A method and system for planning an electric vehicle charging station considering charging demand

Abstract

A kind of electric vehicle charging station planning method and system considering charging demand, the method is first simulated to obtain the charging demand of electric vehicle at each time period of each demand point, then based on charging demand quantity, charging station planning model with the minimum annual loss of charging station and user loss as target is constructed, and coarse-grained solution charging station planning model is solved, to obtain the initial planning scheme of electric vehicle charging station including initial site, the number of fast charging pile and slow charging pile of each initial site, finally the initial planning scheme of electric vehicle charging station is used as the initial population of whale optimization algorithm, and whale optimization algorithm is used for search optimization, to obtain the final electric vehicle charging station planning scheme.The present application not only realizes the maximization of interests of both operators and users, but also effectively improves the solving performance and convergence characteristics.

Description

Technical Field

[0001] This invention belongs to the field of power system planning, and specifically relates to a planning method and system for electric vehicle charging stations that takes into account charging demand. Background Technology

[0002] In recent years, with the advancement of science and technology, electric vehicles have become widely popular due to their numerous advantages, such as energy saving, zero carbon emissions, high energy efficiency, and advanced technology, making them the preferred mode of transportation for many urban residents. By 2023, global electric vehicle sales are projected to reach 14 million units, accounting for approximately one-fifth of total global car sales. The rapidly developing electric vehicle industry has placed new demands on the construction of charging station infrastructure, and the ever-increasing demand for electric vehicle charging reflects the enormous market potential for charging stations and other related infrastructure.

[0003] As operators of the power system, their goal is to reduce construction costs while maximizing annualized returns; while electric vehicle users want to minimize charging costs and waiting times. Therefore, maximizing the overall benefits for both the power system and electric vehicle users while meeting regional electric vehicle charging needs presents a significant challenge.

[0004] However, existing electric vehicle planning methods have certain limitations. Some methods consider construction costs and user losses but neglect the profits of grid operators; others, while considering the impact of external environmental factors on charging station planning, calculate annualized costs on an annual basis, which cannot guarantee the accuracy of the calculations. Furthermore, these approaches merely decouple user selection, capacity allocation, and charging station planning, failing to consider the overall planning perspective, thus exhibiting certain shortcomings. Summary of the Invention

[0005] The purpose of this invention is to address the aforementioned problems in the existing technology by providing a method and system for planning electric vehicle charging stations that takes into account charging demand.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] In a first aspect, the present invention proposes a method for planning electric vehicle charging stations that takes into account charging demand, comprising:

[0008] S1. Simulate the charging demand of electric vehicles at various demand points during different time periods.

[0009] S2. Construct a charging station planning model based on charging demand, with the goal of minimizing the annualized loss of charging stations and user losses.

[0010] S3. Solve the charging station planning model in a coarse-grained manner to obtain the initial planning scheme for electric vehicle charging stations, including the initial site selection points and the number of fast charging piles and slow charging piles at each initial site selection point;

[0011] S4. The initial planning scheme of electric vehicle charging stations is used as the initial population of the whale optimization algorithm. The whale optimization algorithm is used to search for and optimize the electric vehicle charging station planning scheme, including the final site selection point and the number of fast charging piles and slow charging piles at each final site selection point.

[0012] In S2, the objective function of the charging station planning model includes:

[0013] minD(M i ,F Li ,S Li )=α·C G +β·C U

[0014] C G =C CS +C O +C B -C S +C T

[0015]

[0016] In the above formula, Let M be the objective function. i Let i be the decision variable for whether to establish a charging station at the i-th candidate point. C represents the number of fast charging stations and slow charging stations at the i-th candidate point Li, respectively. G C U Let C represent the annualized loss of the charging station and the user loss, respectively, with α and β being the weights of the annualized loss of the charging station and the user loss, respectively. CS The annualized total cost of building a charging station, C O For the operation and maintenance costs of charging stations, C B For the electricity purchase cost of charging stations, C S For the revenue from selling electricity at the charging station, C T The additional line loss cost for charging electric vehicle users, w0 and y0 are the depreciation rate and total service life, respectively, and C is the total service life. fix C land C buy These include the fixed asset cost of the charging station, the land acquisition cost, and the cost of purchasing charging piles. Let EC be the sum of the charging demands of charging points around Li during time period t on day d, EC be the average charging capacity of each electric vehicle, k be the unit operating cost, and a and b be the number of candidate charging points and charging demand points, respectively. Let be the electricity purchase price and the electricity selling price for time period t on day d, respectively, and η be the charging efficiency of the electric vehicle. Energy loss during the journey, To account for the queuing time loss, dist(L i U j () represents the j-th charging demand point N. j Distance to Li These represent the power consumption per unit distance and the cost per unit waiting time, respectively, f(L) i N j ) is N j The decision variable is whether the distance to Li satisfies the distance coverage constraint. N is the time period t on day d. j The charging needs, For distance at r max The total number of charging stations in the area, r max The maximum allowable distance between adjacent charging stations is T(d,t,λ), and the queuing time of users within time period t on day d is T(d,t,λ).

[0017] The constraints include power constraints of the charging station, distance constraints, and maximum waiting time constraints for charging users.

[0018] Specifically, S3 involves: first, using the farthest point set priority algorithm to determine the initial site selection point; then, based on the initial site selection point, using the tabu search algorithm to solve the charging station planning model to obtain the number of fast charging piles and slow charging piles at each initial site selection point.

