Electric Vehicle Load Forecasting And Optimization Strategy Based On Markov And NSGA-Ⅱ Methods

Affected by global energy shortages and environmental pollution,electric vehicles are gradually replacing petrol vehicles.With the popularization of electric vehicles,their large-scale access to the grid for charging is about to become a trend,which will pose a huge challenge to the normal operation of urban power grids.Since a large number of electric vehicle charging will have a greater impact on the grid,it is necessary to predict and optimize the daily load of electric vehicles.To this end,this paper studies related issues in order to realize the daily load forecasting of electric vehicles and the optimization of peak-valley difference.First,Aiming at the problem of electric vehicle daily load forecasting,this paper proposes a forecasting method based on double-chain Markov and decision tree.According to Markov’s no-following nature,and considering the influence of the travel time of electric vehicles on the travel path,a double-chain Markov random path prediction model for electric vehicles is formulated.After simulating the path of the electric vehicle,the parking location,weather and state of charge of the electric vehicle are used as feature vectors,and the decision tree method is used to predict the charging method of the electric vehicle.According to the charging method selected by the electric vehicle,the charging time is calculated,and the Monte Carlo idea is used to predict the daily load distribution of the electric vehicle.Secondly,in view of the large peak-valley difference in the daily load distribution curve of electric vehicles,a corresponding optimization management strategy is proposed.When electric vehicles are operating in cooperation with generator sets,the electric vehicles participating in the discharge are scheduled to construct an optimization model that includes the smallest peak-valley difference between the daily load of electric vehicles and the lowest discharge cost of electric vehicle users and generator sets.Through the improvement of the elite strategy part of the non-dominated sorting genetic algorithm(NSGA-Ⅱ),the optimized NSGA-Ⅱ algorithm is used to solve the constructed optimization model to realize the peak-shaving and valley-filling of the daily load distribution of electric vehicles.Finally,simulations verify that the method proposed in this paper can better realize the daily load forecasting and optimization of electric vehicles.