A charging station planning method and system considering charging station capacity under range uncertainty

Abstract

The present invention discloses a charging station planning method and system considering the capacity of charging stations under the uncertainty of driving range, the method comprising: setting the minimum battery capacity of electric vehicles considering the uncertainty of driving range, combining the maximum battery capacity of electric vehicles and considering the allowable deviation from the shortest path, expanding the original inter-city transportation network diagram, and determining the charging strategy of electric vehicles to charge to a sufficient amount of electricity to travel to the next charging station; constructing the queuing model of the charging station as a multi-service station queuing model; based on the extended inter-city transportation network diagram, under the uncertainty of driving range, from the perspective of inter-city travel, combining the queuing model and the charging strategy to establish a site selection model focusing on cost factors, with the goal of minimizing the amount of electric vehicle charging in transit, the total fixed cost of the charging station, the variable cost of the charging pile, and the waiting time cost of the electric vehicle in the queuing station; solving the site selection model using a simulated annealing algorithm. The site selection method of the present invention is practical and has strong practicality.

Description

Technical Field

[0001] The present invention relates to the field of charging pile planning, and in particular to a charging station planning method and system that considers the capacity of the charging station under uncertainty of battery life. Background Art

[0002] In recent years, the scarcity of non-renewable resources and global warming have aroused global attention to environmental sustainability. In order to achieve a cleaner and sustainable transportation future, it is important to promote new energy vehicles. At present, a considerable number of charging stations have been built in major urban areas in China, and electric vehicles can basically complete urban travel without anxiety. However, the current charging infrastructure construction in the intercity road network is insufficient, and the number of charging facilities is small, which poses certain challenges for long-distance travel of electric vehicles. In addition to the difficulty in finding charging piles, the actual mileage of electric vehicles is also an important factor affecting travel.

[0003] The planning problem of electric vehicle charging stations is mainly about how to select the location of charging stations to meet the travel needs of electric vehicles in the transportation network. Early research on the location of electric vehicle charging stations was mainly based on the location research of gas stations. The earliest gas station location model proposed in this field assumes that all traffic flows follow the shortest path rule, thereby allocating the flow of each OD pair (O: Origin, starting point; D: Destination, end point) in the transportation network. This study established a mixed integer programming model, the goal of which is to determine the location of the gas station by selecting a set of nodes to maximize the number of feasible round trips or meet the maximum coverage requirements. Subsequently, research on range uncertainty began to appear, but there are very few related research projects and results. Moreover, the treatment of range uncertainty is relatively simple, and the existing related models are still in the exploratory stage. The model is simple and the rationality needs to be explored. Similarly, in recent years, research on the planning of charging stations in long-distance transportation networks has appeared, and these studies do not consider demand uncertainty, because the travel demand of traffic flow in long-distance road networks is more stable than that within the city. However, the impact of battery life uncertainty on long-distance travel is foreseeable. Therefore, the present invention studies the planning of charging stations in inter-city road networks that takes into account battery life uncertainty, and further provides a reasonable way to deal with battery life uncertainty. At the same time, path deviations are allowed to make the model closer to reality. The capacity of the charging station is also considered to determine the location and capacity plan of the charging station. Summary of the invention

[0004] Purpose of the invention: In view of the shortcomings of the prior art, the present invention proposes a charging station planning method and system that takes into account the capacity of charging stations under the uncertainty of battery life. It provides a method for determining a reasonable and effective charging station site selection and capacity determination plan under the limitation of charging station capacity when the battery life is uncertain and the path may deviate. This ensures that new energy vehicles can be well replenished during inter-city long-distance travel, alleviating the anxiety of inter-city travel for new energy vehicles.

[0005] Technical solution: According to a first aspect of the present invention, a charging station planning method considering the capacity of the charging station under the uncertainty of driving range is provided, comprising the following steps:

[0006] Considering the uncertainty of driving range, setting the minimum battery capacity r of electric vehicles min , combined with the maximum battery capacity of electric vehicles max It also considers allowing deviation from the shortest path, expands the original inter-city transportation network diagram, and determines that the charging strategy for electric vehicles is to stop charging when the amount of electricity is enough to drive to the next charging station;

[0007] The queuing model of the charging station is constructed as a multi-service station queuing model M / M / C model;

[0008] Based on the extended inter-city transportation network diagram, under the uncertainty of driving range, a location selection model focusing on cost factors is established from the perspective of inter-city travel, combining the queuing model and the charging strategy. The location selection model aims to minimize the amount of electric vehicles charged on the way, the total fixed cost of the charging station, the variable cost of the charging pile, and the waiting time cost of electric vehicles in the queuing station.

[0009] The simulated annealing algorithm is used to solve the site selection model under given constraints to determine the node locations of the charging stations to be established and the number of charging piles installed at each charging station.

[0010] According to a second aspect of the present invention, there is provided a charging station planning system considering the capacity of the charging station under the uncertainty of the driving range, comprising:

[0011] Preprocessing module, considers the uncertainty of battery life to set the minimum battery capacity r of electric vehicles min , combined with the maximum battery capacity of electric vehicles max It also considers allowing deviation from the shortest path, expands the original inter-city transportation network diagram, and determines that the charging strategy for electric vehicles is to stop charging when the amount of electricity is enough to drive to the next charging station;

[0012] The queuing model building module builds the queuing model of the charging station into a multi-service station queuing model M / M / C model;

[0013] The site selection model building module is based on the extended inter-city transportation network diagram. Under the uncertainty of driving range, the site selection model focusing on cost factors is established from the perspective of inter-city travel by combining the queuing model and the charging strategy. The site selection model aims to minimize the amount of electric vehicles charged on the way, the total fixed cost of the charging station, the variable cost of the charging pile, and the waiting time cost of electric vehicles in the queue at the charging station.