[0019] The method of determining the initial site selection point using the farthest point set priority algorithm includes:

[0020] S31. Initialize the location point set P set Candidate point set C set , where P set Initially, it is an empty set;

[0021] S32, Calculate C set The distance between any two candidate points in C is used to determine the two candidate points with the greatest distance from each other. set Move into P set middle;

[0022] S33, Calculate C respectively set From each candidate point to P set The sum of distances to all candidate points within the range, and sorted in descending order of distance;

[0023] S34. Starting with the candidate point with the largest sum of distances, sequentially determine whether the candidate point satisfies the distance constraints between location points, the distance constraints between the candidate point and the demand point, and the coverage constraints of the demand point. Then, select the first candidate point that satisfies all constraints from C. setMove into P set middle;

[0024] S35. Determine whether all charging demand points satisfy the following constraints:

[0025]

[0026] In the above formula, Dist(N) j ,L i () represents the j-th charging demand point N. j The distance r to the i-th candidate point Li max This represents the maximum allowable distance between adjacent charging stations. These are the set of charging demand points and the set of candidate points, respectively.

[0027] If satisfied, then P at this time set Output the initial location point; if the condition is not met, return to S33 for the next iteration.

[0028] S1 uses Monte Carlo simulation to obtain the daily charging demand of electric vehicles at various time periods throughout the year at each demand point, including:

[0029] S11. Within the selected area, determine the demand points for electric vehicles by community, school, or street.

[0030] S12. The travel behavior of electric vehicle users is described using a travel chain. The probability that a user travels at any time and then transfers to another location is represented by the following matrix:

[0031]

[0032] In the above formula, R′ is the transition probability matrix, p ab Let be the transition probability of an electric vehicle from node a to node b;

[0033] S13. The electric vehicle generates the destination and plans the driving route based on the transition probability matrix of the corresponding time period;

[0034] S14. During the simulated electric vehicle driving process, the battery level of each electric vehicle is monitored in real time, and it is determined whether it is lower than the set threshold. If it is lower than the set threshold, its charging demand is recorded, thereby obtaining the charging demand of electric vehicles at each node at different time periods.

[0035] The p ab The data obtained by fitting using a multilayer perceptron includes: first, collecting historical data of each demand point within the selected area, including the date of worship, season category, ambient temperature, and travel data of node a and node b; then, inputting the historical data into the multilayer perceptron neural network to fit and obtain p. abIn this neural network, the hidden layer uses the ReLU function as the activation function, the output layer uses the Sigmoid activation function, and the loss function is the mean square error between the fitted probability value and the actual probability value.

[0036] Secondly, this invention proposes an electric vehicle charging station planning system that considers charging demand, including a charging demand simulation module, a planning model construction module, a coarse-grained solution module, and a fine-grained solution module.

[0037] The charging demand simulation module is used to simulate the charging demand of electric vehicles at various demand points in different time periods.

[0038] The planning model construction module is used to build a charging station planning model based on charging demand. The charging station planning model aims to minimize the annualized loss of the charging station and the loss to users.

[0039] The coarse-grained solution module is used to coarsely solve the charging station planning model to obtain the initial planning scheme for electric vehicle charging stations, including the initial site selection points and the number of fast charging piles and slow charging piles at each initial site selection point.

[0040] The fine-grained solution module is used to take the initial planning scheme of electric vehicle charging stations as the initial population of the whale optimization algorithm, and use the whale optimization algorithm to search and optimize to obtain the electric vehicle charging station planning scheme, including the final site selection point and the number of fast charging piles and slow charging piles at each final site selection point.

[0041] The objective function of the charging station planning model includes:

[0042] minD(M i ,F Li ,S Li )=α·C G +β·C U

[0043] C G =C CS +C O +C B -C S +C T

[0044]

[0045] In the above formula, Let M be the objective function. i Let i be the decision variable for whether to establish a charging station at the i-th candidate point. C represents the number of fast charging stations and slow charging stations at the i-th candidate point Li, respectively. G C ULet C represent the annualized loss of the charging station and the user loss, respectively, with α and β being the weights of the annualized loss of the charging station and the user loss, respectively. CS The annualized total cost of building a charging station, C O For the operation and maintenance costs of charging stations, C B For the electricity purchase cost of charging stations, C S For the revenue from selling electricity at the charging station, C T The additional line loss cost for charging electric vehicle users, w0 and y0 are the depreciation rate and total service life, respectively, and C is the total service life. fix C land C buy These include the fixed asset cost of the charging station, the land acquisition cost, and the cost of purchasing charging piles. Let EC be the sum of the charging demands of charging points around Li during time period t on day d, EC be the average charging capacity of each electric vehicle, k be the unit operating cost, and a and b be the number of candidate charging points and charging demand points, respectively. Let be the electricity purchase price and the electricity selling price for time period t on day d, respectively, and η be the charging efficiency of the electric vehicle. Energy loss during the journey, To account for the queuing time loss, dist(L i U j () represents the j-th charging demand point N. j Distance to Li These represent the power consumption per unit distance and the cost per unit waiting time, respectively, f(L) i N j ) is N j The decision variable is whether the distance to Li satisfies the distance coverage constraint. N is the time period t on day d. j The charging needs, For distance at r max The total number of charging stations in the area, r max The maximum allowable distance between adjacent charging stations is T(d,t,λ), and the queuing time of users within time period t on day d is T(d,t,λ).

[0046] The constraints include power constraints of the charging station, distance constraints, and maximum waiting time constraints for charging users.

[0047] The coarse-grained solution module includes an initial site selection point determination unit and a charging station initial planning scheme determination unit.

