A Method for Optimizing the Location and Capacity of Charging Parking Lots Based on Charging Demand Prediction

Abstract

The present invention discloses an optimization method for site selection and capacity determination of charging parking lots based on charging demand prediction, including: identifying the travel process and charging process according to the changes in the state of electric vehicles; determining the proportion of user travel and the probability transition matrix at each road node according to the land use type and user travel conditions in the research area; constructing a fuzzy logic inference system to simulate user charging decisions and determine the charging probability of users at each road node; determining the spatio-temporal distribution of charging demand at each road node; constructing a model for site selection and capacity determination of charging parking lots; using a genetic algorithm to solve the model for site selection and capacity determination of parking lots, and constructing an algorithm fitness function by using a fuzzy optimization method. The present invention simulates the distribution of user charging demands with actual user travel data, studies public parking lots as alternative addresses for charging facilities, provides technical support for the transformation of urban public parking lots by predicting the charging demands of electric vehicles, accelerates the layout of charging infrastructure, and ensures green travel.

Description

Technical Field

[0001] The present invention relates to the field of traffic planning, and particularly to an optimization method for site selection and capacity determination of a charging parking lot based on charging demand prediction. Background Art

[0002] In recent years, the ownership of new energy vehicles has been increasing, with a rapid growth in quantity and broad development prospects. However, there is still a large gap in the construction of charging facilities compared to the growth of the ownership of electric vehicles overall, making it difficult to meet the increasingly expanding charging demand. There are generally problems such as supply-demand mismatch in local areas, which restricts the further development and popularization of electric vehicles. Therefore, accurately predicting the charging demand of electric vehicle users, reasonably planning charging facilities, and achieving supply-demand balance have important theoretical and practical significance.

[0003] The spatio-temporal distribution of the charging demand of electric vehicles is affected by the subjective behavior of user groups and the objective conditions of urban dynamic and static traffic, and has characteristics such as strong randomness and uneven distribution. In addition, the planning of charging facilities is the key. Reasonable facility planning can achieve supply-demand balance as much as possible and reduce the engineering construction cost. Up to now, in the prediction of the charging demand of electric vehicles, most studies assume that the travel characteristics of fuel vehicles and electric vehicles are the same, and the modeling of their electricity consumption behavior lacks actual data support. In the planning of electric vehicle charging facilities, most studies focus on solving problems such as the division of the service range and site selection and capacity determination of newly built charging stations, lacking consideration of establishing a planning model by combining existing parking facilities and designing an effective solution algorithm for this problem. Summary of the Invention

[0004] Object of the Invention: Aiming at the above problems, the object of the present invention is to provide an optimization method for site selection and capacity determination of a charging parking lot based on charging demand prediction. Based on the actual operation data of electric vehicles, determine the spatio-temporal distribution of the charging demand of electric vehicles in the research area, establish a planning model for site selection and capacity determination of a charging parking lot based on fuzzy bi-objective optimization, and accelerate the effective layout of site selection and capacity determination of a charging parking lot.

[0005] Technical Solution: An optimization method for site selection and capacity determination of a charging parking lot based on charging demand prediction according to the present invention, the method comprising the following steps:

[0006] Step 1, identify the travel process and charging process according to the change of the state of the electric vehicle, and extract the data recorded by the electric vehicle during the travel process and charging process;

[0007] Step 2, determine the proportion of user travel at each road node and the probability transition matrix according to the land use type and user travel situation in the research area;

[0008] Step 3: Determine the charging demand of the electric vehicle based on whether the remaining power can reach the next road node during the travel of the electric vehicle. For the flexible charging demand, taking the maximum chargeable amount of the electric vehicle and the charging price in the current period as inputs, construct a fuzzy logic inference system to simulate the user's charging decision-making and determine the charging probability of the user at each road node; the charging demand includes flexible charging demand and rigid charging demand;

[0009] Step 4: Based on the Monte Carlo method, simulate the travel process and charging process of the user group, determine whether the user generates a charging demand during the travel process according to the charging decision-making, and determine the spatio-temporal distribution of the charging demand at each road node;

[0010] Step 5: Establish an objective function with the lowest user time cost and the highest parking lot benefit as the goals, and construct a charging parking lot location and capacity determination model;

[0011] Step 6: According to the spatio-temporal distribution of the charging demand, use the genetic algorithm to solve the parking lot location and capacity determination model, and construct the fitness function of the genetic algorithm by using the fuzzy bi-objective optimization method.