[0014] The solution module uses the simulated annealing algorithm to solve the site selection model under given constraints to determine the node locations of the charging stations to be established and the number of charging piles installed at each charging station.

[0015] According to a third aspect of the present invention, there is provided a charging station planning device, the device comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the charging station planning method considering the charging station capacity under the uncertainty of battery life described in the first aspect are implemented.

[0016] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a charging station planning method considering the capacity of a charging station under range uncertainty as described in the first aspect of the present invention.

[0017] Beneficial effects: When studying the site selection planning problem considering the capacity of charging stations, the present invention takes minimizing the amount of charging in transit, the construction cost of charging stations, the cost of charging piles, and the cost of queuing time as the goal of modeling. The model is solved by a simulated annealing algorithm. In the process of generating new solutions, the selection of new nodes as charging stations depends on the traffic flow of the nodes, and the addition of charging piles at the charging stations depends on the queuing time of the nodes, and vice versa. The present invention provides a reasonable site selection method for handling the uncertainty of battery life under the consideration of the capacity of charging stations, making the relevant modeling methods of electric vehicle charging closer to reality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is an example diagram of a charging station queue scenario;

[0019] Figure 2 is a flow chart of the method of the present invention;

[0020] Figure 3 To expand the example diagram of the transportation network;

[0021] Figure 4 Example diagram of charging strategy for electric vehicles;

[0022] Figure 5 An example diagram of a queuing model for an electric vehicle charging station;

[0023] Figure 6 It is the flow chart of simulated annealing algorithm;

[0024] Figure 7 To verify the traffic network diagram in the example;

[0025] Figure 8 To verify the numerical performance of the simulated annealing algorithm in the example;

[0026] Fig. 9 This is the node diagram for charging station site selection and capacity determination in the verification example. DETAILED DESCRIPTION

[0027] The technical solution of the present invention is further described below with reference to the accompanying drawings.

[0028] As electric vehicles become more popular, the long queues at charging stations are an important issue that needs to be solved urgently. Figure 1 . The queuing time at the charging station is not only related to the traffic flow, but also to the number of charging piles installed in the charging station. A larger number of charging piles means a higher service rate level. At the same time, the queuing time at the charging station will also affect the travel route selection of electric vehicles. Too long queuing time will cause the driver to give up charging at a certain station, and even choose to detour to other stations to charge. In addition, the construction cost of the charging station is no longer just a fixed cost, but consists of two parts, one is the fixed cost, and the other is the variable cost that changes with the increase in the number of charging piles. The present invention proposes a charging station planning method that considers the capacity of the charging station under the uncertainty of battery life, which is applied to the determination of charging stations between inter-city road networks and charging piles in the station. Since the capacity of the charging station is taken into account, the complexity of the research is significantly increased, because one more decision needs to be made, namely the determination of the number of charging piles. At the same time, this decision will also affect decisions such as the site selection of the charging station and the travel route, and a reasonable model needs to be constructed for analysis.

[0029] Reference Figure 2 , the method of the present invention comprises the following steps:

[0030] Step S1: Considering the uncertainty of driving range and allowing deviation from the shortest path, the intercity transportation network diagram is expanded.

[0031] Since the actual range of electric vehicles is uncertain, even if the range claimed by the manufacturer is known, this cannot guarantee that the vehicle will reach the upper limit of the range in actual driving. To deal with this situation, a minimum battery capacity r can be set min . r min The setting is actually to simulate the driver's control of the electric vehicle's range during the trip. It is not the minimum remaining range displayed on the electric vehicle dashboard, but the power reserved to deal with the uncertainty of the range, or the power that cannot be used for the trip. For example, in the winter in northern my country, the low temperature makes the battery itself consume power faster, and the use of air conditioning in the car also greatly reduces the range achievement rate. At the time of departure, the actual achievement rate of the remaining range displayed on the dashboard is only between 40% and 60%.

[0032] The site selection planning in the present invention adopts a full coverage model, which separates the processing of battery life uncertainty from the site selection model, and pre-processes this problem by expanding the traffic network. This processing method allows the vehicle driving route to deviate from the minimum route to be included in the model. The site selection model, planning model, charging site selection model, and full coverage model in the present invention refer to the same model, namely the charging station site selection planning model, also referred to as the model. Compared with the existing technologies, all assume the shortest path principle for expansion. The expanded traffic network is established as follows: First, consider the original traffic network G 0 =(N 0 ,A 0 ), that is, |N 0 | nodes and |A 0 | arc segments, among which there are |K| OD pairs, O k and D k Respectively represent the starting point and end point of the k∈Kth OD pair. First, add a virtual node and Respectively with O k and D k There are virtual arcs with zero distance between them to ensure the feasibility of vehicle round trip. Then, when the shortest path distance between any two nodes does not exceed (r max -r min ) corresponds to the distance length, where r max is the maximum battery capacity of the electric vehicle, which means that the vehicle can travel between these two nodes without charging. An extended arc segment can be added between the two points, and the length of the arc segment is the shortest distance between the two points. Figure 3 This is an example diagram that demonstrates how to build an extended transportation network diagram.

[0033] Under the above rules, an extended transportation network is established. Figure 3 The percentage on the middle arc segment indicates the percentage of maximum power consumption required to pass through the arc segment, and the black part represents the original traffic network. min When it is set to 30%*R, R is the claimed range of the electric vehicle. The electric vehicle can travel from a to b and then to c without charging. Then a new arc can be added between a and c. Repeating the above steps for any two nodes in the network graph gives Figure 3 The specific mathematical expression for establishing an extended traffic network is as follows: First, add virtual nodes to the original traffic network. and

[0034]

[0035]

[0036]

[0037] Indicates road segment Length, Indicates road segment Length.

