An advanced data driven model for residential electric vehicle charging demand

As the electric vehicle (EV) is becoming a significant component of the loads, an accurate and valid model for the EV charging demand is the key to enable accurate load forecasting, demand respond, system planning, and several other important applications. We propose a data driven queuing model for residential EV charging demand by performing big data analytics on smart meter measurements. The data driven model captures the non-homogeneity and periodicity of the residential EV charging behavior through a self-service queue with a periodic and non-homogeneous Poisson arrival rate, an empirical distribution for charging duration and a finite calling population. Upon parameter estimation, we further validate the model by comparing the simulated data series with real measurements. The hypothesis test shows the proposed model accurately captures the charging behavior. We further acquire the long-run average steady state probabilities and simultaneous rate of the EV charging demand through simulation output analysis.