[0012] Further, the specific steps of Step 1 include:

[0013] Step 11: Identify the travel process of each vehicle respectively, traverse the vehicle operation record data in turn, and take the time when the vehicle status in the original data is normal operation, the start time changes, and the end time is null as the travel start time; take the time when the vehicle status is normal and the end time changes from null to a specific value as the travel end time;

[0014] Step 12: Identify the charging process of each vehicle respectively, traverse the vehicle record data in turn, take the time when the vehicle status changes from normal operation to charging status in the original data as the charging start time, and take the time when it changes from charging to normal operation status as the charging end time;

[0015] Step 13: Eliminate the processes corresponding to missing recording times and abnormal vehicle statuses during the travel process and charging process, and eliminate the travel processes in which the actual driving mileage corresponding to the power consumption during the travel process is greater than the theoretical driving mileage corresponding to the power consumption.

[0016] Further, the specific steps of Step 2 include:

[0017] According to the identified user travel process and charging process, fit the user's first travel start time, travel speed, parking duration, and first travel power;

[0018] According to the starting and ending points of each user's travel and the land use type, determine the proportion of user travel chain types, the proportion of travel from each residential area node, and the user travel probability transfer matrix between nodes of different land use types;

[0019] The travel chain types include two types: home - workplace - home, and home - workplace - business district - home; the probability transition matrix is the proportion of user transfers between various areas from the residential area to the workplace and from the workplace to the business district.

[0020] Further, step 3 specifically includes:

[0021] For each road node passed by the electric vehicle, it is judged whether the remaining power of the current vehicle can reach the next road node of the selected path. If not, it is a rigid charging demand, and the user must choose to charge. Then it is judged whether the parking duration can meet the user's charging demand. If it can meet, charging is carried out according to the charging duration. If it cannot meet, the parking duration is updated to the charging duration; where the destination is regarded as a road node.

[0022] If the current remaining power of the vehicle supports reaching the next road node, it is an elastic charging demand. A fuzzy logic inference system is constructed, and the fuzzy membership functions of the input and output variables are defined. With the maximum chargeable amount ΔSOC max and the charging price c(t) at the current time period as input variables, and the user's charging probability as the output variable, for the maximum chargeable amount ΔSOC max Three fuzzy sets of lower, medium, and higher are defined. c(t) includes two cases of off - peak electricity price and peak - time electricity price. Five fuzzy sets of low, relatively low, medium, relatively high, and high are defined for the charging probability. Inference rules are formulated for the output variables corresponding to the input variables under different sets; when the maximum chargeable amount and the charging price are determined, the membership degrees of each input variable of each inference rule are determined. The minimum value of the membership degrees of the input variables is taken as the output membership degree of each inference rule. Then the figures surrounded by the output membership degrees of each rule are superimposed to find the centroid, and the abscissa of the centroid is the user's charging probability.

[0023] Further, step 4 specifically includes:

[0024] For a certain user individual, the travel chain category is extracted based on the Monte Carlo method. The spatial positions of the user's residential area, workplace, and leisure area nodes are extracted according to the probability transition matrix. The starting time of the first trip, the travel speed, and the power of the first trip are extracted according to the user's trip parameters. The travel path is determined by using the analog multi - path traffic assignment method.

[0025] According to the selected travel path, when the user arrives at each road node, a charging decision is made based on the current remaining power, the parking duration, and the distance to the next road node, and the spatio - temporal distribution of the electric vehicle charging demand at each road node is obtained.

[0026] Further, step 5 specifically includes:

[0027] Obtain the number, location, and scale of public parking lots in the area to be studied. Take whether the public parking lot accepts reconstruction and the number of additional charging piles as decision variables. The constraint conditions are the number limit of charging parking lots and the limit of the number of charging berths in a single charging parking lot.