[0038] remember is the shortest distance between any two nodes i, j, i, j∈N. The addition of extended arcs is expressed as:

[0039]

[0040]

[0041] At this time, r min According to the research idea of ​​the opportunity constraint model, when it is known that the endurance achievement follows a normal distribution, a security threshold is set, that is, the endurance achievement is not less than r max -r min The proportion of vehicles, r min It can be obtained, and then a charging station site selection model covering all OD pairs is established on this basis. This processing method makes it unnecessary for the endurance uncertainty to appear in the subsequent model, so that the allowed path deviation can be included in the model. This processing method is actually to selectively abandon the vehicles with very low endurance in the traffic flow. For example, when the threshold is set to 95%, it means that the 5% of vehicles with the worst endurance are selectively abandoned. Compared with the use of a random programming model, this processing method has more advantages. At the same time, it is expected that for a random programming model, vehicles with the same endurance achievement rate may be able to complete the journey smoothly between some OD pairs, but cannot complete the journey between other OD pairs. That is, it may cause the distribution of charging stations on some paths to have high requirements for endurance, while some other paths have very low requirements for endurance. This solution may be the best for the system, but it is very inconsistent with the actual situation and will cause great confusion to electric vehicle drivers. Although the processing of the model of the present invention seems simple and direct, it sets a fair criterion for the traffic flow of all OD pairs in the system, ensuring the consistency of the final site selection plan within the system. Drivers are rational. When they know that their vehicle's range is low, they will not drive the vehicle on long-distance highways. When dealing with uncertainty, what can be done is to minimize the consequences of uncertainty. And for the construction of charging stations for transportation networks, which is an infrastructure project for the whole society, it is very important to know the consequences of uncertainty.

[0042] Below through Figure 4 An example of a charging strategy for an electric vehicle is given. Figure 4(a) gives the power consumption of each section, (b) and (c) give two different charging strategies. Both can successfully complete the journey while retaining a minimum range of 30%. The charging strategy in (b) is to fully charge at the charging station, while the charging strategy in (c) is to charge enough to reach the next charging station. Obviously, the latter takes less time on the road, so this method adopts the charging strategy of charging as much as needed. For the convenience of research and feasibility considerations, the model has the following assumptions:

[0043] (1) Electric vehicles do not need to be fully charged each time, they only need to be charged to a level sufficient to drive to the next charging station;

[0044] (2) Charging stations are not capacity-constrained, meaning that electric vehicles can charge without queuing when they arrive at a charging station;

[0045] (3) The types of vehicles in the traffic network are the same;

[0046] (4) Electric vehicles are charged at the starting point of each road section;

[0047] (5) When the electric car starts from the starting point, the battery is fully charged, that is, the power is r max In addition, due to the uncertainty of battery life, the battery power should be greater than or equal to the minimum value r during the entire driving process. min .

[0048] Step S2: construct a charging station queuing model in an inter-city travel scenario.

[0049] In inter-city travel scenarios, roadside charging stations will have multiple charging piles. When a vehicle arrives at a charging station, if there is a vacant space, it can be charged. Otherwise, it needs to queue up. Figure 5 As shown, the queuing model of the charging station can be regarded as a multi-server queuing model (M / M / C model). Therefore, for the queuing model of electric vehicles, the present invention has the following assumptions:

[0050] (1) The customer input process is that electric vehicles arrive at the charging station for charging. The number of arrivals in non-overlapping time intervals is independent of each other; the probability of a car arriving in any time interval is independent of the time starting point and only related to the time interval span; when the time interval span is close to 0, the probability that the number of arriving vehicles is greater than 1 is 0. The above assumptions satisfy the Poisson distribution's no aftereffect, stationarity, and orderliness, and it can be assumed that the electric vehicles arriving at the charging station follow a Poisson distribution with parameter λ;

[0051] (2) The queuing system for electric vehicle charging stations is a waiting system. When an electric vehicle arrives at a charging station, there are n fast charging piles in the charging station. Each charging pile has the same basic properties, and the charging services of each charging pile are independent of each other. If all charging piles in the charging station are in use, the electric vehicle will wait in line until there is an idle charging pile available for charging;

[0052] (3) The charging service in the charging station is a parallel charging service with multiple charging piles, and electric vehicle users follow the first-come, first-served rule. The time spent on electric vehicle charging follows a negative exponential distribution with parameter μ.

[0053] Assume that the number of customers arriving is m. When m≤n, there are m charging piles in the charging station that are occupied, and the remaining (nm) charging piles are idle. When m>n, the number of electric vehicle users who need to charge in the charging station exceeds the number of charging piles. All charging piles in the charging station are occupied, and only n electric vehicle users can charge. The remaining (mn) users are waiting in line. Figure 5 Assume that the queue in the charging station is a single column. Based on the above assumptions and combined with the M / M / C queue theory, the equation when the queue system is in equilibrium is:

[0054]

[0055] Where m represents the number of electric vehicle customers who need charging services, P m represents the probability that m electric vehicle users in the queuing system accept charging services, and n represents the number of charging piles in the charging station. And by the regularity condition We can get:

[0056]

[0057] P0 represents the probability that no electric vehicle user in the queuing system accepts charging service.

