Intelligent site selection and capacity determination method for electric vehicle charging station

Abstract

The application discloses a kind of electric vehicle charging station intelligent site selection and sizing method, solve the technical problem that existing electric vehicle site selection and sizing are not intelligent, the application belongs to electric vehicle charging station intelligent planning technical field, specific technical scheme is: first, the multi-objective multi-constrained charging station site selection and sizing model of minimum charging station construction cost, operation and maintenance cost, service user charging cost, queuing time cost and user driving process power consumption cost is established, then using BP neural network, with site selection land rent as evaluation index, the research area is selected, secondly, the preselected address is sized using ant colony algorithm, the best site selection of electric vehicle charging station and its optimal sizing are obtained, finally, the effectiveness of BP neural network for preselecting the research area is verified using site selection and sizing examples, the feasibility of optimal site selection and sizing is adopted using ant colony algorithm, the site selection and sizing are pre-judged, with the characteristics of accurate prediction, site selection speed is fast.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent planning of electric vehicle charging stations, and relates to an intelligent planning model of electric vehicle charging stations and an ant colony algorithm site selection and capacity determination method based on BP neural network pre-site selection. BACKGROUND

[0002] In the past 10 years, with the rapid development of science and technology in the field of power batteries, electric vehicles have attracted great attention, and have initially formed a scale market in Europe and the United States and other countries. At present, the number of electric vehicles in China is growing rapidly. Compared with traditional vehicles, electric vehicles are low-carbon, environmentally friendly, and economical and efficient, and are widely used. According to the data statistics in 2021, the sales volume of electric vehicles in China is 3.5 million, which has been the largest in the world for seven consecutive years. The number of electric vehicles in China is 7.84 million. The rapid growth of sales volume has brought huge charging demand, thus stimulating the comprehensive deployment and rapid growth of electric vehicle charging station infrastructure.

[0003] Under the above background, the optimal planning of electric vehicle charging stations has become an important problem worthy of study. Through scientific layout planning of charging stations, it is beneficial to optimize power resource allocation, improve power grid efficiency, reduce economic cost and power grid loss, and improve user convenience.

[0004] At present, the research on electric vehicle charging station planning problem is mainly divided into the establishment of charging station site selection and capacity determination model and the research on model solving method.

[0005] In terms of model, only single site selection or capacity determination is studied. On the other hand, charging station site selection and capacity determination involve electric vehicle charging facilities, service objects, and other constraints. The current models only consider the factors concerned, which are not comprehensive enough. The application discusses the service objects and influencing factors, establishes a multi-objective and multi-constraint planning charging station site selection and capacity determination model with the minimum charging cost of service users, user travel power consumption cost and queuing waiting time cost, and compares the overall planning pattern of Chinese electric vehicle charging stations more scientifically and comprehensively.

[0006] In terms of model solving algorithm, bionic intelligent algorithm and optimization theory algorithm are currently used, which has the disadvantages of large amount of calculation and slow convergence speed. In the application, the rent is used as the evaluation index, and the BP neural network is used to pre-select a number of station addresses, and then the ant colony algorithm is used to solve the optimal address and its capacity based on the pre-selected address. The calculation amount is small, and the convergence is fast. SUMMARY

[0007] To solve the technical problems in the prior art, the application provides an intelligent site selection and capacity determination method for an electric vehicle charging station, analyzes constraint conditions, establishes a charging station site selection and capacity determination model, solves the charging station optimization model by using a BP neural network and an ant colony algorithm, and obtains the optimal site selection and capacity determination of the charging station.

[0008] To achieve the above object, the technical scheme adopted by the application is as follows:

[0009] I. Optimize the planning target of the charging station, and calculate the investment cost.

[0010] The charging optimization model is established according to the minimization of the total charging construction cost, and the construction investment cost of the charging station is:

[0011]

[0012] In formula (1), T is the use time limit of the charging station, M is the number of electric vehicle charging stations, CI g,i is the construction cost of the i th charging station, P ch,i is the rated capacity of the charging pile, C r is the construction cost of one charging pile, c i is the number of charging piles of the i th charging station, s i is the floor area of the i th charging station, C f,i is the rent of the i th charging station within the use time limit.

