International Conference on Artificial Intelligence: Technologies and Applications (ICAITA 016) Optial Planning for Electric Vehicle Charging Station Considering the Constraint of Battery Capacity Ming Zeng1,, Xiaohui Zhan1, and uanfei Li1, 1 State Key Laboratory for Alternate Electrical Power Syste with Renewable Energy Sources, orth China Electric Power University, Beiing 1006, China School of Econoics and Manageent, orth China Electric Power University, Beiing 1006, China the distribution of resident load. AHP ethod is used to calculate the weight coefficient of each candidate station, based on which the optial econoic odel of EV charging station is established. However, all these ethod fail to take soe factors, especially factors fro VE itself, which affect the charging station planning into account when building atheatic odels. Abstract Planning EV charging stations reasonably is significant to the developent of whole EV industry. Taking constraints of charging capacity and investent liitation into account, the optiization odel of EV charging station planning with the obective function of iniu annual cost is established. Constraint of service radius is replaced by constraint of battery capacity which takes the factor of EV itself in order to eliinate the possible planning failure. The odel is solved by the chaos and harony search algorith and the optial result of the charging stations planning is obtained. Case analysis shows that the ethod proposed in this paper has a certain practical and scientific, and can provide soe reference for the planning of EV charging stations. Aiing at this proble, this paper, based on the existing planning ideas, will introduce capacity constraints of EV into the odel. Therefore the axiu econoic benefits of the charging station location, which takes both the traffic network and the electric vehicle itself into account can be realized. Using chaos and harony search algorith to solve the odel. The scientificalness of the proposed ethod is deonstrated by a case analysis. Keywords-electric vehicle; charging station; battery capacity, chaos and harony search algorith I. ITRODUCTIO II. With the econoic developent and the growing contradiction between energy supply and environent pollution, the existing developent odel is unsustainable. So social econoic ust shift to the odel of high efficiency and low eission. Against this background, the research and discussion of the global power industry about icrogrid, EVs and so on, has becoe the hot topic in recent years [1-] As an iportant part of the EV facilities construction, the construction of EV charging station is essential to the developent of entire EV industry. EV charging station planning is a nonlinear optiization proble with ulti variables and ulti constraints. In order to solve this kind of proble, any ethods have been proposed. Coon ethods include atheatical optiization ethods (such as linear prograing [3] and nonlinear prograing [4]), heuristic optiization ethods and intelligent optiization ethods (such as tabu searching ethod [5], genetic algorith [6], siulated annealing ethod [7], and particle swar optiization ethod [8], etc.) Many ethods for solving the proble of the planning of EV charging stations were presented by soe papers. Paper [9] analyses factors that affect the capacity of charging station. In paper [10], through the analysis of the electric vehicle charging deand, the factors affecting the electric vehicle charging station planning are proposed, and the principles of the layout planning are put forward. Paper [11] siulates the nuber of EV according to 016. The authors - Published by Atlantis Press 349 MODEL FOR EV CHARGIG STATIO PLAIG A. Cost iniization Obective Function Fro the point of view of atheatics, the proble of EV charging station planning is a typical proble of location selecting and capacity deterining. In the area of traffic and atheatics, there have been a very good location theory to help decision akers or analysts to weigh different planning obectives [1]. The traditional planning idea is to divide the charging deand area in the selected planning area. Then, the total annual cost iniization obective function is constructed. By optiization, the capacity and location of EV charging stations can be deterined. According to that idea, the atheatical odel can be described by following equations. in C C1 C C3 (1) C1 A [ 1 i (1 i )n ] (1 i ) n 1