[0058] According to the queuing theory, the following indicators can be calculated:

[0059] Charging station service intensity:

[0060]

[0061] Utilization rate of charging stations:

[0062]

[0063] Average queue length at charging stations, including is ρ1 to the power of n:

[0064]

[0065] Average captain of a charging station:

[0066]

[0067] Average waiting time at a charging station:

[0068]

[0069] Average duration of stay at a charging station:

[0070]

[0071] The relevant symbols and parameters in the model are as follows:

[0072] G(N,A): transportation network;

[0073] N: the set of nodes in the transportation network;

[0074] A: the set of road segments arc(i,j) in the traffic network, arc(i,j)∈A,i∈N,j∈N;

[0075] l ij : The length of the road segment arc(i,j);

[0076] k: OD pair k, k∈K;

[0077] O k : The starting point of OD for k;

[0078] D k : The endpoint of OD versus k;

[0079] d k : OD traffic flow to k;

[0080] f i : Fixed fee for a single charging station;

[0081] f h : The cost of a single charging station;

[0082] μ: service rate of a single charging pile;

[0083] λ i : Customer arrival rate at node i;

[0084] η1: weight coefficient;

[0085] η2: Charging time calculation coefficient;

[0086] θ: time value coefficient;

[0087] The queuing time at node i;

[0088] R: claimed range of electric vehicles;

[0089] r min : The minimum battery level of an electric vehicle during its journey, between [0,1], which is the percentage of full range.

[0090] The decision variables in the model are as follows:

[0091] y i : 0-1 variable, if a charging station is built at node i, it is 1, otherwise it is 0;

[0092] x ijk : 0-1 variable, 1 if the path of OD pair k passes through segment arc(i,j), 0 otherwise;

[0093] t k : The amount of electricity that needs to be charged for the vehicles between OD and k to complete the trip;

[0094] r i : The charge at the starting point and the end point, i∈U k∈K {O k ,D k};

[0095] n i : The number of charging piles established at node i.

[0096] Step S3: Establish an inter-city road network charging station site selection model that takes into account queuing time under uncertain driving range.

[0097] Based on the basic assumptions of the charging station site selection problem, under the uncertainty of battery life, from the perspective of inter-city travel, a site selection model focusing on cost factors is established in combination with the queuing model, with the goal of minimizing the amount of charging on the way, construction cost, and queuing waiting time cost:

[0098]

[0099] The objective function consists of three parts. The first part is to minimize the on-the-way charging time cost required to complete all OD trips. At the same time, due to the charging strategy of this method, it also indirectly minimizes the deviation of the driving path from the shortest path; the second part is to minimize the total fixed cost of building charging stations and the variable cost of charging piles; the third part is the waiting time cost of electric vehicles at charging stations. The total cost of the objective function is to minimize the relevant costs from a comprehensive perspective.

[0100] The model follows the following constraints:

[0101]

[0102] ∑ i:(i,j)∈A x ijk ≤yi ,j∈N o ,k∈K (2)

[0103]

[0104]

[0105]

[0106] λ i =∑ k∈K d k x ijk ,(i,j)∈A,k∈K (6)

[0107] μ i =y i μ,(i,j)∈A,k∈K (7)

[0108]

[0109] t k ≥0,k∈K (9)

[0110] r i ≥0,i∈{O k ∪D k},k∈K (10)

[0111] x ijk ∈{0,1},(i,j)∈A,k∈K (11)

[0112] y i ∈{0,1},i∈N o (12)

[0113] Constraint (1) is a flow balance constraint that ensures that each OD pair has a feasible path. Constraint (2) indicates that when a road segment is used, a charging station needs to be built at its starting point, except for virtual stations. Constraint (3) defines the decision variable t k , which is the amount of charging required for an OD trip. Constraints (4) and (5) ensure the assumptions about the battery power of the vehicle at the starting point and the end point. Constraint (4) indicates that if there is a charging station at the starting point, the battery percentage is 100%, otherwise it is 50%, to ensure round trip; similarly, constraint (5) indicates that if there is a charging station at the end point, the actual available power status does not need to be considered, otherwise it needs to be no less than 50%. The battery constraint here does not need to consider the uncertainty of the battery life, because the previously expanded traffic network model ensures that the vehicle can pass the distance between any two candidate charging stations. Constraint (6) calculates the arrival rate λ of each node i, which is the average traffic volume arriving at the point per unit time. Constraint (7) represents the service rate at the node. If there is no charging station at the node, the service rate is 0. Constraint (8) calculates the queuing time of each node according to the queuing formula. Constraints (9)-(12) represent the value range of the decision variables.

[0114] Step S4: using simulated annealing algorithm to solve the site selection model.

[0115] The simulated annealing algorithm applied to solve the location planning model contains two processes. The first is an iterative process, which generates new solutions (at which nodes to build charging stations, how many charging piles to install at each charging station) by perturbing the solution in the previous iteration, which is called the adjacent solution. Then, the objective function value of the adjacent solution needs to be calculated. To do this, the charging behavior of the user (driver) needs to be determined based on the route selection and the choice of charging station among the existing options in the current solution. Since each user's charging decision is affected by the charging decisions of other users, it is necessary to find an equilibrium state by solving the queuing congestion problem, which specifies the routes and charging station choices for all users. Users choose routes to minimize the total delay, including charging time (charge amount) and waiting time in queues. Therefore, the algorithm needs a second process to find this equilibrium state. Once the equilibrium state is reached, the objective function value of the adjacent solution can be calculated. In the simulated annealing algorithm, the new objective function value of the adjacent solution needs to be compared with the objective function value of the current solution. In this algorithm, if the objective function value is improved, the adjacent solution will be determined to be accepted as the new current solution. Based on the difference between the old and new objective function values, it has a probability of being accepted even if it does not improve the objective function. This ensures that the algorithm avoids getting stuck in a local optimum. Once a neighboring solution is accepted as the new current solution or rejected (so there is no change in the current solution), another neighboring solution will be randomly generated based on the current solution in the next iteration. This process will continue for a certain number of iterations to find the optimal solution for the objective function value. As the iterative process progresses, the probability of accepting a neighboring solution with a worse function value becomes smaller and smaller, thereby ensuring that the optimal function value is obtained.

[0116] 4.1. Algorithm design:

[0117] In order to apply the simulated annealing algorithm framework to solve the proposed model, assumptions and definitions are required to generate feasible initial solutions and neighboring solutions.