[0013] The operation and maintenance cost of the charging station is:

[0014]

[0015] In formula (2), k is a proportional factor, the operation and maintenance is generally 1.5 times the construction cost, and k = 1.5%.

[0016] The charging cost of the service user is:

[0017]

[0018] In formula (3), P e is the battery capacity of the electric vehicle, P c is the charging price difference, N i is the number of vehicles charged by the i th charging station per day.

[0019] The objective function of the investment and operation cost in formula (4) is obtained from formula (1)-(3):

[0020] F1 = CI + CO - CC (4)

[0021] The time spent by the owner to charge and the cost of the power loss when the owner is on the way to the charging station.

[0022] The time spent by the electric vehicle to reach the charging station:

[0023]

[0024] In formula (5), N i is the number of the electric vehicles charged by the i-th charging station per day, D is the distance to the charging station, q is the power consumption per 100 kilometers, and c is the cost of charging per kilowatt-hour.

[0025] The time spent by the electric vehicle to wait for charging at the charging station:

[0026]

[0027] In formula (6), E q represents the expected value of the time spent by the owner of the electric vehicle to wait in line, and the parking fee for the waiting time within 30 minutes is 1 yuan.

[0028] According to formula (5) and (6), formula (7) can be obtained, and the time cost objective function of the owner to charge is:

[0029] F2=365β(C VTI +C W ) (7)

[0030] In formula (7), 365 is the number of days in most years, and β is the cost coefficient of the urban travel time, generally taken as 1.1%.

[0031] II. Optimization of the charging station planning model

[0032] From the classification of electric vehicles, charging methods, charging facilities, service objects, and influencing factors, a multi-objective charging station site selection and capacity determination model is established, in which the construction cost of the charging station, the operation and maintenance cost, the charging cost of the service user, the power consumption cost of the user driving, and the queuing time cost are minimized.

[0033] Formula (8) is obtained by combining formula (4) and formula (7), and the objective function of the charging station planning is:

[0034] minF=min(F1+F2) (8)

[0035] Power grid constraint, node voltage amplitude constraint:

[0036] V hmin ≤V h ≤V hmax (9)

[0037] In formula (9), V h is the voltage of the distribution network, and Vhmin Vmin is the minimum voltage of the distribution network hmax Vmax is the maximum voltage of the distribution network

[0038] Current constraint:

[0039] Imin is the minimum current of the distribution network g,min Imax is the maximum current of the distribution network g Imin is the minimum current of the distribution network g,max (10)

[0040] In equation (10), I is the current of the distribution network g Imin is the minimum current of the distribution network g,min Imin is the minimum current of the distribution network g,max Imax is the maximum current of the distribution network

[0041] Capacity constraint:

[0042] Pmin is the minimum capacity of the charging device ch,i Pmax is the maximum load of the substation or transformer i,max (11)

[0043] In equation (11), P is the rated capacity of the charging device ch,i Pmax is the maximum load of the substation or transformer i,max Pmax is the maximum load of the substation or transformer

[0044] In the most extreme conditions, at least to ensure that the car at the charging station is full of electricity to reach the adjacent charging station:

[0045] 2r i <D (12)

[0046] In equation (12), r is the charging service range i The service range is determined by the endurance and average charge of the electric vehicle:

[0047]

[0048] In equation (13), d is the shortest endurance in the electric vehicle m SOC is the average charge in the i-th range m,i μ is the discount on the endurance of the car in actual driving i

[0049] The human patience in line is 15-30 minutes, taking 20 minutes as an example, one person's patience in one hour is 1 / 3, so the expectation value is taken as:

[0050]

[0051] Charging station charging pile quantity constraint:

[0052] Cmin is the minimum number of charging piles i Cmax is the maximum number of charging piles i,max (15) ​

[0053] In formula (15), C i,max Determined by the maximum capacity limit of the charging station;

[0054] The number of electric vehicles is restricted:

[0055]

[0056] In formula (16), T l The number of cars on the road L, ζ is the market share of electric vehicles in this range, N il The path information generated by each vehicle.

