Research On The Prediction Method Of Photovoltaic Power Generation And Electric Vehicle Charging And Discharging

At present,with the deepening of economic and social development,the consumption of traditional energy sources continues to grow,and the use of polluting energy sources and the environment are deteriorating.The crisis of energy supply and environmental degradation has broken out in the world.As a result,renewable energy and electric vehicles have emerged,becoming the "darling" of the world.Due to the instability of new energy generation and the increase in electric vehicle load,it will inevitably have an adverse impact on the microgrid,triggering a new round of power problems.Therefore,the prediction of the power generation of new energy and the charging and discharging of electric vehicles is crucial to the safe and stable operation of the power system.This paper studies the two aspects of photovoltaic power generation forecasting and electric vehicle charging and discharging forecasting.On the one hand,aiming at the power generation characteristics of photovoltaic systems,this paper proposes an Elman neural network photovoltaic power prediction method using association rules and kernel principal component analysis.First,a gray correlation analysis method is used to select several similar days that are highly similar to the predicted daily meteorological features as training samples;Secondly,the kernel principal component analysis method is used to reduce the dimension of the input number of the training samples,and extract the principal component sequence;Then,the Elman neural network optimized by ant colony algorithm is used to predict the data.Finally,aiming at the large error of Elman neural network prediction model at the peak of power fluctuation and the volatility of prediction,the Markov method is used to modify the preliminary predicted value and obtain the final predicted result.On the other hand,there has a strong randomness in time and space with the behavior of the vehicle owner in the electric vehicle charging and discharging,the electric vehicle load forecasting model is proposed to consider a different charging mode ratio.Firstly,the charging area is spatially divided into office areas,residential areas,and entertainment areas according to regional functions.Secondly,electric private cars are randomly sampled to obtain daily travel distances from different locations,and to calculate battery load status.Then,the charging mode of vehicles that need to be charged is judged,and the starting charging time is extracted according to the charging mode.Finally,the Monte Carlo algorithm is used to calculate the load of private cars under different requirements.It is verified that the charging and discharging of electric vehicles are closely related to the vehicle size,spatial distribution,charging behavior and charging mode of users.