[0048] The initial site selection point determination unit is used to determine the initial site selection point using the farthest point set priority algorithm. The specific process includes:

[0049] A1. Initialize the location point set P set Candidate point set C set , where Pset Initially, it is an empty set;

[0050] A2. Calculate C set The distance between any two candidate points in C is used to determine the two candidate points with the greatest distance from each other. set Move into P set middle;

[0051] A3. Calculate C respectively. set From each candidate point to P set The sum of distances to all candidate points within the range, and sorted in descending order of distance;

[0052] A4. Starting with the candidate point with the largest sum of distances, sequentially determine whether the candidate point satisfies the distance constraints between location points, the distance constraints between the candidate point and the demand point, and the coverage constraints of the demand point. Then, select the first candidate point that satisfies all constraints from C. set Move into P set middle;

[0053] A5. Determine whether all charging demand points satisfy the following constraints:

[0054]

[0055] In the above formula, Dist(N) j ,L i () represents the j-th charging demand point N. j The distance r to the i-th candidate point Li max This represents the maximum allowable distance between adjacent charging stations. These are the set of charging demand points and the set of candidate points, respectively.

[0056] If satisfied, then P at this time set Output the initial address point; if the condition is not met, return to A3 for the next loop.

[0057] The initial planning scheme determination unit for the charging station is used to solve the charging station planning model based on the determined initial site selection point using a tabu search algorithm, and to obtain the number of fast charging piles and slow charging piles at each initial site selection point.

[0058] The charging demand simulation module uses Monte Carlo simulation to obtain the daily charging demand of electric vehicles at various time periods throughout the year at each demand point. The specific process includes:

[0059] a1. Within the selected area, determine the demand points for electric vehicles by community, school, or street.

[0060] a2. The travel chain is used to describe the travel behavior of electric vehicle users. The probability that a user travels at any time and transfers to another location is represented by the following matrix:

[0061]

[0062] In the above formula, R′ is the transition probability matrix, p ab Let be the transition probability of an electric vehicle from node a to node b;

[0063] a3. Electric vehicles generate destinations and plan driving routes based on the transition probability matrix of the corresponding time period;

[0064] a4. During the simulated driving of electric vehicles, the battery level of each electric vehicle is monitored in real time, and it is determined whether it is lower than the set threshold. If it is lower than the set threshold, its charging demand is recorded, thereby obtaining the charging demand of electric vehicles at each node at different time periods.

[0065] The p ab The data obtained by fitting using a multilayer perceptron includes: first, collecting historical data of each demand point within the selected area, including the date of worship, season category, ambient temperature, and travel data of node a and node b; then, inputting the historical data into the multilayer perceptron neural network to fit and obtain p. ab In this neural network, the hidden layer uses the ReLU function as the activation function, the output layer uses the Sigmoid activation function, and the loss function is the mean square error between the fitted probability value and the actual probability value.

[0066] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0067] 1. This invention proposes an electric vehicle charging station planning method that considers charging demand. First, it simulates the charging demand of electric vehicles at various demand points during different time periods. Then, based on the charging demand, it constructs a charging station planning model with the objective of minimizing the annualized loss of the charging station and the user's loss. The model is then solved in a coarse-grained manner to obtain an initial planning scheme for electric vehicle charging stations, including initial site selection points and the number of fast and slow charging piles at each initial site. Finally, this initial planning scheme is used as the initial population for a whale optimization algorithm to search for and optimize the final electric vehicle charging station plan. On one hand, this method considers the annual revenue of the power system operator and the loss of electric vehicle users, simultaneously maximizing the interests of both operators and users. On the other hand, this method employs a two-stage solution approach, sequentially using coarse-grained and fine-grained (whale optimization) search optimization, improving solution performance and exhibiting good convergence characteristics.

[0068] 2. The coarse-grained solution of the electric vehicle charging station planning method considering charging demand proposed in this invention includes first using the farthest point set priority algorithm to determine the initial site selection points, and then using the tabu search algorithm to solve the charging station planning model based on the determined initial site selection points, obtaining the number of fast charging piles and slow charging piles at each initial site selection point. The farthest point set priority algorithm can effectively reduce the complexity of subsequent model solution; the tabu search algorithm avoids getting trapped in local optima too early.

[0069] 3. The electric vehicle charging station planning method proposed in this invention fully considers the changes in electric vehicle users' travel behavior over time, taking into account the impact of seasonal, weekday, holiday, and 24-hour variations on user travel behavior. It simulates users' charging demand over time based on travel behavior and uses a multilayer perceptron to fit the travel probability during the simulation process. This method can adapt to the variability of environmental factors, find the relationship between the input nodes, seasonal changes, and the output travel probability, and efficiently fit the complex nonlinear relationship between the two. Attached Figure Description

[0070] Figure 1 This is a flowchart of the method described in Example 1.

[0071] Figure 2 This is a schematic diagram of the structure of the multilayer perceptron (MLP) neural network in Example 1.

[0072] Figure 3 This is a flowchart of the first stage model solution in Example 1.

[0073] Figure 4 This is a flowchart of the whale optimization algorithm in Example 1.

[0074] Figure 5 This is a schematic diagram of the system described in Example 2. Detailed Implementation

[0075] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings.

[0076] Example 1:

[0077] A method for planning electric vehicle charging stations that takes into account charging demand, such as Figure 1 As shown, the specific implementation steps are as follows:

[0078] 1. Within the selected urban area, taking residential communities, schools, or blocks as units, for areas with more than 400 motor vehicles and an electric vehicle ownership rate of not less than 8%, the geometric center of these areas will be designated as electric vehicle demand points. Historical data for each electric vehicle demand point will be collected, including dates of worship. Seasonal categories Ambient temperature T i The travel data of node a and node b form the input vector. Inputting the multilayer perceptron (MLP) neural network, we obtain p by fitting. ab The structure of a multilayer perceptron (MLP) neural network is as follows: Figure 2 As shown, the hidden layer includes hidden layer 1 (containing 64 neurons) and hidden layer 2 (containing 32 neurons), both of which use the ReLU function as the activation function; the output layer uses the Sigmoid activation function, which maps the output to a probability between 0 and 1. The loss function is the mean squared error between the fitted probability value and the actual probability value. The actual probability value is the proportion of the number of travel data sets from node a to node b on the same date to the total number of travel data sets.

