Research On Pv Power Forecasting Based On Adaptive Modeling

Influenced by multiple factors including natural environments and the operating conditions of the devices,the photovoltaic power involves great randomness and fluctuation.In the power system with strict requirements of supply and demand balance,the grid-connected operation in the large-scale photovoltaic power plants brings grave challenges to the safety and stability of the power system.Timely and precise predication about the working conditions in the photovoltaic power plants for a certain time in the future is of great value to the operation and maintenance of the power grid.The rotation of day and night as well as the seasonal climate change heavily affect the Photovoltaic power,and there are certain regulations about the overall working conditions.The adaptive modeling of the photovoltaic power prediction model can effectively enhance the prediction precision of the photovoltaic power prediction model.In light of the above reasons,this paper engages in the following researches:(1)Influenced by natural factors,there are certain changing regulations in the photovoltaic power.According to the above features,we adopted the method of t-distributed stochastic neighbor embedding(t-SNE)and Density-Based Spatial Clustering of Applications with Noise(DBSCAN)to make clustering analysis on the changing conditions of the photovoltaic power in different climate change environment conditions.We made good preparations for the adaptive modeling of the following prediction model when reducing the complexity of feature division.(2)Integrated with the result of clustering analysis,we adopted the Extreme Learning Machine(ELM)based on the Particle swarm optimization(PSO)to build adaptive modeling of the photovoltaic power prediction model.At the same time,the setting results of different model parameters will exert a certain influence on the prediction precision of the photovoltaic power.Therefore,the phased optimizing setting on the training parameters of different models during the training process improved the final prediction precision of the photovoltaic power prediction model.(3)With the actual algorithmic examples,this paper involves the comparisons and analyses about the model in this paper and the traditional,unadaptive modeling and the prediction result of the optimized photovoltaic power prediction model,further demonstrating the effectiveness of the model in this paper on the prediction of the photovoltaic power.