[0057] The charging station planning model is solved, a BP neural network is constructed, and the BP network is used to pre-select the site of the electric vehicle charging station. The BP network is used for prediction, the neural network is trained using relevant data, and the experience is accumulated. The trained network judges new events. In essence, the BP algorithm is to take the square of the network error as the objective function, and use the gradient descent method to calculate the minimum value of the objective function. The transfer function is a nonlinear transformation function-Sigmoid function (i.e. S function).

[0058] Construction of BP neural network: Taking the built charging station as the research object, traffic convenience, land cost, candidate land area, accessibility, spatial characteristics, and population density are divided into five grades of excellent, good, general, poor, and bad, with corresponding values of 5, 4, 3, 2, and 1, and are analyzed as input layer data. Among them, on traffic convenience, more than 10 bus lines within 500 meters of the selected address are excellent, 8-10 are good, 5-8 are general, 3-5 are poor, and less than 3 are poor; on land cost, 5 and 6 are excellent, 4 is good, 3 is general, 2 is poor, and 1 is poor; on candidate land area, more than 50 parking spaces are excellent, 40-50 are good, 30-40 are general, 20-30 are poor, and less than 20 are poor; on accessibility, more than 5 roads are excellent, 4 roads are good, 3 roads are general, 2 roads are poor, and 1 road is poor; on spatial characteristics, it is determined by the degree of prosperity of the surrounding shopping malls, office buildings, residential buildings, hospitals, and schools; on population density, it is determined by the flow of people in the area, and the flow of people is especially poor. Therefore, the rent as the output layer index.

[0059] Establishment of charging station pre-site selection model based on BP network: The order of 150 groups of data of the input layer and output layer of the built charging station is disturbed, among which the first 130 groups are training data and the last 20 groups are test data. The activation function is a single-pole Sigmoid function, which is between (0, 1). The BP neural network is established, the parameters are set, and the training is performed.

[0060] The input layer is 6, and the output layer is 1, namely the rent. The number of intermediate layers is obtained from formula (17):

[0061]

[0062] In formula (17): n is the number of input layers; m is the number of intermediate layers; l is the number of output layers; a ∈ (1, 10), after multiple training, m = 10, learning rate η = 0.09, training error is 0.01, and the maximum number of training is 1000.

[0063] Through continuous iteration of the BP network, the relationship between the charging station site selection and the output value rent is well presented. The predicted 20 data are very close to the expected values. After 1000 times of training, the prediction value and the expected value fitting degree reach 0.9678, so the established model is feasible.

[0064] Charging station site selection and capacity determination based on ant colony algorithm: the ant colony algorithm is a probability algorithm used to find an optimized path, has the characteristics of distributed calculation, information positive feedback and heuristic search, and is essentially a heuristic global optimization algorithm in evolutionary algorithms.

[0065] After the BP network prediction model, R optimal sites are obtained, the optimal sites are converted into the positions of the ant colony algorithm, the number of ants is A, and they are placed on R paths to solve the site selection and capacity determination model.

[0066] According to the expected value of each address and the pheromone released on the line, the transition probability of the kth ant at the address is obtained, and then the path is selected. The higher the rent, the smaller the expected value of selecting the path. The number of ants A = 15, the maximum number of cycles N c_max = 200, the pheromone factor α = 2, the heuristic factor β = 4, the pheromone evaporation factor ρ = 0.5, the pheromone Q = 100, and the charging station usage time limit T = 50.

[0067] First, the related parameters are initialized, the R optimal addresses are input, the ants are placed in different positions, and their respective transition probabilities are calculated to determine their next path addresses, and the process stops when all ants have passed through; the roads passed by each ant are counted and compared, and the current iteration number of the best solution is recorded, and then the pheromone concentration on the roads passed by the ants is updated; finally, it is judged whether the maximum iteration number is reached, if yes, the optimal address and capacity are selected; otherwise, the number of cycles N c = N c + 1 continues to be calculated.

[0068] The application establishes a feasible planning model and a solving method for electric vehicle charging station site selection and capacity determination, that is, an intelligent site selection and capacity determination method for electric vehicle charging stations, which can provide method reference for subsequent related theoretical research, provide station capacity planning method for the government, provide station capacity technology for the operator, and also can relieve the charging anxiety of electric vehicle owners. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1 Error comparison chart of predicted value and expected value of charging station site selection and output.