[0118] First, the variable related to the number of charging posts needs to be modified. If the number of charging posts at each station is allowed to be any value, there will be an infinite number of feasible solutions, which greatly increases the difficulty of solving. Therefore, four levels ({low, medium, high, very high}, corresponding to {5, 10, 20, 30} charging posts) are defined for the number of points in each candidate station instead of a continuous integer variable. Each candidate station selected as part of any feasible solution must have one of these levels as its number of charging posts.

[0119] In order to obtain an initial solution, rather than a random initial solution, set y for all candidate nodes i =1 to ensure the feasibility of the initial solution. This assumption ensures that there is at least one feasible path for each OD pair in the network. For the number of points, a moderate level of all candidate charging stations in the initial solution is assumed.

[0120] In order to transfer from the current solution to the neighboring solution, a weighted random method is used to improve the efficiency of the algorithm. There are multiple possible perturbations in the current solution, such as adding or removing candidate stations or increasing or decreasing the number of points at the candidate station level. Depending on the type of perturbation, each location is associated with a weight factor (e.g., total traffic, total delay). Then, the location is selected based on weighted random selection. Specifically, define:

[0121]

[0122] Among them, φ i is the weight coefficient of node i, Φ(N) is the cumulative weight coefficient, and node l is selected if it meets the following conditions:

[0123]

[0124] Among them, ρ∈[0,1], obeys uniform distribution.

[0125] When generating adjacent solutions, one of the following four rules is randomly followed. If a rule cannot generate a feasible solution, it is randomly switched to another rule:

[0126] Rule 1: When a new node is selected, it is not currently in the current solution, and its weight value is the traffic flow passing through the node, that is, φ i =∑ k∈K d k x ijk This means that nodes with large traffic flow will be preferentially selected to join the current solution and be equipped with a medium number of charging piles.

[0127] Rule 2: When a new node is removed, it is currently in the current solution, and its weight value is the inverse of the traffic flow passing through the node, that is, φ i =1 / ∑k∈K d k x ijk This means that nodes with low traffic volume will be selected first to eliminate the current solution. However, when eliminating a selected node makes an OD pair infeasible, the perturbation process will be repeated. In this case, the weight coefficient is calculated based on a new set, which includes all current nodes and the node mentioned above.

[0128] Rule 3: When adding a charging pile to a node, the node chooses which of the low, medium, and high solutions in the current solution, which depends on the queue time of the node, that is, This means that nodes with longer queuing times will be selected to add charging piles first.

[0129] Rule 4: When the number of charging piles is reduced at a node, the node chooses which of the low, medium, and high options in the current solution, which depends on the inverse of the queue time of the node, that is, This means that nodes with shorter queue times will be selected to reduce charging piles first.

[0130] 4.2. Traffic Assignment Problem

[0131] Due to the congestion and queues at charging stations and the goal of minimizing the total amount of charging, travelers cannot choose their own routes. Therefore, it is necessary to model the problem. Under the premise of the current solution, it can be considered that in the current iteration, the construction of charging stations and charging piles in the transportation network is completed, then the initial site selection planning model is simplified to minimize the total queue time and total charging amount, that is, the objective function is simplified to:

[0132]

[0133] Constraints: (1), (2), (3), (6), (7), (8), (9), (10), (11).

[0134] The above model only has one decision variable x ijk , where y i 、n i is input as a parameter. In addition, constraints (4) and (5) have been satisfied in the process of generating the current solution, so r i There is no need to consider it anymore. The model only needs to make one decision, which is the driver's driving route. Since this model adopts the charging strategy of "charging as much as used", this decision can also be regarded as the driver's charging strategy. The model can be solved directly by the solver.

[0135] 4.3. Simulated annealing algorithm steps

[0136] Based on the design of the simulated annealing algorithm above, the implementation steps of the algorithm are as follows:

[0137] Step 1. Let T = T0, i.e., the initial temperature for starting annealing, generate a feasible initial solution, and calculate the corresponding objective function value Γ0 based on the original objective function.

[0138] Step 2. Set T equal to the next value T in the cooling schedule i .

[0139] Step 3. Perform sub-iteration and, according to rules 1-4, perturb the current solution, change the number of charging piles of the selected node, or add and remove nodes to generate a new solution.

[0140] Step 4. Perform traffic flow analysis on the road network based on the new solution to obtain the driver path and queue time, and calculate the corresponding objective function value Γ according to the original objective function j , we get ΔΓ=Γ j -Γ0.

[0141] Step 5. If ΔΓ<0, the new solution is accepted as the new current solution; if ΔΓ>0, the new solution is added with probability exp(-ΔΓ / T i )accept.

[0142] Step 6. Update the sub-iteration count.

[0143] Step 7. If the sub-iteration count is less than the maximum number of sub-iterations, go to step 3.

[0144] Step 8. When the sub-iteration count reaches the maximum number of sub-iterations, update the main iteration count and go to step 2.

[0145] Step 9. When the main iteration count reaches the maximum number of main iterations, terminate the algorithm.

[0146] The simulated annealing algorithm process designed for the charging station planning problem considering the charging station capacity under the uncertainty of battery life is as follows: Figure 6 shown.

[0147] In order to verify the effectiveness of the proposed model and algorithm, the present invention has been specifically implemented. The traffic network example used is Figure 7 As shown. There are 9 city nodes in total, and the nodes are numbered AI. In addition, a candidate station is established every 30-40 kilometers in the road section. The interval is not fixed to ensure that each road section of different lengths can be equally divided. 31 road section candidate points are obtained, and the nodes are numbered 1-31. Finally, a total of 40 nodes can be used as candidate points for establishing charging stations. Each station is bilateral and can be used for charging vehicles in both directions. The traffic demand and the shortest distance between cities are shown in Tables 1 and 2.