[0079] 2. Monte Carlo simulation was used to obtain the daily charging demand of electric vehicles at various time periods throughout the year at different demand points, including:

[0080] 2.1 Within the selected area, determine the demand points for electric vehicles by community, school, or street.

[0081] 2.2 The travel behavior of electric vehicle users is described using a travel chain. The probability that a user travels at any given time and then moves to another location is represented by the following matrix:

[0082]

[0083] In the above formula, R′ is the transition probability matrix, p ab Let be the transition probability of an electric vehicle from node a to node b;

[0084] 2.3 The electric vehicle generates the destination and plans the driving route based on the transition probability matrix of the corresponding time period;

[0085] 2.4 During the simulated electric vehicle driving process, the battery level of each electric vehicle is monitored in real time, and it is determined whether it is lower than the set threshold. If it is lower than the set threshold, its charging demand is recorded, thereby obtaining the charging demand of electric vehicles at each node at different time periods.

[0086] 3. Construct a charging station planning model.

[0087] The charging needs of electric vehicle users in different regions can be represented by a series of demand points: Each point is uniquely identified by its latitude and longitude. The charging demand points N within time period t on day d are... j The number of electric vehicles requiring charging, i.e., the charging demand, is expressed as Users at each charging point will only travel to a distance of r. maxWhen charging is done at multiple charging stations within the area, the average demand is directed to each station. Therefore, the sum of the charging demands of all charging stations surrounding the i-th candidate point Li (the candidate location of the charging station is called the candidate point) within time period t on day d is... Calculate using the following formula:

[0088]

[0089] In the above formula, For distance at r max The total number of charging stations within the area, f(L) i N j ) is N j The decision variable is whether the distance to Li satisfies the distance coverage constraint. It is set to 1 if the distance coverage constraint is satisfied, and 0 otherwise. b is the number of charging demand points.

[0090] Considering the sum of fixed assets, land, and charging pile costs, the annualized total cost C for the power system to complete the construction of all charging stations is... CS Calculate using the following formula:

[0091]

[0092] In the above formula, w0 and y0 are the depreciation rate and the total operating years, respectively, and C... fix C land C buy These are the fixed asset costs of the charging station, the land acquisition costs, and the costs of purchasing charging piles, respectively. i Let be the decision variable for whether to establish a charging station at the i-th candidate point; 1 is taken if a charging station is established, and 0 otherwise. The unit fixed asset cost and unit land acquisition cost for each charging station. The unit purchase cost for slow charging piles and fast charging piles are respectively. denoted by , where represents the number of fast charging stations and slow charging stations for the i-th candidate point Li, and a represents the number of candidate points.

[0093] The operation and maintenance cost of charging stations C O Electricity purchase cost C B Electricity sales revenue C S The additional line loss cost C for electric vehicle users when charging T Calculate using the following formula:

[0094]

[0095] In the above formula, EC is the average charging capacity of each electric vehicle, and k is the unit operating cost, i.e., the operating cost per kilowatt-hour. Let be the electricity purchase price and the electricity selling price for time period t on day d, respectively, and η be the charging efficiency of the electric vehicle.

[0096] Considering the sum of construction costs and electricity purchase and sales revenue within one year, the annualized loss C of the charging station is... G The calculation is as follows:

[0097] C G =C CS +C O +C B -C S +C T .

[0098] Electric vehicle users incur energy consumption losses and queuing time losses while traveling to charging stations, resulting in a user loss of C. U for:

[0099]

[0100] In the above formula, Energy loss during the journey, This is due to the loss of queuing time.

[0101] The travel loss incurred by a user while traveling to a charging station is calculated using the following formula:

[0102]

[0103] In the above formula, dist(L i U j () represents the j-th charging demand point N. j Distance to Li Cost per unit distance.

[0104] The loss incurred by users due to queuing is calculated using the following formula:

[0105]

[0106] In the above formula, Let T(d,t,λ) be the cost per unit of waiting time, and let T(d,t,λ) be the user's queuing time within time period t on day d. Let ρ be the total number of charging piles, ρ be the service intensity, λ be the number of vehicles entering the charging station, and P0 be the probability of an idle charging station. The average number of vehicles that are charged at each charging station in one hour.

[0107] The optimization objective is the weighted sum of the annualized losses of electric charging stations and the losses of electric vehicle users, aiming to maximize annualized revenue while minimizing user losses. The specific optimization objective is expressed as follows:

[0108]

[0109] In the above formula, Let M be the objective function, α and β be the weights of the annualized loss of the charging station and the loss of the user, respectively, α,β∈[0,1], and the decision variables include M. i ,

[0110] The constraints include:

[0111] The power constraints of charging stations are as follows:

[0112]

[0113] In the above formula, P F P S These represent the maximum power of fast charging stations and slow charging stations, respectively, in P. max This is the maximum power allowed by the charging station;

[0114] The distance constraints between charging stations are as follows:

[0115]

[0116] In the above formula, d min This is the minimum distance between charging stations. For the set of candidate points;

[0117] The distance constraints between charging stations and user demand points are as follows:

[0118]

[0119] In the above formula, r max This represents the maximum allowable distance between adjacent charging stations.