[0070] Figure 2 Regression analysis chart of predicted value and expected value.

[0071] Figure 3 Error comparison chart of predicted value and expected value.

[0072] Figure 4 Optimization result chart of ant colony algorithm after iteration. DETAILED DESCRIPTION

[0073] In order to make the technical problems, technical solutions and beneficial effects to be solved by the application more clear and obvious, the application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application, and are not used to limit the application.

[0074] The intelligent site selection and capacity determination method for electric vehicle charging stations has the following specific steps:

[0075] I. Optimize the planning target of the charging station, and calculate the investment cost:

[0076] According to the minimization of the total cost of charging construction, a charging optimization model is established, and the construction investment cost of the charging station is:

[0077]

[0078] In formula (1), T is the use time limit of the charging station, M is the number of electric vehicle charging stations, CI g,i is the construction cost of the i th charging station, P ch,i is the rated capacity of the charging pile, C r is the construction cost of one charging pile, c i is the number of charging piles of the i th charging station, s i is the floor area of the i th charging station, C f,i is the rent of the i th charging station within the use time limit;

[0079] Charging station operation and maintenance cost:

[0080]

[0081] In formula (2), k is a proportional factor, the operation and maintenance is generally 1.5 times of the construction cost, k=1.5%.

[0082] The charging cost of the service user:

[0083]

[0084] In formula (3), P e is the battery capacity of the electric vehicle, P c is the charging price difference, N i is the number of vehicles charged per day at the i-th charging station.

[0085] The objective function of the investment operation cost of formula (4) is obtained from formula (1)-(3):

[0086] F1=CI+CO-CC (4)

[0087] The time consumed by the vehicle owner for charging and the electricity consumption cost of the vehicle owner when going to the charging station are taken into account.

[0088] The time required by the charging vehicle to arrive at the charging station:

[0089]

[0090] In formula (5), N i is the number of vehicles charged per day at the i-th charging station, D is the distance to the charging station, q is the electricity consumption per 100 kilometers, and c is the electricity fee paid by the vehicle for charging one kilowatt-hour.

[0091] The waiting time of the charging vehicle at the charging station for charging completion:

[0092]

[0093] In formula (6), E q represents the expected value of the queuing time of the electric vehicle owner, and the parking fee for the queuing time within 30 minutes is 1 yuan.

[0094] According to formula (5) and (6), formula (7) is obtained, which is the time cost objective function of the vehicle owner for charging:

[0095] F2=365β(C VTI +C W ) (7)

[0096] In formula (7), 365 is the number of days in most years, and β is the urban travel time cost coefficient, generally taken as 1.1%.

[0097] II. Optimization of the charging station planning model

[0098] From the classification of electric vehicles, charging methods, charging facilities, service objects, and influencing factors, a multi-objective charging station site selection and capacity determination model is established, which minimizes the charging station construction cost, operation and maintenance cost, service user charging cost, user driving power consumption cost, and queuing waiting time cost.

[0099] From formula (4) and formula (7), formula (8) is obtained, and the objective function of the charging station planning is:

[0100] minF = min (F1 + F2) (8)

[0101] Power grid constraint, node voltage amplitude constraint:

[0102] V hmin ≤V h ≤V hmax (9)

[0103] In formula (9), V h is the voltage of the distribution network, V hmin is the minimum voltage of the distribution network, and V hmax is the maximum voltage of the distribution network;

[0104] Current constraint:

[0105] I g,min ≤I g ≤I g,max (10)

[0106] In formula (10), I g is the current of the distribution network; I g,min is the minimum current of the distribution network, and I g,max is the maximum current of the distribution network;

[0107] Capacity constraint:

[0108] P ch,i ≤S i,max (11)

[0109] In formula (11), P ch,i is the rated capacity of the charging device, and S i,max is the maximum load of the transformer substation or transformer;

[0110] Under the most extreme conditions, at least the electric vehicle charged at the charging station has enough electricity to reach the adjacent charging station:

[0111] 2r i <D (12)

[0112] In formula (12), r i is the charging service range, which is determined by the endurance and average charge of the electric vehicle:

[0113]