[0148] Table 1. Traffic demand between cities (unit: vehicles / minute)

[0149]

[0150]

[0151] Table 2. Shortest distance between cities (in kilometers)

[0152]

[0153] The range R of the electric vehicle is set to 400 km, and the security level α is set to 80%. Other parameter settings are shown in Table 3.

[0154] Table 3. Model and algorithm parameter settings

[0155]

[0156] There are three key factors to consider during the simulated annealing algorithm calculation process: initial temperature and acceptance probability, cooling rate, and algorithm termination conditions.

[0157] (1) Initial temperature and acceptance probability: From the algorithm steps, we can see that the probability of accepting a worse solution in the i-th iteration is That is, it depends on the temperature of the i-th time. When the temperature is high, the probability of accepting the new solution will be higher, and when the temperature is low, the opposite is true. In order to determine the initial temperature, considering that the initial iteration hopes to try more solutions, an initial temperature that can give an initial acceptance probability of 80% can be selected, that is, through Find the initial temperature.

[0158] (2) Cooling rate: T i+1 =T i ×τ,0<τ<1,i=0…N trans For the cooling process, a maximum number of iterations N is set trans , reflects the number of iterations allowed in the two-iteration process. In each iteration, the temperature is multiplied by a decreasing coefficient. If it is too large, the temperature will decrease very slowly and convergence will be slow. On the other hand, if the value is too small, the temperature will decrease too quickly, so the algorithm may quickly get stuck in the local optimum.

[0159] (3) Algorithm termination condition: When the temperature drops to T0·ε, the algorithm terminates. ε is the termination criterion coefficient. Setting ε is to prevent the algorithm from stopping the cooling process when there is no progress due to too many iterations or too fast cooling speed. When the temperature is low enough, the algorithm will only accept better solutions and will eventually converge to the nearest local optimum. The setting of this coefficient depends on different problems and can be determined by pre-calculus.

[0160] The simulated annealing parameters are initially set as follows:

[0161] Set the maximum number of iterations both inside and outside to N trans =50;

[0162] The temperature reduction coefficient is τ = 0.85;

[0163] Termination standard coefficient ε=10 -4 .

[0164] In addition, in order to ensure the finite number of solutions, the number of charging piles at each station is not an arbitrary continuous integer, but is selected from {5, 10, 20, 30}. This rule is not only for the convenience of solving, but also more in line with reality, because in the actual highway network, charging stations of different specifications will be established according to certain standards, and there will not be a situation where the number of charging piles at each charging station is uneven.

[0165] Numerical experiments are performed to verify whether the selected number of main iterations and internal iterations can make the algorithm converge. The total cost changes with the number of iterations as shown in Figure 8 As shown, the number of iterations on the horizontal axis is based on consecutive main iterations and internal sub-iterations. Figure 8 It shows that as the number of iterations increases, the objective function value gradually converges. Figure 8 As shown in Figure 2, when the total number of iterations of the algorithm reaches 1500 (30 main iterations and 50 sub-iterations), the subsequent iterations cannot further optimize the iteration points. Therefore, the maximum number of iterations N set in the numerical experiment trans =50 is reasonable.

[0166] Table 4. Number of charging station nodes and charging piles

[0167]

[0168]

[0169] The solution results are shown in Table 4 and Fig. 9 As shown in Figure 5, 11 nodes are selected as charging stations, of which 4 nodes {1, C, 15, 21} are selected to install 5 charging piles; 2 nodes {A, D} are selected to install 10 charging piles; 3 nodes {E, H, I} are selected to install 20 charging piles; and 2 nodes {F, G} are selected to install 30 charging piles. The travel paths and charging plans of each OD pair are shown in Table 5.

[0170] Table 5. Travel routes and charging plans for each OD pair

[0171]

[0172] The present invention considers the location planning problem of charging station capacity, establishes a model with the goal of minimizing the amount of charging in transit, the construction cost of the charging station, the cost of the charging pile and the cost of the queuing time, and adopts the simulated annealing algorithm to solve it. In the process of generating a new solution, the selection of a new node as a charging station depends on the traffic flow of the node, and the addition of a charging pile at the charging station depends on the queuing time of the node, and vice versa. In terms of the number of charging piles, the choice between {5, 10, 20, 30} charging piles, recorded as {low, medium, high, very high} avoids infinite choices, and is also more in line with the establishment standards of charging stations in actual situations. Finally, it is solved that a charging station containing 4 low, 2 medium, 3 high, and 2 very high charging piles needs to be established.

[0173] Based on the same technical concept as the method embodiment, the present invention also provides a charging station planning system considering the capacity of the charging station under the uncertainty of the driving range, including:

[0174] Preprocessing module, considers the uncertainty of battery life to set the minimum battery capacity r of electric vehicles min , combined with the maximum battery capacity of electric vehicles max It also considers allowing deviation from the shortest path, expands the original inter-city transportation network diagram, and determines that the charging strategy for electric vehicles is to stop charging when the amount of electricity is enough to drive to the next charging station;

[0175] The queuing model building module builds the queuing model of the charging station into a multi-service station queuing model M / M / C model;

[0176] The site selection model building module is based on the extended inter-city transportation network diagram. Under the uncertainty of driving range, the site selection model focusing on cost factors is established from the perspective of inter-city travel by combining the queuing model and the charging strategy. The site selection model aims to minimize the amount of electric vehicles charged on the way, the total fixed cost of the charging station, the variable cost of the charging pile, and the waiting time cost of electric vehicles in the queue at the charging station.

[0177] The solution module uses the simulated annealing algorithm to solve the site selection model under given constraints to determine the node locations of the charging stations to be established and the number of charging piles installed at each charging station.