[0120] The maximum waiting time constraint for charging users is as follows:

[0121]

[0122] In the above formula, T(d,t) represents the user's waiting time within time period t on day d, and t max Z represents the maximum user wait time within a given time period. * Represents an integer.

[0123] 4. Use the farthest point set first algorithm to determine the initial location point, such as... Figure 3 As shown, it includes:

[0124] 4.1 Initialize the location point set P set Candidate point set C set , where P set Initially, it is an empty set;

[0125] 4.2 Calculate C setThe distance between any two candidate points in C is used to determine the two candidate points with the greatest distance from each other. set Move into P set middle;

[0126] 4.3 Calculate C respectively set From each candidate point to P set The sum of distances to all candidate points within the range, and sorted in descending order of distance;

[0127] 4.4 Starting with the candidate point with the largest sum of distances, sequentially determine whether the candidate point satisfies the distance constraints between location points, the distance constraints between the candidate point and the demand point, and the coverage constraints of the demand point. Then, select the first candidate point that satisfies all constraints from C. set Move into P set middle;

[0128] 4.5 Determine whether all charging demand points satisfy the following constraints:

[0129]

[0130] In the above formula, Dist(N) j ,L i () represents the j-th charging demand point N. j The distance r to the i-th candidate point Li max This represents the maximum allowable distance between adjacent charging stations. These are the set of charging demand points and the set of candidate points, respectively.

[0131] If satisfied, then P at this time set Output the initial location point; if the condition is not met, return to step 4.3 for the next iteration.

[0132] 5. Based on P set The tabu search algorithm is used to solve the charging station planning model, obtaining the number of fast charging piles and slow charging piles at each initial site selection point. The specific process is as follows: Figure 3 As shown, it includes:

[0133] 5.1 Initialize hyperparameters; the maximum number of iterations for the tabu search is T. tabu The tabu search operator is to increase or decrease the number of fast and slow charging stations by ΔL. j and ΔS j The tabu list stores operators that have already been performed. The set P containing the initial optimal solutions is initialized. best The optimal solution X best The evaluation function is

[0134] 5.2. A search operator is used to perform a neighborhood search on the number of fast and slow charging piles at all charging stations. Based on the tabu search rule, the solution for the next iteration is updated using the solution obtained from the neighborhood search.

[0135]

[0136] Among them, the taboo search rules include: using Perform a neighborhood search and calculate the evaluation function value of the neighborhood solution. If the following conditions are met... Breaking the taboo rule will maximize the evaluation function. Assigned to X best If this condition is not met, then the solution with the largest corresponding evaluation function among the solutions obtained by the untabulated operators will be selected. As the solution for the next iteration And the operator (ΔL) j ,ΔS j Add it to the taboo list. If The evaluation function value under the condition is better than P. best The evaluation function value of the worst solution in the set is used to update the optimal solution set P. best : The solution X n+1 Add to set P best And replace the solution in the set that makes the evaluation function the worst;

[0137] 5.3. Iterate until the iteration termination condition is met, and output the set P at this point. best .

[0138] 6. Using the initial planning scheme of electric vehicle charging stations as the initial population for the whale optimization algorithm, the algorithm is used to search and optimize, resulting in a planning scheme for electric vehicle charging stations, including the final site selection points and the number of fast charging piles and slow charging piles at each final site selection point, such as... Figure 4 As shown, it includes:

[0139] 6.1 Update the optimal solution in the population using the following formula:

[0140] X * =argminFITNESS(X n );

[0141] 6.2. For each individual in the population, initialize the hyperparameters: the random probability p is a random value between [0,1], and the coefficients A and C are calculated as follows:

[0142] A = a·r1-a

[0143] C = 2·r²

[0144] In the above formula, r1 and r2 are both random values ​​between [0,1], and a is a parameter that decreases linearly from 2 to 0 during the iteration process;

[0145] 6.3. Based on the update rules of the whale optimization algorithm, determine the new solution. If p≤0.5, enter the prey search mode. The solution update method in the prey search mode is as follows:

[0146] X(t+1)=X * (t)-A·D

[0147] A = 2a·rand(1) - a

[0148] D = |C*X*(t) - X(t)|

[0149] C = 2rand(1)

[0150] If p > 0.5 and |A| ≤ 1, then enter the encirclement mode. The solution update method for the encirclement mode is as follows:

[0151] X(t+1)=X rand (t)-A·D

[0152] D = |C·Xrand(t) - X(t)|

[0153] If p > 0.5 and |A| > 1, then enter the shrinking encirclement mode. The solution update method for the shrinking encirclement mode is as follows:

[0154]

[0155] 6.4 Determine if the new solution satisfies the constraints of the planning model. If not, proceed to step 6.2 to update the hyperparameters and search for a new solution; if the constraints are satisfied, proceed according to the objective function. Calculate the fitness of the new solution.

[0156] 6.5 Calculate the new solution for the next individual, i.e., jump to step 6.2; if all individuals in the population have completed the update, this round of iteration ends, and jump to step 6.1. If the maximum number of iterations is reached, the algorithm ends and returns the solution X with the best fitness. * , which serves as the final solution and corresponding fitness obtained from the solution.

[0157] Example 2:

[0158] An electric vehicle charging station planning system that takes into account charging demand, such as Figure 4 As shown, it includes a charging demand simulation module, a planning model construction module, a coarse-grained solution module, and a fine-grained solution module.

[0159] The charging demand simulation module is used to obtain the daily charging demand of electric vehicles at various time periods throughout the year using Monte Carlo simulation. The specific process includes:

[0160] a1. Within the selected area, determine the demand points for electric vehicles by community, school, or street.