[0114] In formula (13), d m is the shortest endurance in the electric vehicle, SOC m,i is the average state of charge in the ith range, μ i is the discount on the endurance when the vehicle is actually running;

[0115] The human patience in line is 15-30 minutes, and 20 minutes is taken as the calculation, and the patience of one person in one hour is 1 / 3, so the expectation is taken as:

[0116]

[0117] The charging pile number constraint of the charging station is:

[0118] C i ≤C i,max (15)

[0119] In formula (15), C i,max is determined by the maximum capacity limit of the charging station;

[0120] The number constraint of the electric vehicle is:

[0121]

[0122] In formula (16), T l is the number of vehicles on the road L, ζ is the market share of electric vehicles in the range, N il is the path information generated by each vehicle.

[0123] The charging station optimization planning model is solved, a BP neural network is constructed, the electric vehicle charging station pre-site selection based on the BP network is adopted, the BP network prediction is adopted, the neural network is trained by using relevant data, and then new events are judged.

[0124] The BP network has an input layer, a hidden layer and an output layer, and the BP algorithm is to take the network error square as the objective function and adopt the gradient descent method to calculate the minimum value of the objective function.

[0125] The BP neural network is a multi-level feedforward neural network, and the learning process includes two processes of forward propagation of signals and backward propagation of errors. The first process is from the input layer to the output layer through the hidden layer. If the actual output does not conform to the expectation, the second process is entered, which is from the output layer to the input layer through the hidden layer. The adjustment of the weight and offset of each layer in the two processes is repeated.

[0126] The transfer function of BP network is a nonlinear transformation function--Sigmoid function (S function), and the S function has two types: unipolar S function and bipolar S function. The unipolar S function is defined as follows:

[0127]

[0128] The BP network is used for prediction, and the neural network is trained by using relevant data to "accumulate" experience, and then the new event is judged.

[0129] Construction of BP neural network:

[0130] Taking the built charging stations in A city as the research object, traffic convenience, land cost, candidate land area, accessibility, spatial characteristics, and population density are analyzed as the input layer, and divided into five grades of excellent, better, general, worse, and poor, corresponding to the values of 5, 4, 3, 2, and 1. Among them, on the traffic convenience, more than 10 bus lines within 500 meters of the selected address are excellent, 8-10 are better, 5-8 are general, 3-5 are worse, and less than 3 are poor; on the land cost, the benchmark land price of A city is shown in Table 1, 5 and 6 are excellent, 4 is better, 3 is general, 2 is worse, and 1 is poor; on the candidate land area, more than 50 parking spaces are excellent, 40-50 are better, 30-40 are general, 20-30 are worse, and less than 20 are poor; on the accessibility, more than 5 roads are excellent, 4 roads are better, 3 roads are general, 2 roads are worse, and 1 road is poor; on the spatial characteristics, it is determined by the degree of prosperity of the surrounding shopping malls, office buildings, residential buildings, hospitals, schools, etc.; on the population density, it is determined by the flow of people in the area, and the flow of people is particularly poor. The rent price of different land can reflect the development of the local area, therefore, the rent is used as the output layer index.

[0131] Table 1 A city benchmark land price table (commercial, residential, industrial)

[0132]

[0133] Based on the BP network, the charging station pre-site selection model is established: 150 groups of data of the input layer and the output layer of the built charging station are shuffled, among which the first 130 groups are training data and the last 20 groups are test data. The activation function is Sigmoid function, which is between (0, 1), the BP neural network is established, the parameters are set, and the training is performed.

[0134] The input layer is 6, and the output layer is 1, i.e. rent. The number of intermediate layers is obtained from formula (18):

[0135]

[0136] In formula (19), n is the number of input layers; m is the number of intermediate layers; l is the number of output layers; a is in (1, 10), after multiple training, m = 10, learning rate η = 0.09, training error is 0.01, and the maximum number of training is 1000.

[0137] As shown in Figure 1 , through the continuous iteration of the BP network, the relationship between the charging station site selection and the output value rent is well presented, and 16 of the 20 predicted data are very close to the expected value; as shown in Figure 2 , after 1000 times of training, the fitting degree of the predicted value and the expected value of the 130 groups of data participating in the training and the 20 groups of data participating in the test reaches 0.9678, so the established model is feasible.