[0028] The site selection and capacity determination model of charging parking lots considers the optimization objectives from the user's perspective and the parking lot's perspective. Among them, the optimization objective from the user's perspective is to minimize the user's time cost, that is, the sum f1 of the time that the user group travels more to meet the charging demand and the queuing waiting time should be minimized. It is obtained by adding the time t for each user i to reach the parking lot i and the queuing time w for the user to wait for the charging service i and the expression is:

[0029]

[0030] The optimization objective from the parking lot's perspective is to maximize the difference f2 between the charging fee and the charging facility construction cost. It is obtained by subtracting the sum C of the single-day costs of building charging piles in each parking lot from the sum of the charging fees c that each user i needs to pay to meet the charging demand i and the expression is: j

[0031]

[0032] where j represents each public parking lot, j = 1, 2, ……, n; n represents the number of public parking lots.

[0033] Furthermore, the specific steps of step 6 include:

[0034] Adopt the fuzzy bi-objective optimization method, and establish a fitness function through the fuzzification method. The expression is:

[0035] F = min{μ1(F1), μ2(F2)}

[0036] where μ1(F1) and μ2(F2) are the membership degrees of the values F1 and F2 of the objective functions f1 and f2 respectively. When only solving the optimization objective from the user's perspective, the optimal solution corresponds to the value F of the objective f1 1m , and at this time the value of the objective f2 is F 2m ; when only solving the optimization objective from the target parking lot's perspective, the optimal solution corresponds to the value F of the objective f2 2M , and at this time the value of the objective f1 is F 1M , and the membership degree expressions are:

[0037]

[0038]

[0039] ​Randomly generate N feasible solutions as the initial parent population P, and perform crossover operation and mutation operation in sequence to obtain a new population P' to form a temporary parent population P temp =[P, P'], calculate the fitness using the fitness function, sort according to the fitness size, and take the top N individuals in P temp as the new population O to complete one iteration; repeat the genetic algorithm until the maximum number of iterations is reached and stop the iteration, and take the individual with the highest fitness in the whole iteration process as the optimal solution of the site selection and capacity determination model.

[0040] Beneficial effects: Compared with the prior art, the remarkable advantages of the present invention are:

[0041] 1. Based on the actual travel and charging behavior data of electric vehicles, the present invention conducts research on the site selection and capacity determination of charging parking lots, and balances the benefits of both from the perspectives of electric vehicle users and charging parking lots. For the electric vehicle user group, since the time for the driver to go to the charging parking lot and the waiting time for parking are considered in the site selection planning, the time loss caused by the electric vehicle users due to charging during the journey can be greatly shortened, and the problem of detouring for charging can be avoided as much as possible; for the charging parking lot, the optimization goal considers the charging revenue and the construction cost of charging facilities, which can ensure the benefits of the parking lot and avoid the situation that the parking lot has too large expenditure costs and cannot make up for the losses within a certain period and thus cannot operate.

[0042] 2. Considering the actual travel and charging behavior of electric vehicles, a more accurate prediction of the spatio-temporal distribution of charging demand can be obtained, so that the site selection planning of charging stations can better achieve the balance between charging supply and demand and improve the utilization efficiency of charging infrastructure;

[0043] 3. The method of adding charging piles based on public parking lots avoids the problems of long construction period and high land use cost of newly built charging parking lots. Description of the drawings

[0044] Figure 1 is the flow chart of the site selection and capacity determination planning of the charging parking lot;

[0045] Figure 2 is the flow chart of the charging decision-making process;

[0046] Figure 3 is the example diagram of solving the charging probability of users at each road node;

[0047] Figure 4 is the flow chart of the simulation calculation of the spatio-temporal distribution of charging demand. Detailed implementation manners

[0048] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments.

