Research On Load Forecasting Method Of Charging Station Based On Support Vector Regression Machine

During the peak period of power consumption,a large number of electric vehicles will be charged centrally,which will lead to a significant increase in the peak-valley load difference of the power grid,local overload in the distribution network,increase the operational burden of power equipment,and make it difficult to regulate and control the peak load of the power sys-tem.As an important part of the power load,the load forecasting of electric vehicle charging station is of great significance to the safe and stable operation of power system,and can also provide corresponding power data support for charging station itself.Firstly,by analyzing the data characteristics of charging load of electric vehicle charging station,the load characteristics type of charging station are selected.Then,a load forecasting model of charging station based on linear,self-defined polynomial and radial basis function is designed by using the method of selecting parameter combination.Through the analysis of ac-tual data,the appropriate characteristic data and load data are selected for example experi-ment.Aiming at the problem that the parameters of support vector regression machine model depend too much on human experience selection,this paper uses particle swarm optimization to optimize the parameters of support vector regression machine model,and synthetically im-proves the particle swarm optimization algorithm by adaptively adjusting inertia weight and acceleration coefficient,and constructs a load forecasting model based on IPSO-SVR charging station.The forecasting effect of the improved particle swarm optimization model is verified by an example analysis.Secondly?solving the problem that particle swarm optimization(PSO)may be trapped in local optimal solution region,this paper introduces grid search algorithm,uses its global ergodicity characteristics,combines the grid large step search method with the comprehensive improved particle swarm optimization(IPSO-SVR),proposes a parameter op-timization way of support vector regression machine(SVR)based on grid particle swarm opti-mization,achieves global optimization;constructs a charging station based on GS IPSO-SVR.The load forecasting model is compared with the experimental results of standard particle swarm optimization and improved particle swarm optimization model.The practicability and peculiarity of grid particle swarm optimization model in electric vehicle charging station load forecasting are verified.Through experiments,this paper proves that the radial basis function(RBF)has type ad-vantage in constructing the load forecasting model of charging station based on support vector regression(SVR).At the same time,the grid particle swarm optimization model can effectively solve the problem of particle swarm optimization easily sinking into local optimum solution and low eff-iciency of grid search.The integration of GS effectively improves the convergence accuracy of particle swarm optimization.IPSO optimization ensures the operation efficiency and prediction effect of the model.Therefore,the GS_IPSO-SVR model proposed in this paper has certain guiding significance for load forecasting of charging station.