[0161] a2. The travel chain is used to describe the travel behavior of electric vehicle users. The probability that a user travels at any time and transfers to another location is represented by the following matrix:

[0162]

[0163] In the above formula, R′ is the transition probability matrix, p ab The transition probability of an electric vehicle from node a to node b is obtained by fitting a multilayer perceptron. The fitting process includes: first, collecting historical data of each demand point in the selected area, including the date of worship, season category, ambient temperature, and travel data of nodes a and b; then, inputting the historical data into the multilayer perceptron neural network to fit p. ab In this neural network, the hidden layer uses the ReLU function as the activation function, the output layer uses the Sigmoid activation function, and the loss function is the mean square error between the fitted probability value and the actual probability value.

[0164] a3. Electric vehicles generate destinations and plan driving routes based on the transition probability matrix of the corresponding time period;

[0165] a4. During the simulated driving of electric vehicles, the battery level of each electric vehicle is monitored in real time, and it is determined whether it is lower than the set threshold. If it is lower than the set threshold, its charging demand is recorded, thereby obtaining the charging demand of electric vehicles at each node at different time periods.

[0166] The planning model construction module is used to construct a charging station planning model based on charging demand. The objective function of the charging station planning model includes:

[0167]

[0168] C G =C CS +C O +C B -C S +C T

[0169]

[0170]

[0171] In the above formula, Let M be the objective function. i Let i be the decision variable for whether to establish a charging station at the i-th candidate point. C represents the number of fast charging stations and slow charging stations at the i-th candidate point Li, respectively. G C U Let C represent the annualized loss of the charging station and the user loss, respectively, with α and β being the weights of the annualized loss of the charging station and the user loss, respectively. CS The annualized total cost of building a charging station, C O For the operation and maintenance costs of charging stations, C B For the electricity purchase cost of charging stations, C S For the revenue from selling electricity at the charging station, C T The additional line loss cost for charging electric vehicle users, w0 and y0 are the depreciation rate and total service life, respectively, and C is the total service life. fix C land C buy These include the fixed asset cost of the charging station, the land acquisition cost, and the cost of purchasing charging piles. Let EC be the sum of the charging demands of charging points around Li during time period t on day d, EC be the average charging capacity of each electric vehicle, k be the unit operating cost, and a and b be the number of candidate charging points and charging demand points, respectively. Let be the electricity purchase price and the electricity selling price for time period t on day d, respectively, and η be the charging efficiency of the electric vehicle. Energy loss during the journey, To account for the queuing time loss, dist(L i U j () represents the j-th charging demand point N. j Distance to Li These represent the power consumption per unit distance and the cost per unit waiting time, respectively, f(L) i N j ) is N j The decision variable is whether the distance to Li satisfies the distance coverage constraint. N is the time period t on day d. j The charging needs, For distance at r max The total number of charging stations in the area, r max The maximum allowable distance between adjacent charging stations is T(d,t,λ), and the queuing time of users within time period t on day d is T(d,t,λ).

[0172] The constraints include:

[0173] The power constraints of charging stations are as follows:

[0174]

[0175] In the above formula, PF P S These represent the maximum power of fast charging stations and slow charging stations, respectively, in P. max This is the maximum power allowed by the charging station;

[0176] The distance constraints between charging stations are as follows:

[0177]

[0178] In the above formula, d min This is the minimum distance between charging stations. For the set of candidate points;

[0179] The distance constraints between charging stations and user demand points are as follows:

[0180]

[0181] In the above formula, r max This represents the maximum allowable distance between adjacent charging stations.

[0182] The maximum waiting time constraint for charging users is as follows:

[0183]

[0184] In the above formula, T(d,t) represents the user's waiting time within time period t on day d, and t max Z represents the maximum user wait time within a given time period. * Represents an integer.

[0185] The coarse-grained solution module is used to coarsely solve the charging station planning model to obtain an initial planning scheme for electric vehicle charging stations, including initial site selection points and the number of fast charging piles and slow charging piles at each initial site selection point. It includes an initial site selection point determination unit and a charging station initial planning scheme determination unit.

[0186] The initial site selection point determination unit is used to determine the initial site selection point using the farthest point set priority algorithm. The specific process includes:

[0187] A1. Initialize the location point set P set Candidate point set C set , where P set Initially, it is an empty set;

[0188] A2. Calculate C set The distance between any two candidate points in C is used to determine the two candidate points with the greatest distance from each other. set Move into P set middle;

[0189] A3. Calculate C respectively. set From each candidate point to Pset The sum of distances to all candidate points within the range, and sorted in descending order of distance;

[0190] A4. Starting with the candidate point with the largest sum of distances, sequentially determine whether the candidate point satisfies the distance constraints between location points, the distance constraints between the candidate point and the demand point, and the coverage constraints of the demand point. Then, select the first candidate point that satisfies all constraints from C. set Move into P set middle;

[0191] A5. Determine whether all charging demand points satisfy the following constraints:

[0192]

[0193] In the above formula, Dist(N) j ,L i () represents the j-th charging demand point N. j The distance r to the i-th candidate point Li max This represents the maximum allowable distance between adjacent charging stations. These are the set of charging demand points and the set of candidate points, respectively.

[0194] If satisfied, then P at this time set Output the initial address point; if the condition is not met, return to A3 for the next loop.

[0195] The initial planning scheme determination unit for the charging station is used to solve the charging station planning model based on the determined initial site selection point using a tabu search algorithm, and to obtain the number of fast charging piles and slow charging piles at each initial site selection point.