[0049] The present invention provides an optimization method for the location and capacity determination of a charging parking lot based on charging demand prediction. By analyzing the actual data recorded by electric vehicles, the charging and travel processes of users are identified, and the travel behavior characteristics of users and the operation characteristics of vehicles are extracted. According to the land use conditions in the research area and the travel behavior characteristics of users, the travel ratio of users at each road node is determined, as well as the probability transfer matrix of user travel between different road nodes. Based on factors such as the maximum chargeable amount and charging price, a fuzzy logic inference system is constructed to simulate the charging decisions of users. Combining with the path selection theory, the Monte Carlo method is used to simulate the travel and charging behaviors of the user group, and the spatio-temporal distribution of the charging demand of electric vehicles at each road node is determined. Taking the public parking lots in the research area as alternative sites for charging facilities, a location and capacity determination planning model for charging parking lots considering user costs and parking lot benefits is constructed. An adaptation function is established through a fuzzification method, and a genetic algorithm is designed to solve the problem.

[0050] As Figure 1 , an optimization method for the location and capacity determination of a charging parking lot based on charging demand prediction according to this embodiment includes the following steps:

[0051] Step 1: Identify the travel process and charging process according to the changes in the state of electric vehicles, and extract the data recorded by electric vehicles during the travel process and charging process:

[0052] Step 11: Identify the travel process of each vehicle separately. Traverse the vehicle operation record data in sequence. The time when the vehicle state is normal operation, the start time changes, and the end time is null in the original data is taken as the travel start time. The time when the vehicle state is normal and the end time changes from null to a specific value is taken as the travel end time. Table 1 lists the start-stop part of the record data in the original data of electric vehicles, and the data acquisition time interval is 30 - 60s.

[0053] Table 1 Record form of the start-stop part of the original data

[0054]

[0055]

[0056] In the column of "vehicle state", "1" represents normal operation, "3" represents the charging state, and "9" represents a malfunction.

[0057] Step 12: Identify the charging process of each vehicle separately. Traverse the vehicle operation record data in sequence. The time when the vehicle state changes from normal operation to the charging state in the original data is taken as the charging start time, and the time when the vehicle changes from the charging state to the normal operation state is taken as the charging end time. Table 2 lists the charging part of the record data in the original data of electric vehicles.

[0058] Table 2 Record form of the original data for the charging part

[0059]

[0060] In the column of "vehicle status", "1" indicates normal operation, "3" indicates charging status, and "9" indicates a fault.

[0061] Step 13: Eliminate the processes corresponding to missing recording times and abnormal vehicle statuses during the travel process and charging process, and eliminate the travel processes in which the actual driving mileage corresponding to the power consumption during the travel process is greater than the theoretical driving mileage corresponding to the power consumption.

[0062] Specifically, processes corresponding to a collection time interval exceeding 5 minutes, abnormal vehicles, and a vehicle status of "9" should all be eliminated; or eliminate the travel processes in which the actual driving mileage corresponding to the power consumption during the travel process is greater than the theoretical driving mileage corresponding to the power consumption. As shown in the following formula, if the power consumption (%) for a single trip exceeds 5 percentages of the ratio of the travel mileage ΔM to the cruising range M, it is regarded as abnormal power consumption:

[0063]

[0064] Step 2: Determine the proportion of user trips and the probability transition matrix for each road node according to the land use type in the research area and the user travel situation:

[0065] According to the identified user travel process and charging process, fit the start time of the user's first trip, travel speed, parking duration, and the power of the first trip;

[0066] According to the start and end points of each user trip and the land use type, determine the proportion of user trip chain types, the proportion of trips to each residential area node, and the user travel probability transition matrix between nodes of different land use types;

[0067] The travel chain types include two types: home-work area-home and home-work area-commercial area-home; the probability transition matrix is the proportion of user transfers between each area from the residential area to the work area and from the work area to the commercial area.

[0068] According to the map of urban land use types, it is possible to determine whether the land use type is residential, working, or commercial land. Then, if a road node is in a certain area or close to a certain area, it is regarded as belonging to the corresponding road node type. The land use type refers to determining the road node type of the user's travel according to the longitude and latitude of each user trip, then extracting the travel starting point type as the residential area road node type, and finally statistically obtaining the travel ratio of different residential areas. For example, Residential Area 1:Residential Area 2 = 0.4:0.6. When predicting the charging demand and allocating the travel flow of each place, the travel volume of different regions can be determined according to the total travel volume and the ratio.