C A (1 ) 1 (3) E y PD bb b (7) t fr( E )cos u y d C3 c0d bsbbb 1 bb (4) In equations above, C 1 stands for the annual fixed cost of building a charging station; C stands for the annual operation and aintenance costs of charging stations; C 3 stands for the annual charging costs of EV users. stands for the nuber of charging stations needs to be built; A stands for the present value of building charging station ; i stands for the discount rate; n stands for years in return of capital investent, stands for variable, 1 eans that the charging station is selected; stands for conversion coefficient; considering EV will loss while driving, c 0 stands for the loss cost of a y single EV in unit distance traveling; D stands for the nuber of days per year, d stands for the ties of charging per user on average, which can be deterined by the following equation. d ph (5) Q p stands for the consuption of electricity of EV per hundred kiloeters; h stands for daily driving distance of EV on average; Q stands for the battery capacity of EV. B stands for the collection in which users in the counity b get their EVs charged in station. is 0-1 variable, 1 eans b that users in the counity b only goes to station in a certain tie period. Besides, this variable should satisfy another equation as follows. 1 1 b b (6) S b stands for the distance between the counity b and charging station ; b stands for the road condition coefficient; stands for the nuber of EV in the counity b. B. Constraints 1) Constraint of charging capacity of charging stations bb P E e( E )cos (8) b In equations above, E stands for the capacity of the charging station. P stands for the charging power of each EV. t u stands for the daily tie under charging ode u. At present, two odes are considered, which are fast charge and conventional charge. stands for charging efficiency. f stands for the deand factor of charger; ee ( ) stands the load rate of charging station; cos stands for the power factor of charging station. ) Constraint of investent liitation 1 M stands for the liitation of investent. A M (9) 3) Consideration of EV battery capacity constraint Most of the existing planning ethods, only consider the service radius of charging stations and take it as a constraint. However, in urban traffic, the service radius and the length of the EV driving path are different. ot considering the characteristics of the traffic network will lead to increased driving distance of electric vehicles, soeties beyond the ileage of EV. An exaple is given in Fig.1. A user lives in point A, there are charging stations in both point A and B which service radius are 5KM. The ileage of EV is 50KM. It s obviously that this station planning can satisfy the service radius constraint. However if the user wants to drive to point C and returns back to point A. The actual driving distance of EV is 80KM, which is longer than 50KM, the distance between point A and B. A C B 40KM 10KM FIGURE I. CASE OF PLAIG FAILURE At this circustances, the user will have to drive to point B and get his EV charged or he can t return back to point A. Therefore, if there is only one station on the driving path, it s possible that EV can t reach the destination or coplete a return trip soothly (tie of user is wasted in this exaple, it can be worse in other cases). To solve this proble, this paper replace the service radius constraint with the battery capacity constraint based on the structure of traffic network. It s assued in this paper that the electric quantity of EV at starting point (F) is x% Q (If starts fro charging station, 350
