Research On Load Modeling Of Electric Vehicles And Its Application

Large deployment of electric vehicles (EVs) can not only have a negativeeffect on security and stability of the grid but also provide ancillary services interms of frequency, voltage and peak load regulation, spinning service,compensating the intermittency of renewable energy etc. In other words, EVcharging is a “double-edged sword” to the grid. Whether the grid can play upstrengths and avoid weaknesses or not depends on the appropriate dispatch andcontrol to the charging process of the EVs. Hence, researching this problem haslarge significance. Also, this is the exerting point of the paper.Firstly, this paper deeply interprets the physical charging process as a diffusionprocess which includes a great deal of charging units transporting from low SOC tohigh with a stochastic disturbance variable constructed by paralleling speed ofelectric vehicles. With a detailed mathematic analysis of the movement of twoinfinitesimal energy volumes, the diffusion charging load model is proposed. Basedon a great deal of realistic and statistical data, the charging load of private electricvehicles is calculated through both the diffusion load state space model and themonte-carlo load model. The result shows that diffusion load model is as precise asthe monte-carlo load model. Compared to the monte-carlo load model, the diffusionload model divides the charging power into reference power and controllable powerand grasps the internal physical process between the charging loads at differentsequent times. Hence, the regional agent can integrate EVs into a “virtual powerplant” to supply concrete dispatch and control strategies for ancillary services basedon the hierarchical load control architecture.Secondly, this paper proposes an optimal charging strategy of electric vehiclesbased on diffusion load model, of which essential methodology is to minimize thepower loss during the charging time and optimally allocate the number of EVs ineach time block of dispatching period so as to optimal the charging power of eachtime block. This charging strategy considers four charging issues faced by both thegrid side and user side, ie, satisfying the grid operating conditions, contenting thetransporting demand of user, avoiding the damage on the battery and preventinglarge dimensions. The IEEE33distribution system is analyzed by bothuncoordinated and optimal charging in terms of comparing load profile, power lossand node voltage. The results show that the optimal charging strategy canremarkably achieve smaller peak-valley difference, higher load rate, lower powerloss and batter voltage quality and get a win-win between the grid and the user. Thirdly， this paper analyzes the problem of integrating EVs into ancillaryservices. The “virtual power plant” is aggregated by electric vehicles under thehierarchical load control architectures. Then this paper analyzes this issuemathematically utilizing the diffusion load model of electric vehicles and definesthe positive-definite Lyapunov candidate function of the tracking error between theACE desired power and controllable charging power. Then based on this candidatefunction, a closed-loop robust control system is derived with an appropriateconsideration of uncertainties among the number of electric vehicles and ACEdesired power. After proving the stability of this control system through theLyapunov stability theory, we examine the performance of the controller usingnumerical simulations on a real AEC desired power. The results show that thecontrol system ensures the “virtual power plant” achieve a precise response withouthaving negative effect on the normal using of EVs.