[0069] Based on the urban land use type map and the starting and ending positions of the user's trips, the type of each user trip can be determined, such as residential area → work area, work area → leisure area. Each type has a probability transition matrix. The rows and columns of each probability transition matrix represent the proportion of trips between different land use areas, which is determined by the proportion of the number of each node to other nodes. Table 3 lists the user trip probability transition matrix from residential area nodes to work area nodes, where WID represents the work area node ID, HID represents the residential area node ID, and the value represents the probability from the corresponding residential area node → work area node.

[0070] Table 3 Example of User Trip Probability Transition Matrix from Residential Area Nodes to Work Area Nodes

[0071]

[0072]

[0073] Step 3: Judge the charging demand of the electric vehicle according to whether the remaining power of the electric vehicle can reach the next road node during the trip. For the flexible charging demand, taking the maximum chargeable amount of the electric vehicle and the charging price at the current time period as inputs, construct a fuzzy logic inference system to simulate the user's charging decision and determine the charging probability of the user at each road node; the charging demand includes flexible charging demand and rigid charging demand.

[0074] As Figure 2 shown, every time the electric vehicle passes through a road node, judge whether the remaining power of the current vehicle can reach the next road node of the selected path. If not, it is a rigid charging demand and the user must choose to charge. Judge whether the parking duration can meet the user's charging demand. If it can meet, charge according to the charging duration. If it cannot meet, update the parking duration to the charging duration; where the destination is regarded as a road node.

[0075] If the current remaining power of the vehicle supports reaching the next road node, it is a flexible charging demand. Construct a fuzzy logic inference system, define the fuzzy membership function between input and output variables, and use the maximum chargeable amount ΔSOC max and the charging price c(t) at the current time period as input variables, and the user's charging probability as the output variable. Define 3 fuzzy sets of lower, medium, and higher for the maximum chargeable amount ΔSOC max . c(t) includes the valley-time electricity price c vally and the peak-time electricity price c peakIn two cases, five fuzzy sets of low, relatively low, medium, relatively high, and high are defined for the charging probability. Inference rules are formulated for the input variables corresponding to the output variables under different sets, as shown in Table 4. There are a total of 6 rules. According to the inference rules, the user's charging probability is determined. Under the condition of determining the maximum chargeable amount and the charging price, the membership degrees of each input variable of each rule are determined. The minimum value of the membership degrees of the input variables is taken as the output membership degree of each rule. Then, the centroids of the graphs formed by the output membership degrees of each rule are superimposed to find the centroid. The abscissa of the centroid is the user's charging probability. As Figure 3 shown in the example, when the maximum chargeable amount ΔSOC max takes 58.5 and the charging price c(t) takes 0.125, the fuzzy membership degrees of the chargeable amount and the fuzzy membership degrees of the charging price of the six inference rules can be obtained. The minimum value of the two is taken as the fuzzy membership degree of the charging probability of each rule. According to the fuzzy membership degree function of the output variable, the fuzzy membership degree image of each rule can be determined. The abscissa of the centroid obtained by superimposing the images and solving is the charging probability 0.758 of the vehicle at the current road node.

[0076] Table 4 Fuzzy Inference Rules for Charging Decision

[0077]

[0078]

[0079] The maximum chargeable amount ΔSOC max is determined according to the parking duration and the current remaining power. The expression is as follows:

[0080] ΔSOC max = min(P * T / C * 100, 100 - SOC)

[0081] where P is the charging power (kw), T is the parking duration (h), C is the battery capacity (kwh), and SOC is the current state of charge.

[0082] Step 4: Based on the Monte Carlo method, simulate the travel process and charging process of the user group. According to the charging decision, judge whether there is a charging demand during the user's travel process, and determine the spatio-temporal distribution of the charging demand at each road node.