assue EV is full charged.) Then EV drives to the nest point (T) on the path. The consuption of this distance is, so EF T the electric quantity arriving at next point is ET EF EF T. If there is a charging station on this point, ET Q. When the reaining power is not enough to aintain EV traveling to the next node, or can not return on any point on the path to last node, consider that the EV charging station planning can not eet the charging needs of this path. Considering the battery capacity constraint, an inspection procedure will be added in order to test the planning results of the traditional ethod. III. CHAOS HARMO SEARCH ALGORITHM In this paper, the Chaos Harony Search Algorith (CHS) is used to solve the traditional ethod [13]. In this practical proble, the harony of usical instruent tones in CHS represents the optial operation of each equipent. The aesthetic evaluation of the harony corresponds to the function value of the obective function. The worst harony is the vector of using capacity of each device which axiizes the obective function value. And the best harony is the vector of using capacity of each device which iniizes the obective function value. The calculation steps of Chaos harony search algorith as follows: 1) Initialization paraeter: the initial value of tones(the harony of each usical instruent tone represents a group of Chaos), the harony eory(hm), the nuber of harony which can be saved in HM(HMS), the largest nuber of iterations( ax ), the retention of harony eory(hmcr), eory disturbance probability(pr) and so on. ) Initialize the harony eory: First, using chaotic systes to apping the ergodicity of chaos of equation when the control paraeter u=4. Then, initializing the feasible solution, and preferentially selecting HMS solutions into HM as the initial solution group of the algorith to initialize the feasible solutions. 3) Through the retention of harony eory, rando selection of tones and rando disturbance, it will produce a new solution X new, and copared with the worst solution of harony eory, then eliinating the poor to update the HM. 4) When the nuber of iterations has reached the axiu nuber of iterations ax, the output is the optial result, and the algorith is over. Perfor inspection procedure on the obtained optial result. If the constraint of battery capacity is satisfied, then take the result as final result. Unless, re-initialize the harony eory bank and perfor the CHS again. The specific process as shown in figure 3: FIGURE II. ISPECTIO PROCEDURE OF BATTER CAPACIT COSTRAIT 351
Start Initialize harony eory bank HM =1 Creat rando nubers within (0,1) rand<hmcr? Variable selection within HM Perfor rando perturbation on the variable with probability PR =+1 Variable selection within Peritted zone <? ew solution X new better then the worst one in HM? Replace the worst solution in HM with X new The nuber of iterations has reached ax? Inspection procedure Battery capacity constraint satisfied? FIGURE IV. TRAFFIC ETWORK AD LOCATIOS OF CHARGIG DEMAD COMMUITIES AD CHARGIG STATIO CADIDATES End FIGURE III. THE SPECIFIC PROCESS OF CHS AD IPECTIO PROCEDURE IV. CASE AALSIS The proportion of planning area is 10.5k. The nuber of EV during planning tie period is 190. The coordinate and nuber of each counity is given in table 1 and the coordinate of charging station candidates are given in table. The location of charging deand counities, charging station candidates and the traffic network are showed in figure 4. It s assued that the nuber of EV doesn t change during planning tie period, the ratio of fast charge user is 80% and the charging power of this ode is 50KW; the ratio of conventional charge user is 0% and the charging power of this ode is 0KW. On average, every EV is charged for every two days. TABLE I. EV OWERSHIP AD LOCATIO OF CHARGIG DEMAD COMMUITIES uber of counities Abscissa Ordinate Ownership of EV 1.30 1.19 130.04 3.10 85 3.07 1.0 13 4 1.9 3.3 100 5 1.1 3.4 10 6 1.1 3.01 75 7 1.4.54 80 8 1.65 1.45 130 9 0.19 3.50 95 10 0.86.40 90 11-0.44 3.90 9 1 0.08 3.41 78 13 1.4 1.73 75 14 0.53 1.59 10 uber of counities Abscissa Ordinate Ownership of EV 15-0.75 3.15 13 16-0.4 0.74 37 17-0.14.80 117 18-0.73 1.73 100 19 0.84 1.36 107 0-0.57 1.96 113 35
