Robust Optimal Dispatch Of Islanded Microgrid Based On Interval Prediction Of Electric Vehicle Charging Load Scene

Electric vehicle(EV)charging load has a strong randomness of time series.With the continuous increase of EV,the randomness of EV’s disordered charging poses a significant challenge to islanded microgrid dispatching’s economy and reliability of the power supply.Simultaneously,renewable energy sources are inherently unstable,and their output power is volatile and uncertain.How to reduce the impact of the strong randomness of EV charging load on the economy and reliability of the power supply of islanded microgrid operation is a complex problem to be solved.Aiming at this problem,a more effective EV charging load interval prediction method is proposed.An islanded microgrid robust optimization dispatching model that considers the constraints of EV users charging satisfaction is constructed.In order to reduce the impact of the strong random charging load of EV on the operation of the islanded microgrid,an interval prediction method of EV charging load based on scene generation with multiple correlation days is proposed.Firstly,considering that EV charging behavior is greatly influenced by historical charging behavior,the Spearman rank correlation coefficient is used to analyze the correlation between the EV charging load on the day to be forecasted and the EV charging load on the historic day.Find historical days that have an extremely strong correlation with the day to be forecasted.The original multi-correlation day charging scene set is constructed to describe the charging behavior of EV.Then,massive generating multi-correlation day charging scenes with similar probability distribution and different timing distribution with the original multi-correlation day charging scene set are generating based on improved variational autoencoder.Finally,appropriate scenes with high relevance to EV charging load data of the known historical days are selected from generating a multi-correlation day charging scene set to constitute a related scene set.According to the data average and data interval of the last day of the related scene set,the deterministic prediction result and the interval prediction result of EV charging load of the day to be forecasted are obtained.Aiming at the islanded microgrid containing wind turbine(WT)and photovoltaic(PV),a robust optimization dispatching scheme for the microgrid that considers the demand response of EV is proposed to improve the reliability of the power supply of the microgrid and the operating economy.Firstly,in traditional research,EV participate in microgrid dispatch as a fully responsive load.Whether EV users participate in the EV demand response scheme formulated by microgrid operators is considered in the model,which depends on the subjective will of EV users.The charging satisfaction constraint of EV users is defined to ensure EV users’ enthusiasm to participate in demand response.Then,considering the uncertainty of EV charging load,WT output,PV output,and residents load,a day-ahead robust optimization scheduling model that considers EV demand response and EV users charging satisfaction constraints is constructed.Finally,a column constraint generation algorithm is used to obtain the day-ahead robust optimal dispatching plan for the islanded microgrid.The data used is the plug-in charging profile of EV households randomly selected from the residential energy consumption survey data set in the Midwestern United States.The prediction effects of the new prediction method and the traditional probability prediction method-Gaussian process regression method are analyzed separately.The contrast experiment proves that the new way has a more reliable prediction interval and a narrower prediction interval.For islanded microgrid with a relatively high proportion of renewable energy,a two-stage day-ahead robust dispatch model takes into account the strong randomness of EV charging load,wind power output volatility,photovoltaic output volatility,and residents load volatility,is formulated.The column constraint generation algorithm is used to obtain a robust dispatch scheme under multi-source-load prediction interval.Simulation results show that the day-ahead demand response dispatch model of EV can improve the power supply reliability and renewable energy utilization rate of the islanded microgrid and reduce the power generation cost of the islanded microgrid.