[0083] As Figure 4As shown in the figure, for a certain individual user, travel chain categories are extracted based on the Monte Carlo method. The spatial positions of nodes in the user's residential area, work area, and leisure area are extracted according to the probability transition matrix. The starting time of the first trip, the travel speed, and the battery power of the first trip are extracted according to the user's trip parameters. The travel path is determined by using the analog multi-path traffic assignment method. According to the selected travel path, when the user arrives at each road node, a charging decision is made based on the current remaining battery power, the parking duration, and the distance to the next road node, and the spatio-temporal distribution of the electric vehicle charging demand at each road node is obtained.

[0084] The analog multi-path traffic assignment method determines the travel path. It is assumed that the user's path selection behavior is random, and its probability is related to the lengths of each effective path that meets the user's travel purpose. An effective path is a path connected by effective road segments between the user's departure place and the destination. When the user travels along the effective path, they will definitely get closer to the destination. The probability P(k) that each effective path k is selected by the user is calculated using the Logit-type discrete choice model, and the expression is:

[0085]

[0086] where σ is a dimensionless assignment parameter, l(i) is the length of the i-th effective travel path, is the average length of m effective paths.

[0087] Step 5: Establish an objective function with the goal of minimizing the user's time cost and maximizing the benefit of the parking lot, and construct a charging parking lot site selection and capacity determination model:

[0088] Obtain the number, location, and scale of public parking lots in the area to be studied. With whether the public parking lot accepts reconstruction and the number of additional charging piles as decision variables, the constraint conditions are the limit on the number of charging parking lots and the limit on the number of charging berths in a single charging parking lot;

[0089] The charging parking lot site selection and capacity determination model considers the optimization objectives from the user's perspective and the parking lot's perspective. Among them, the optimization objective from the user's perspective refers to minimizing the user's time cost, that is, the sum f1 of the additional travel time and the queuing waiting time for the user group to meet the charging demand should be minimized. It is obtained by adding the time t i when each user i arrives at the parking lot and the queuing time w i for the user to wait for the charging service. The time to arrive at the parking lot depends on the distance d i from user i to the parking lot and the current speed v i , and the expression is:

[0090]

[0091] where the user's queuing time w iReferring to the calculation idea of vehicle queuing delay, the time distribution curve of the number of vehicles with charging demand is denoted as Let the number of charging berths in the parking lot be C. When the demand number at time t a to t b During the time period exceeds C, the area where the curve exceeds the horizontal line C can approximately represent the charging delay of users during this period, denoted as D ab , and its calculation method is as follows:

[0092]

[0093] The optimization goal from the perspective of the parking lot is to maximize the difference f2 between the charging fee and the construction cost of charging facilities. The charging fee c that each user i who meets the charging demand needs to pay i The sum of and the single-day cost C of building charging piles in each parking lot j The difference obtained by taking the sum is expressed as:

[0094]

[0095] Among them, j represents each public parking lot, j = 1, 2,..., n; n represents the total number of public parking lots. To convert the construction cost into a single-day cost, it is necessary to determine the number of charging piles N added in each parking lot j , the service life T of the charging pile, and the unit price P of the charging pile.

[0096] Step 6: According to the spatio-temporal distribution of charging demand, use the genetic algorithm to solve the parking lot location and capacity determination model, and adopt the fuzzy double-objective optimization method to construct the fitness function of the genetic algorithm.

[0097] Adopt the fuzzy double-objective optimization method, establish the fitness function through the fuzzy method, and the expression is:

[0098] F = min{μ1(F1), μ2(F2)}

[0099] Among them, μ1(F1) and μ2(F2) are the membership degrees of the values F1 and F2 of the objective functions f1 and f2 respectively. When solving only from the optimization goal of the user perspective, the optimal solution corresponds to the value of the objective f1 as F 1m , and at this time the value of the objective f2 is F 2m ; when solving only from the optimization goal of the target parking lot perspective, the optimal solution corresponds to the value of the objective f2 as F 2M , and at this time the value of the objective f1 is F 1M , and the membership degree expressions are respectively:

[0100]

[0101]

[0102] Randomly generate N feasible solutions as the initial parent population P, and perform crossover operation and mutation operation in sequence to obtain a new population P' to form a temporary parent population P temp =[P, P'], calculate the fitness using the fitness function, sort according to the fitness size, and select the top N individuals in P temp as the new population O to complete one iteration; repeat the genetic algorithm until the maximum number of iterations is reached and stop the iteration, and select the individual with the highest fitness in the entire iteration process as the optimal solution of the site selection and capacity determination model.