[0196] The fine-grained solution module is used to take the initial planning scheme of electric vehicle charging stations as the initial population of the whale optimization algorithm, and to use the whale optimization algorithm to search for optimization, so as to obtain an electric vehicle charging station planning scheme including the final site selection point, the number of fast charging piles and slow charging piles at each final site selection point.

Claims

1. A method for planning electric vehicle charging stations considering charging demand, characterized in that, The method includes: S1. Simulate the charging demand of electric vehicles at various demand points during different time periods. S2. Construct a charging station planning model based on charging demand. The charging station planning model aims to minimize the annualized loss of charging stations and the loss of users. Its objective function includes: C G =C CS +C O +C B -C S +C T In the above formula, Let M be the objective function. i Let i be the decision variable for whether to establish a charging station at the i-th candidate point. The i-th candidate point L i The number of fast charging stations and slow charging stations, C G C U Let C represent the annualized loss of the charging station and the user loss, respectively, with α and β being the weights of the annualized loss of the charging station and the user loss, respectively. CS The annualized total cost of building a charging station, C O For the operation and maintenance costs of charging stations, C B For the electricity purchase cost of charging stations, C S For the revenue from selling electricity at the charging station, C T The additional line loss cost for charging electric vehicle users, w0 and y0 are the depreciation rate and total service life, respectively, and C is the total service life. fix C land C buy These include the fixed asset cost of the charging station, the land acquisition cost, and the cost of purchasing charging piles. L is the time period t on day d. i The sum of charging demands at surrounding charging points, EC is the average charging capacity per electric vehicle, k is the unit operating cost, and a and b are the number of candidate charging points and charging demand points, respectively. Let be the electricity purchase price and the electricity selling price for time period t on day d, respectively, and η be the charging efficiency of the electric vehicle. Energy loss during the journey, To account for the queuing time loss, dist(L i U j () represents the j-th charging demand point N. j To L i distance, These represent the power consumption per unit distance and the cost per unit waiting time, respectively, f(L) i N j ) is N j To L i The decision variable is whether the distance satisfies the distance coverage constraint. N is the time period t on day d. j The charging needs, For distance at r max The total number of charging stations in the area, r max The maximum allowable distance between adjacent charging stations is T(d,t,λ), and the queuing time of users within time period t on day d is T(d,t,λ). The constraints include the power constraints of the charging station, the distance constraints, and the maximum waiting time constraints for charging users; S3. Solve the charging station planning model in a coarse-grained manner to obtain the initial planning scheme for electric vehicle charging stations, including the initial site selection points and the number of fast charging piles and slow charging piles at each initial site selection point; S4. The initial planning scheme of electric vehicle charging stations is used as the initial population of the whale optimization algorithm. The whale optimization algorithm is used to search for and optimize the electric vehicle charging station planning scheme, including the final site selection point and the number of fast charging piles and slow charging piles at each final site selection point.

2. The electric vehicle charging station planning method considering charging demand according to claim 1, characterized in that, Specifically, S3 involves: first, using the farthest point set priority algorithm to determine the initial site selection point; then, based on the initial site selection point, using the tabu search algorithm to solve the charging station planning model to obtain the number of fast charging piles and slow charging piles at each initial site selection point. The method of determining the initial site selection point using the farthest point set priority algorithm includes: S31. Initialize the location point set P set Candidate point set C set , where P set Initially, it is an empty set; S32, Calculate C set The distance between any two candidate points in C is used to determine the two candidate points with the greatest distance from each other. set Move into P set middle; S33, Calculate C respectively set From each candidate point to P set The sum of distances to all candidate points within the range, and sorted in descending order of distance; S34. Starting with the candidate point with the largest sum of distances, sequentially determine whether the candidate point satisfies the distance constraints between location points, the distance constraints between the candidate point and the demand point, and the coverage constraints of the demand point. Then, select the first candidate point that satisfies all constraints from C. set Move into P set middle; S35. Determine whether all charging demand points satisfy the following constraints: In the above formula, Dist(N) j ,L i () represents the j-th charging demand point N. j To the i-th candidate point L i The distance, r max This represents the maximum allowable distance between adjacent charging stations. These are the set of charging demand points and the set of candidate points, respectively. If satisfied, then P at this time set Output the initial location point; if the condition is not met, return to S33 for the next iteration.

3. The electric vehicle charging station planning method considering charging demand according to claim 1, characterized in that, S1 uses Monte Carlo simulation to obtain the daily charging demand of electric vehicles at various time periods throughout the year at each demand point, including: S11. Within the selected area, determine the demand points for electric vehicles by community, school, or street. S12. The travel behavior of electric vehicle users is described using a travel chain. The probability that a user travels at any time and then transfers to another location is represented by the following matrix: In the above formula, R′ is the transition probability matrix, p ab Let be the transition probability of an electric vehicle from node a to node b; S13. The electric vehicle generates the destination and plans the driving route based on the transition probability matrix of the corresponding time period; S14. During the simulated electric vehicle driving process, the battery level of each electric vehicle is monitored in real time, and it is determined whether it is lower than the set threshold. If it is lower than the set threshold, its charging demand is recorded, thereby obtaining the charging demand of electric vehicles at each node at different time periods.

4. The electric vehicle charging station planning method considering charging demand according to claim 3, characterized in that, The p ab The data obtained by fitting using a multilayer perceptron includes: first, collecting historical data of each demand point within the selected area, including the date of worship, season category, ambient temperature, and travel data of node a and node b; then, inputting the historical data into the multilayer perceptron neural network to fit and obtain p. ab In this neural network, the hidden layer uses the ReLU function as the activation function, the output layer uses the Sigmoid activation function, and the loss function is the mean square error between the fitted probability value and the actual probability value.