[0178] It should be understood that the charging station planning system that considers the capacity of the charging station under the uncertainty of the battery life in the embodiment of the present invention can implement all the technical solutions in the above-mentioned method embodiment, and the functions of its various functional modules can be specifically implemented according to the methods in the above-mentioned method embodiments. The specific implementation process can refer to the relevant description in the above-mentioned embodiments, which will not be repeated here.

[0179] The present invention also provides a charging station planning device, which includes: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the charging station planning method that considers the charging station capacity under the uncertainty of battery life as described above are implemented.

[0180] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the charging station planning method considering the charging station capacity under the uncertainty of the driving range as described above are implemented.

[0181] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices (systems), computer equipment or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0182] The present invention is described with reference to a flowchart of a method according to an embodiment of the present invention. It should be understood that each process in the flowchart and a combination of processes in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A device that specifies functions in a process or multiple processes.

[0183] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A function specified in a process or multiple processes.

[0184] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 The steps of a specified function in a process or multiple processes.

Claims

1. A charging station planning method considering charging station capacity under range uncertainty, characterized in that: The following steps are involved: Considering the uncertainty of driving range, setting the minimum battery capacity r of electric vehicles min , combined with the maximum battery capacity of electric vehicles max It also considers allowing deviation from the shortest path, expands the original inter-city transportation network diagram, and determines that the charging strategy for electric vehicles is to stop charging when the amount of electricity is enough to drive to the next charging station; The queuing model of the charging station is constructed as a multi-service station queuing model M / M / C model; Based on the extended inter-city transportation network diagram, under the uncertainty of driving range, a location selection model focusing on cost factors is established from the perspective of inter-city travel, combining the queuing model and the charging strategy. The location selection model aims to minimize the amount of electric vehicles charged on the way, the total fixed cost of the charging station, the variable cost of the charging pile, and the waiting time cost of electric vehicles in the queuing station. The simulated annealing algorithm is used to solve the site selection model under given constraints to determine the node locations of the charging stations to be established and the number of charging piles installed at each charging station; Among them, the method of expanding the original intercity transportation network graph is as follows: For the original traffic network G 0 =(N 0 ,A 0 ), that is, |N 0 | nodes and |A 0 | arc segments, among which there are |K| OD pairs, O k and D k Respectively represent the starting point and end point of the k∈Kth OD pair. First, add a virtual node and Respectively with O k and D k There are virtual arcs with zero distance between them, and then when the shortest path distance between any two nodes does not exceed r max -r min When the corresponding distance length is calculated, an extended arc segment is added between the two nodes, and the arc segment length is the shortest distance between the two nodes; The queue model is constructed as follows: Assume that the arrival of electric vehicles at charging stations follows a Poisson distribution with parameter λ, the charging service for electric vehicles follows the principle of first-come, first-served, the queue at the charging station is a waiting system, and the time spent on charging electric vehicles follows a negative exponential distribution with parameter μ; Assume that the number of cars arriving for charging is m. When m≤n, there are m charging piles in the charging station that are occupied, and the remaining (nm) charging piles are idle. When m>n, the number of electric vehicles that need to be charged in the charging station exceeds the number of charging piles. All charging piles in the charging station are occupied, and only n electric vehicles can be charged. The remaining (mn) are waiting in line. Combined with the M / M / C queuing theory, the equation when the queuing system is in equilibrium is: Where P m represents the probability that m electric vehicle users in the queuing system accept charging services, and the regularity condition Get the probability that no electric vehicle user in the queuing system accepts charging service The various indicators of the queuing model are calculated through queuing theory, including: Charging station service intensity: Utilization rate of charging stations: Average queue length at a charging station: Average captain of a charging station: Average waiting time at a charging station: Average duration of stay at a charging station: The objective function of the site selection model is as follows: The model follows the following constraints: ∑ i:(i,j)∈A x ijk ≤y i ,j∈N o ,k∈K (2) λ i =∑ k∈K d k x ijk ,(i,j)∈A,k∈K (6) m i =y i ·μ,(i,j)∈A,k∈K (7) t k ≥0,k∈K (9) r i ≥0,i∈{O k ∪D k },k∈K (10) x ijk ∈{0,1},(i,j)∈A,k∈K (11) and i ∈{0,1},i∈N o (12) Where N represents the set of nodes in the expanded traffic network; A represents the set of road segments arc(i,j) in the expanded traffic network, arc(i,j)∈A,i∈N,j∈N; N o represents the node set in the original traffic network; l ij represents the length of the road segment arc(i,j); k represents the OD pair k, k∈K; O k Indicates the starting point of OD to k; D k Indicates the end point of OD to k; d k represents the traffic flow of OD to k; f i represents the fixed cost of a single charging station; f h represents the cost of a single charging pile; μ represents the service rate of a single charging pile; λ i represents the customer arrival rate at node i; η1 represents the weight coefficient; η2 represents the charging time calculation coefficient; θ represents the time value coefficient; represents the queuing time at node i; R represents the claimed range of the electric vehicle; r min Indicates the minimum power of the electric vehicle during its trip, between [0,1], that is, the percentage of full range; y i is a 0-1 variable, indicating whether a charging station is established at node i. If yes, it is 1, otherwise it is 0; x ijk is a 0-1 variable, indicating whether the path of OD pair k passes through the segment arc(i,j). If yes, it is 1, otherwise it is 0; t k represents the amount of electricity required for the vehicles between OD pair k to complete their trip; r i Represents the electric charge at the starting point and the end point, i∈U k∈K {O k ,D k };n i Indicates the number of charging piles established at node i.

2. The method according to claim 1, characterized in that When solving the site selection model, each feasible solution uses the traffic flow of the node as the basis for the location of the charging station, and the queuing time of the node as the basis for adding charging piles; the number of charging piles in each candidate charging station is based on the pre-defined grade standard.