[0103] Crossover operation: The solution X is a 2×n vector. The first row is the site selection solution, and the 0 and 1 values represent not setting and setting charging piles respectively. The second row is the capacity determination solution. For two paired solutions X1 and X2. First, slide and select two columns of decision variables, and swap the decision variables according to the probability P c1 Secondly, select a column of capacity determination where the site selection decision variables are all 1, and perform linear crossover of real values according to the probability P c2

[0104] Mutation operation: According to the probability P m1 select a site selection decision variable to invert it and change its associated capacity determination. According to the probability P m2 select a capacity determination with a site selection decision variable of 1 to float up and down by a certain range.​

Claims

1. A method for optimizing the location and capacity determination of a charging parking lot based on charging demand prediction, characterized in that, The method includes the following steps: Step 1, identify the travel process and charging process according to the changes in the state of the electric vehicle, and extract the data recorded by the electric vehicle during the travel process and charging process; Step 2, determine the proportion of user travel and the probability transition matrix of each road node according to the land use type of the research area and the user travel situation; Step 3, judge the charging demand of the electric vehicle according to whether the remaining power of the electric vehicle can reach the next road node during the travel process. For the flexible charging demand, taking the maximum chargeable amount of the electric vehicle and the charging price at the current time period as inputs, construct a fuzzy logic inference system to simulate the user's charging decision, and determine the charging probability of the user at each road node; the charging demand includes flexible charging demand and rigid charging demand; Step 4, simulate the travel process and charging process of the user group based on the Monte Carlo method, judge whether the user's travel process generates a charging demand according to the charging decision, and determine the spatio-temporal distribution of the charging demand at each road node; Step 5, establish an objective function with the lowest user time cost and the highest parking lot benefit as the goals, and construct a charging parking lot location and capacity determination model; Step 6, according to the spatio-temporal distribution of the charging demand, use the genetic algorithm to solve the parking lot location and capacity determination model, and construct the fitness function of the genetic algorithm by using the fuzzy multi-objective optimization method; The specific content of Step 3 includes: When the electric vehicle passes through each road node, judge whether the remaining power of the current vehicle can reach the next road node of the selected path. If not, it is a rigid charging demand, and the user must choose to charge. Judge whether the parking duration can meet the user's charging demand. If it can meet the demand, charge according to the charging duration. If it cannot meet the demand, update the parking duration to the charging duration; where the destination is regarded as a road node; If the current remaining power of the vehicle supports reaching the next road node, it is an elastic charging demand. A fuzzy logic inference system is constructed to define the fuzzy membership functions of the input and output variables, with the maximum chargeable amount ΔSOC max and the charging price c(t) at the current time period as input variables, and the user's charging probability as the output variable. For the maximum chargeable amount ΔSOC max Three fuzzy sets of low, medium, and high are defined. c(t) includes two cases: off-peak electricity price and peak electricity price. Five fuzzy sets of low, relatively low, medium, relatively high, and high are defined for the charging probability. Inference rules are formulated for the output variables corresponding to the input variables under different sets; under the condition of determining the maximum chargeable amount and the charging price, determine the membership degree of each input variable for each inference rule, take the minimum value of the input variable membership degree as the output membership degree of each inference rule, and then superimpose the graphs surrounded by the output membership degrees of each rule to find the centroid. The abscissa of the centroid is the user's charging probability; The specific content of Step 4 includes: For a certain user individual, extract the travel chain category based on the Monte Carlo method, extract the spatial positions of the user's residential area, workplace and leisure area nodes according to the probability transition matrix, extract the start time of the first trip, travel speed and the power of the first trip according to the user's trip parameters, and determine the travel path by using the analogy multi-path traffic assignment method; According to the selected travel path, when the user arrives at each road node, make a charging decision according to the current remaining power, parking duration and the distance to the next road node, and obtain the spatio-temporal distribution of the charging demand of the electric vehicle at each road node.

