed.The wavelet analysis method is used to complete the division of photovoltaic frequency bands corresponding to frequency modulation and peak shaving operations in power grid operation.The time-frequency analysis index of PV convergence effect is constructed,and the analysis of PV cluster convergence effect in different seasons is completed through the index.The frequency bands that should be focused on in different seasons are given to guide the safe and stable operation of the power grid.The analysis of indicators based on the measured data reveals that the output volatility of PV clusters gradually declines as the range widens,the rate at which it does so tends to flatten out as the range widens,which verifies that the convergence effect has a certain trend.Secondly,a quantitative method based on statistical analysis of the convergence effect is proposed.In this paper,the continuous power curve is used as a statistical analysis tool to describe the convergence effect,and a combined forecasting model of the continuous power curve by scenario is established so as to complete the prediction of the continuous power curve of the expanded cluster,that is,the trend quantification of the convergence effect.Firstly,the hierarchical order of clusters with different scales is determined based on the hierarchical clustering algorithm.According to this sequence,the continuous power curve under different scales is divided into output scenarios,and a curve combination prediction model considering multiple error criteria is established under each scenario.Finally,the predicted curves under each scenario are spliced to obtain the continuous power curve of the cluster to be built.Through analysis of numerical examples,it is shown that the combined prediction model considering multiple error criteria is more suitable for quantitative analysis of the PV convergence trend in practical projects.Finally,a method for predicting the power of PV power generation systems with the consideration of the convergence effect is proposed.Three coefficient calculation methods are used to measure the similarity,consistency,and monotony between output sequences and determine the representative power station;The key features are extracted through Convolutional Neural Network-Long Short Term Memory Neural Network Model(CNN-LSTM)to realize the output prediction of the benchmark power plant;Considering the photovoltaic convergence effect,the fluctuation uncorrelation between the power stations is represented by the trend inconsistency coefficient proposed in Chapter 3,and the output prediction value of the representative power stations after linear amplification is corrected to finally obtain the output prediction value of the photovoltaic cluster.The results show that how well the reference power station selection method worked and verify that the CNN-LSTM model has more advantages in the prediction of representative power stations.They also effectively verify the accuracy of the photovoltaic power generation cluster prediction method considering the convergence effect.