[0138] Charging station site selection and capacity determination based on ant colony algorithm: the ant colony algorithm is a probabilistic algorithm for finding an optimal path, has the characteristics of distributed computing, information positive feedback and heuristic search, and is essentially a heuristic global optimization algorithm in evolutionary algorithms.

[0139] After the BP network prediction model, R optimal sites are obtained, the optimal sites are converted into the positions of the ant colony algorithm, the number of ants is A, and the R paths are placed on the R paths for solving.

[0140] According to the expected value of each address and the pheromone released on the line, the transfer probability of the kth ant on the address is obtained, and then the path is selected. The higher the rent, the smaller the expected value of selecting the path. The number of ants A = 15, the maximum number of cycles N c_max = 200, the pheromone factor α = 2, the heuristic factor β = 4, the pheromone volatilization factor ρ = 0.5, the pheromone Q = 100, and the charging station usage time limit T = 50.

[0141] First, the related parameters are initialized, the selected addresses are input, the ants are placed in different positions, and their respective transfer probabilities are calculated to determine their next path addresses, and the process is stopped when all ants have passed through; the roads passed by each ant are counted and compared, and the current iteration number of the best solution is recorded, and then the pheromone concentration on the road is updated; finally, it is judged whether the maximum iteration number is reached, if yes, the optimal capacity is selected; otherwise, the cycle number N c = N c + 1 continues to be calculated.

[0142] Example simulation and analysis:

[0143] Taking B city as an example, the address preselected by the BP neural network prediction model is close to the predicted rent and the survey rent, and these addresses are used as the initial addresses of the ant colony algorithm, and the ant colony algorithm is used to optimally solve the address and capacity of the charging station.

[0144] Select the eight square, the people's government, wanda square, city God's temple, peace hospital, city people's hospital, west bus station, east bus station as the prediction address, from the traffic, land cost, land area, accessibility, spatial characteristics, population density analysis, rent for the final site selection, divided into 5, 4, 3, 2, 1 five levels.

[0145] Through the BP neural network pre site selection model to predict the expected value of the 8 groups of data, as shown in Figure 3 The error comparison of the predicted value and the expected value is obtained, and the first, fourth, seventh and eighth groups of data are very close to the expected value, so the eight square, city God's temple, west bus station and east bus station are selected as the optimal address.

[0146] The four candidate addresses obtained by prediction are applied to the ant colony algorithm, and under the constraint conditions of formula (9)-(16), the optimal result of charging station site selection and capacity is obtained according to the objective function formula (8).

[0147] From Figure 4 It can be seen that after the ant colony algorithm is iterated for 200 times, the best cost of building charging station can reach about 300,000 yuan. From table 2, under the premise of minimum cost, the number of charging piles built in city God's temple is the largest.

[0148] Table 2 optimization calculation result

[0149]

[0150] In table 3, it can be seen that the average annual net cost is only 300,000 yuan, which can build 42 charging piles in city God's temple for electric vehicle owners to use.

[0151] Table 3 construction cost of charging station based on ant colony algorithm

[0152]

[0153] The relevant data of B city electric vehicle charging station planning model are shown in table 4.

[0154] Table 4 relevant data of electric vehicle charging station planning model

[0155]

[0156] From the classification of electric vehicles, charging mode, charging facilities are researched, from the service object, influence factor are discussed, the multi-objective charging station intelligent site selection and fixed capacity model of charging station construction cost, operation and maintenance cost, service user charging cost, user driving power consumption cost and queuing waiting time cost minimum is established, the planning layout of electric vehicle charging station is compared scientific and comprehensive. First, the research area is selected by using BP neural network, and then the ant colony algorithm is used for fixed capacity of the selected address. The simulation example proves the effectiveness of the method.

[0157] The above merely describes preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement and improvement within the spirit and principle of the present application shall be included in the scope of the present application.