3. The method according to claim 2, characterized in that The predefined grading standards are four levels {low, medium, high, very high}, corresponding to {5, 10, 20, 30} charging piles respectively.

4. The method according to claim 2, characterized in that: In the simulated annealing algorithm, in each round of iteration, a weighted random method is used to transfer from the current solution to the adjacent solution according to the following rules. If a rule cannot generate a feasible solution, it is randomly switched to another rule: When a new node is selected that is not currently in the current solution, its weight value is the traffic flow through the node; When a new node is removed, it is currently in the current solution, and its weight value is the inverse of the traffic flow passing through the node; When adding a charging pile at a node, the node is in the current solution, then the choice of the charging pile quantity level standard depends on the queue time of the node; When reducing charging piles at a certain node, the node is in the current solution, and the selection of the charging pile quantity level standard depends on the inverse of the queuing time of the node.

5. A charging station planning system considering charging station capacity under range uncertainty, characterized in that: include: Preprocessing module, considers the uncertainty of battery life to set the minimum battery capacity r of electric vehicles min , combined with the maximum battery capacity of electric vehicles max Considering the deviation from the shortest path, the original inter-city transportation network diagram is expanded, and the charging strategy of electric vehicles is determined to stop charging when the electric power is enough to drive to the next charging station; the method of expanding the original inter-city transportation network diagram is as follows: For the original traffic network G 0 =(N 0 ,A 0 ), that is, |N 0 | nodes and |A 0 | arc segments, among which there are |K| OD pairs, O k and D k Respectively represent the starting point and end point of the k∈Kth OD pair. First, add a virtual node and Respectively with O k and D k There are virtual arcs with zero distance between them, and then when the shortest path distance between any two nodes does not exceed r max -r min When the corresponding distance length is calculated, an extended arc segment is added between the two nodes, and the arc segment length is the shortest distance between the two nodes; the queuing model construction module constructs the queuing model of the charging station as a multi-service station queuing model M / M / C model, and the method is as follows: Assume that the arrival of electric vehicles at charging stations follows a Poisson distribution with parameter λ, the charging service for electric vehicles follows the principle of first-come, first-served, the queue at the charging station is a waiting system, and the time spent on charging electric vehicles follows a negative exponential distribution with parameter μ; Assume that the number of cars arriving for charging is m. When m≤n, there are m charging piles in the charging station that are occupied, and the remaining (nm) charging piles are idle. When m>n, the number of electric vehicles that need to be charged in the charging station exceeds the number of charging piles. All charging piles in the charging station are occupied, and only n electric vehicles can be charged. The remaining (mn) are waiting in line. Combined with the M / M / C queuing theory, the equation when the queuing system is in equilibrium is: Where P m represents the probability that m electric vehicle users in the queuing system accept charging services, and the regularity condition Get the probability that no electric vehicle user in the queuing system accepts charging service The various indicators of the queuing model are calculated through queuing theory, including: Charging station service intensity: Utilization rate of charging stations: Average queue length at a charging station: Average captain of a charging station: Average waiting time at a charging station: Average duration of stay at a charging station: The site selection model construction module is based on the extended inter-city transportation network diagram. Under the uncertainty of driving range, a site selection model focusing on cost factors is established from the perspective of inter-city travel combined with the queuing model and charging strategy. The site selection model aims to minimize the amount of electric vehicle charging in transit, the total fixed cost of the charging station, the variable cost of the charging pile, and the waiting time cost of the electric vehicle in the charging station. The objective function form of the site selection model is as follows: The model follows the following constraints: ∑ i:(i,j)∈A x ijk ≤y i ,j∈N o ,k∈K (2) λ i =∑ k∈K d k x ijk ,(i,j)∈A,k∈K (6) m i =y i ·μ,(i,j)∈A,k∈K (7) t k ≥0,k∈K (9) r i ≥0,i∈{O k ∪D k },k∈K (10) x ijk ∈{0,1},(i,j)∈A,k∈K (11) and i ∈{0,1},i∈N o (12) Where N represents the set of nodes in the expanded traffic network; A represents the set of road segments arc(i,j) in the expanded traffic network, arc(i,j)∈A,i∈N,j∈N; N o represents the node set in the original traffic network; l ij represents the length of the road segment arc(i,j); k represents the OD pair k, k∈K; O k Indicates the starting point of OD to k; D k Indicates the end point of OD to k; d k represents the traffic flow of OD to k; f i represents the fixed cost of a single charging station; f h represents the cost of a single charging pile; μ represents the service rate of a single charging pile; λ i represents the customer arrival rate at node i; η1 represents the weight coefficient; η2 represents the charging time calculation coefficient; θ represents the time value coefficient; represents the queuing time at node i; R represents the claimed range of the electric vehicle; r min Indicates the minimum power of the electric vehicle during its trip, between [0,1], that is, the percentage of full range; y i is a 0-1 variable, indicating whether a charging station is established at node i. If yes, it is 1, otherwise it is 0; x ijk is a 0-1 variable, indicating whether the path of OD pair k passes through the segment arc(i,j). If yes, it is 1, otherwise it is 0; t k represents the amount of electricity required for vehicles between OD pair k to complete their trip; r i Represents the electric charge at the starting point and the end point, i∈U k∈K {O k ,D k };n i Indicates the number of charging piles established at node i; The solution module uses the simulated annealing algorithm to solve the site selection model under given constraints to determine the node locations of the charging stations to be established and the number of charging piles installed at each charging station.

6. A charging station planning device, characterized in that: The device comprises: one or more processors; Memory; and One or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the charging station planning method considering the charging station capacity under the range uncertainty as described in any one of claims 1-4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a charging station planning method considering charging station capacity under range uncertainty as described in any one of claims 1 to 4 are implemented.

Images