5. An electric vehicle charging station planning system that considers charging demand, characterized in that, The system includes a charging demand simulation module, a planning model construction module, a coarse-grained solution module, and a fine-grained solution module. The charging demand simulation module is used to simulate the charging demand of electric vehicles at various demand points in different time periods. The planning model construction module is used to construct a charging station planning model based on charging demand. The charging station planning model aims to minimize the annualized loss of charging stations and the loss to users. Its objective function includes: C G =C CS +C O +C B -C S +C T In the above formula, Let M be the objective function. i Let i be the decision variable for whether to establish a charging station at the i-th candidate point. The i-th candidate point L i The number of fast charging stations and slow charging stations, C G C U Let C represent the annualized loss of the charging station and the user loss, respectively, with α and β being the weights of the annualized loss of the charging station and the user loss, respectively. CS The annualized total cost of building a charging station, C O For the operation and maintenance costs of charging stations, C B For the electricity purchase cost of charging stations, C S For the revenue from selling electricity at the charging station, C T The additional line loss cost for charging electric vehicle users, w0 and y0 are the depreciation rate and total service life, respectively, and C is the total service life. fix C land C buy These include the fixed asset cost of the charging station, the land acquisition cost, and the cost of purchasing charging piles. L is the time period t on day d. i The sum of charging demands at surrounding charging points, EC is the average charging capacity per electric vehicle, k is the unit operating cost, and a and b are the number of candidate charging points and charging demand points, respectively. Let be the electricity purchase price and the electricity selling price for time period t on day d, respectively, and η be the charging efficiency of the electric vehicle. Energy loss during the journey, To account for the queuing time loss, dist(L i U j () represents the j-th charging demand point N. j To L i distance, These represent the power consumption per unit distance and the cost per unit waiting time, respectively, f(L) i N j ) is N j To L i The decision variable is whether the distance satisfies the distance coverage constraint. N is the time period t on day d. j The charging needs, For distance at r max The total number of charging stations in the area, r max The maximum allowable distance between adjacent charging stations is T(d,t,λ), and the queuing time of users within time period t on day d is T(d,t,λ). The constraints include the power constraints of the charging station, the distance constraints, and the maximum waiting time constraints for charging users; The coarse-grained solution module is used to coarsely solve the charging station planning model to obtain the initial planning scheme for electric vehicle charging stations, including the initial site selection points and the number of fast charging piles and slow charging piles at each initial site selection point. The fine-grained solution module is used to take the initial planning scheme of electric vehicle charging stations as the initial population of the whale optimization algorithm, and use the whale optimization algorithm to search and optimize to obtain the electric vehicle charging station planning scheme, including the final site selection point and the number of fast charging piles and slow charging piles at each final site selection point.

6. The electric vehicle charging station planning system considering charging demand according to claim 5, characterized in that, The coarse-grained solution module includes an initial site selection point determination unit and a charging station initial planning scheme determination unit. The initial site selection point determination unit is used to determine the initial site selection point using the farthest point set priority algorithm. The specific process includes: A1. Initialize the location point set P set Candidate point set C set , where P set Initially, it is an empty set; A2. Calculate C set The distance between any two candidate points in C is used to determine the two candidate points with the greatest distance from each other. set Move into P set middle; A3. Calculate C respectively. set From each candidate point to P set The sum of distances to all candidate points within the range, and sorted in descending order of distance; A4. Starting with the candidate point with the largest sum of distances, sequentially determine whether the candidate point satisfies the distance constraints between location points, the distance constraints between the candidate point and the demand point, and the coverage constraints of the demand point. Then, select the first candidate point that satisfies all constraints from C. set Move into P set middle; A5. Determine whether all charging demand points satisfy the following constraints: In the above formula, Dist(N) j ,L i () represents the j-th charging demand point N. j To the i-th candidate point L i The distance, r max This represents the maximum allowable distance between adjacent charging stations. These are the set of charging demand points and the set of candidate points, respectively. If satisfied, then P at this time set Output the initial address point; if the condition is not met, return to A3 for the next iteration. The initial planning scheme determination unit for the charging station is used to solve the charging station planning model based on the determined initial site selection point using a tabu search algorithm, and to obtain the number of fast charging piles and slow charging piles at each initial site selection point.

7. The electric vehicle charging station planning system considering charging demand according to claim 5, characterized in that, The charging demand simulation module uses Monte Carlo simulation to obtain the daily charging demand of electric vehicles at various time periods throughout the year at each demand point. The specific process includes: a1. Within the selected area, determine the demand points for electric vehicles by community, school, or street. a2. The travel chain is used to describe the travel behavior of electric vehicle users. The probability that a user travels at any time and transfers to another location is represented by the following matrix: In the above formula, R′ is the transition probability matrix, p ab Let be the transition probability of an electric vehicle from node a to node b; a3. Electric vehicles generate destinations and plan driving routes based on the transition probability matrix of the corresponding time period; a4. During the simulated driving of electric vehicles, the battery level of each electric vehicle is monitored in real time, and it is determined whether it is lower than the set threshold. If it is lower than the set threshold, its charging demand is recorded, thereby obtaining the charging demand of electric vehicles at each node at different time periods.

8. An electric vehicle charging station planning system considering charging demand according to claim 7, characterized in that, The p ab The data obtained by fitting using a multilayer perceptron includes: first, collecting historical data of each demand point within the selected area, including the date of worship, season category, ambient temperature, and travel data of node a and node b; then, inputting the historical data into the multilayer perceptron neural network to fit and obtain p. ab In this neural network, the hidden layer uses the ReLU function as the activation function, the output layer uses the Sigmoid activation function, and the loss function is the mean square error between the fitted probability value and the actual probability value.

Images