Claims

1. An intelligent site selection and sizing method for electric vehicle charging stations, characterized in that, The specific steps are as follows: According to the minimization of the total cost of charging construction, a charging optimization model is established, and the construction investment cost of the charging station is: In formula (1), T is the time limit for use of the charging station, M is the number of electric vehicle charging stations, CI g,i is the construction cost of the i-th charging station, P ch,i is the rated capacity of the charging pile, C r is the construction cost of a charging pile, c i is the number of charging piles of the i-th charging station, s i is the floor area of the i-th charging station, C f,i is the rent of the i-th charging station within the time limit for use; The operation and maintenance cost of the charging station is: In formula (2), k is a proportional factor; The charging cost of the service user is: In formula (3), P e is the battery capacity of the electric vehicle, P c is the price difference of charging, N i is the number of vehicles charged at the i-th charging station per day; The objective function of the investment and operation cost of formula (4) is obtained from formula (1)-(3): F1=CI+CO-CC (4) The time required for the charging vehicle to arrive at the charging station is: In formula (5), N i is the number of times of charging the car at the i-th charging station per day, D is the distance to the charging station, q is the power consumption per 100 km, and c is the electricity fee paid for charging one unit of electricity at the charging station. The waiting time for the charging vehicle to complete charging at the charging station is: In formula (6), E q represents the expected value of the time that the electric vehicle owner waits in line. According to formula (5) and (6), formula (7) is obtained, and the time cost objective function of the vehicle owner charging is: F2 = 365 β (C VTI + C W ) (7) In formula (7), β is the urban travel time cost coefficient; Formula (8) is obtained by integrating formula (4) and formula (7), and the objective function of the charging station planning is: minF=min(F1+F2) (8) Node voltage amplitude constraint: V hmin ≤V h ≤V hmax (9) In formula (9), V h is the distribution network voltage, V hmin is the minimum distribution network voltage, V hmax is the maximum distribution network voltage; Current constraint: I g,min ≤I g ≤I g,max (10) In formula (10), I g is the distribution network current; I g,min is the distribution network minimum current, I g,max is the distribution network maximum current; Capacity constraint: P ch,i ≤S i,max (11) In formula (11), P ch,i is the rated capacity of the charging device, S i,max is the maximum load of the substation or transformer; At least ensure that the car charged at the charging station has enough electricity to reach the adjacent charging station: 2r i <D (12) In formula (12), r i The charging service range is determined by the cruising range and average charge level of the electric vehicle. In formula (13), d m is the shortest cruising range in the electric vehicle, SOC m,i is the average state of charge in the ith range, μ i is the discount on the cruising range when the vehicle is actually driven; Expected value E q Take: Charging pile number constraint of the charging station: C i ≤C i,max (15) In formula (15), C i,max determined with the maximum capacity limit of the charging station; Electric vehicle number constraint: In formula (16), T l is the number of cars on the road L, ζ is the market share of electric cars in this range, N il is the path information generated per car; The objective function formula (8) and the constraint formula (9)-(16) constitute the intelligent site selection and capacity optimization model of the electric vehicle charging station. 2.The method of claim 1, wherein, Build a BP neural network, shuffle the data of the input layer and output layer of the built charging station, use Sigmoid function as the activation function, which is between (0, 1), establish the BP neural network, set the parameters, and train; The input layer is 6, the output layer is 1, that is, the rent, and the number of intermediate layers is obtained from formula (17): In formula (17), n is the number of input layers, m is the number of intermediate layers, l is the number of output layers, and a∈(1,10); Through the continuous iteration of the BP network, the relationship between the charging station site selection and the output value rent is shown. 3.The method of claim 2, wherein, After the BP network prediction model, R optimal sites are obtained, the optimal sites are converted into the positions of the ant colony algorithm, the number of ants is A, and they are placed on R paths to solve the established electric vehicle site selection and capacity optimization objective function formula (8); According to the expected value of each address and the pheromone released on the line, the transition probability of the kth ant at the address is obtained, then the path is selected, and the higher the rent, the smaller the expected value of selecting the path; First, initialize the related parameters, input the selected address, place the ants in different positions, and calculate their respective transition probabilities to determine their next path address, and stop when all ants have passed through; Compare the roads passed by each ant, record the best solution of the current iteration number, and then update the pheromone concentration on the road passed by the ant; Finally, it is judged whether the maximum iteration number is reached, if so, the optimal capacity is selected; Otherwise, the number of loops N is set to c = N c + 1 and the calculation continues.

Images