TABLE II. LOCATIOS OF CHARGIG STATIO CADIDATES uber of uber of Location candidates candidates Location 1-0.59,.84 6 0.71, 1.84-0.4, 3.36 7 0.75, 1.0 3-0.63,.4 8-0.11, 1.10 4 1.63, 3.16 9-0.47, 1.45 5 1.80,.56 10 1.84, 1.38 Other paraeters in the case are given in table 3. TABLE III. OTHER PARAMETERS I THE CASE Paraeters Value Paraeters Value Charging station Power loss per construction basic 100 0 100KM(KWH) cost(10000uan) Charging station construction capacity cost(10000uan/kva) 0.5 Investent liitation 5000 Load rate of stations 0.75 ears in return of capital investent(year) 0 Discount rate 0.1 Deand factor 0.8 Charging efficiency 0.9 Average daily distance(km) Road condition coefficient Annual charging ties Charger siultaneity rate Loss cost(uan/km) Battery capacity of VE(KWH) Initial electric charge for EV(%) Maxiu iteration liit 00 All paraeters are substituted into the odel and CHS algorith is used to solve it. After 600 ties iterations, the third planning obtained is able to satisfy the constraint of battery capacity. This eans that the battery capacity constraint can, to a certain extent, avoid the proble of the service radius constraint. The location and capacity of charging station after optiization are shown in table 4. 00 1. 180 TABLE IV. LOCATIO AD CAPACIT OF CHARGIG STATIOS AFTER OPTIMIZATIO uber Location Capacity -0.4, 3.36 900 4 1.63, 3.16 3000 6 0.71, 1.84 3000 9-0.47, 1.45 3700 10 1.84, 1.38 3000 As can be seen fro the table above, after considering all constraints, the deand for EV charging in the planning area can be et by the above 5 charging stations, the total annual cost is 1. illion uan. It can also be seen that due to the different geographical location of charging stations, the planning of the capacity is not the sae, which is conducive to the full and efficient use of resources. 0.7.5 60 50 and the traditional service radius constraint is replaced by the battery capacity constraints, which akes the planning odel ore rigorous and scientific. The satisfactory solution of the odel is obtained by using the chaos and harony search algorith, which akes the charging deand of the electric vehicle users in the planning area be satisfied. Exaple analysis shows that the odel and algorith used in this paper has soe scientificalness, and can provide soe reference for engineering planning. REFERECES [1] SU Ling, ZHAG Jianhua, WAG Li, et al, Study on soe key probles and technique related to icrogrid, Power Syste Protection and Control, vol.38, no.19, pp.35-39, 010. [] KHODAAR M, BARATI M, SHAHIDEHPOUR M, Integration of high reliability distribution syste in icrogrid operation, Sart Grid, vol.3,no.4, pp.1997-006, 01. [3] LI in, ZHAG Boing, ZHAO Jinquan, et al. An online optial power flow approach based on extended linear prograing[j], Autoation of Electric Power Systes, vol.30,no.5, pp.18-3, 006. [4] CHE C C, WAG S C. Branch and bound scheduling for theral generating units[j]. IEEE Trans on Energy Conversion, vol.8,no., pp184-189,1993. [5] WAG Chengshuan, LIU Tao, XIE inghua, Sbustation locating and sizing based on hybrid genetic algorith[j]. Autoation of Electric power Syste, vol.5, no.6, pp40-44, 006. [6] R. icole, Title of paper with only first word capitalized, J. ae Stand. Abbrev., in press. [7] TURTO H, MOURA F, Vehicle-to-grid systes for sustainable developent:an integrated energy analysis, Technological Forecasting and Social Change, vol.75, no.8, pp.1091-1108, 008. [8] ADERSE P, MATHEWS J, MORTE R, Integrating private transport into renewable energy policy : the strategy of creating intelligent recharging grids for electric vehicles, Energy Policy, vol.37, no.7, pp.481-486, 009. [9] SABER A, VEAAGAMOORTH G, Plug-in vehicles and renewable energy sources for cost and eission reductions, IEEE Transactions on Industrial Electronics, vol.58, no.4, pp.19-138, 011. [10] U Dayang, HUAG Haili, LEI Ming, et al, CO reduction benefit by coordinated dispatch of electric vehicle charging and wind power, Autoation of Electric Power Systes, vol.36, no.10, pp.14-18, 01. [11] KOU Lingfeng, LIU Zifa, ZHOU Huan. Modeling algorith of charging station planning for regional electric vehicle[j]. Modern Electric Power, vol.7, no.4, pp.44-48, 010. [1] TAG Xianggang, LIU Junyong, LIU ongbo. Electric Vehicle Charging Station Planning Based on Cotational Geoetry Method[J]. Autoation of Electric Power Syste, vol.36, no.8, pp4-30, 011. [13] LI uehong, WA Pin, WAG onghua, et al, Optial linear cooperation spectru sensing ethod based on chaos harony search algorith, Journal of Coputer Applications, vol.3, no.9, pp.41-417, 01. V. COCLUSIOS The planning of EV charging station is a coplex prograing proble involving the user's deand, econoic benefit, geographical topology and any other factors. In this paper, the iniu annual cost function odel is established, 353