2. The method for optimizing the location and capacity determination of a charging parking lot according to claim 1, characterized in that, The specific content of Step 1 includes: Step 11, identify the travel process of each vehicle respectively, traverse the vehicle operation record data in turn, take the time when the vehicle state in the original data is normal operation, the start time changes and the flameout time is null as the start time of the trip, and take the time when the vehicle state is normal and the flameout time changes from null to a specific value as the end time of the trip; Step 12, identify the charging process of each vehicle respectively, traverse the vehicle operation record data in turn, take the time when the vehicle state changes from normal operation to charging state in the original data as the start time of charging, and take the time when it changes from charging to normal operation state as the end time of charging; Step 13: Eliminate the processes with missing recorded times and abnormal vehicle states during the travel and charging processes, and eliminate the travel processes in which the actual driving mileage corresponding to the power consumption during the travel is greater than the theoretical driving mileage corresponding to the power consumption.

3. The method for optimizing the location and capacity determination of a charging parking lot according to claim 2, characterized in that, The specific steps of Step 2 include: According to the identified user travel and charging processes, fit the start time of the user's first trip, travel speed, parking duration, and the power of the first trip. Based on the starting and ending points of each user trip and the land use type, determine the proportion of user travel chain types, the proportion of trips from each residential area node, and the user travel probability transition matrix between different land use type nodes. The travel chain types include two types: home-work area-home, and home-work area-commercial area-home; the probability transition matrix is the proportion of user transfers between each area from the residential area to the work area and from the work area to the commercial area.

4. The method for optimizing the location and capacity determination of a charging parking lot according to claim 1, characterized in that, The specific steps of Step 5 include: Obtain the number, location, and scale of public parking lots in the area to be studied. Taking whether the public parking lot accepts reconstruction and the number of additional charging piles as decision variables, the constraint conditions are the number limit of charging parking lots and the limit of the number of charging berths in a single charging parking lot. The charging parking lot site selection and capacity determination model considers the optimization objectives from the user's perspective and the parking lot's perspective. Among them, the optimization objective from the user's perspective refers to minimizing the user's time cost, that is, the sum f1 of the additional travel time and the queuing waiting time for the user group to meet the charging demand should be minimized. It is obtained by summing the arrival time t of each user i at the parking lot i and the queuing time w for the user to wait for the charging service i and is expressed as: The optimization goal of the parking lot angle is to maximize the difference f2 between the charging fee and the charging facility construction cost, which is obtained by subtracting the sum of the charging costs c that each user i who meets the charging demand needs to pay from the sum of the single-day costs C of building charging piles in each parking lot. The expression is as follows: i The sum of j The difference between the sum and the single-day cost C of building charging piles in each parking lot. The expression is as follows: Among them, j represents each public parking lot, j = 1, 2, ……, n; n represents the number of public parking lots.

5. The method for optimizing the location and capacity determination of a charging parking lot according to claim 4, characterized in that, The specific steps of Step 6 include: Adopt a fuzzy bi-objective optimization method, and establish a fitness function through a fuzzification method. The expression is: F = min{μ1(F1), μ2(F2)} Among them, μ1(F1) and μ2(F2) are the membership degrees of the values F1 and F2 of the objective functions f1 and f2 respectively. When optimizing the objective solution only from the user's perspective, the value of the objective f1 corresponding to the optimal solution is F 1m , and at this time the value of the objective f2 is F 2m ; when optimizing the objective solution only from the perspective of the target parking lot, the value of the objective f2 corresponding to the optimal solution is F 2M , and at this time the value of the objective f1 is F 1M , and the membership degree expressions are respectively: Randomly generate N feasible solutions as the initial parent population P, and perform crossover operation and mutation operation in sequence to obtain a new population P' to form a temporary parent population P temp =[P, P'], calculate the fitness using the fitness function, sort according to the fitness size, and take P temp The top N individuals in are used as the new population O to complete one iteration; repeat the genetic algorithm until the maximum number of iterations is reached and then stop the iteration, and take the individual with the highest fitness in the entire iteration process as the optimal solution of the site selection and capacity determination model.

Images