A Method for Selecting and Planning Electric Vehicle Charging Facilities Based on Multi-Source Data

Abstract

This invention discloses a method for selecting and planning electric vehicle charging facilities based on multi-source data. First, a user optimal choice model is constructed. Based on this model, a set of optimal user choices is obtained, which is then used as initial parameters to solve the electric vehicle charging facility location model. This set includes a set of optimal search results and a set of driver optimal addresses. Next, an electric vehicle charging facility location model is constructed. Based on the set of optimal search results and the driver optimal address set, the model is solved, leading to a charging facility planning scheme. This scheme includes adjustments to the number of charging piles in current charging facilities and future charging facility location planning. This invention can provide a scientific number of charging piles for candidate sites without existing charging facilities and solve the problem of resource waste caused by excess charging piles for candidate power stations with existing charging facilities.

Description

Technical Field

[0001] This invention relates to a method for selecting and planning electric vehicle charging facilities. Its main function is to recommend the optimal charging facilities when electric vehicles have charging needs by acquiring and integrating various data, and to provide the best solution for urban charging facility planning by combining multi-vehicle charging data within a region. Background Technology

[0002] The development of gasoline-powered vehicles has spanned over a century. During this time, the automotive industry has faced severe challenges due to development, increased vehicle ownership, depletion of oil resources, and environmental pollution. Following energy development trends and strategies, the research and development and use of energy-saving and emission-reducing new energy vehicles have become the only way to solve energy and environmental problems. Currently, more than a dozen countries worldwide have explicitly stated or intend to announce a ban on the sale and manufacture of traditional gasoline-powered vehicles in the coming decades. Therefore, gasoline-powered vehicles, after more than a century of service, may become history, replaced by entirely new energy vehicles. With the rapid growth in the number of electric vehicles, my country's charging infrastructure is far from meeting the increasing demand. Therefore, building a reasonable, efficient, and convenient charging infrastructure is an urgent task.

[0003] Currently, most electric vehicle charging station planning methods aim to ultimately construct charging stations. However, China has a huge population, dense urban roads, and extremely valuable urban land, making the cost of building charging stations, especially near city centers, prohibitively high. Furthermore, most existing planning methods, after determining the charging station site, do not determine the optimal number of charging piles to be built within the station. These issues affect the rationality of charging facility site selection and reduce electric vehicle user satisfaction. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application proposes a method for selecting and planning electric vehicle charging facilities based on multi-source data. The method obtains a set of optimal user choices based on different user preferences, serving as the initial parameters for the electric vehicle charging facility site selection model. The objective function of the charging facility site selection model simultaneously considers construction costs and user satisfaction. By solving the model, the optimal site selection scheme and its corresponding number of charging piles are obtained simultaneously.

[0005] The technical solution adopted in this invention is as follows:

[0006] The method for selecting and planning electric vehicle charging facilities based on multi-source data includes the following steps:

[0007] S1. Construct a user's optimal choice model based on travel time, road congestion, queuing time, and electricity price; represented as:

[0008] min W=θ1W1+θ2W2+θ3W3+θ4W4

[0009] In the formula: W1 is the degree of road congestion; W2 is the travel time; W3 is the queuing time; W4 is the electricity price; θ1, θ2, θ3, and θ4 are the weights corresponding to W1-W4, and θ1+θ2+θ3+θ4=1;

[0010] S2. Based on the user optimal choice model, obtain the set of user optimal choices, and use this set of user optimal choices as the initial parameters for solving the electric vehicle charging facility location model. The set of user optimal choices includes the set of optimal search results and the set of driver optimal addresses.

[0011] S3. The site selection model for electric vehicle charging facilities is represented as follows:

[0012] min F=ω1F1+ω2F2+ω3F3+F4

[0013] In the formula: F represents the total operating cost of the charging facility to the target lifespan; F1 represents the construction cost of the charging facility to the target lifespan; F2 represents the distance cost for users to travel to the charging facility; F3 represents the queuing cost for users; F4 represents the penalty function; ω1, ω2, and ω3 are the weights corresponding to F1, F2, and F3, respectively.

[0014] S4. Based on the optimal search result set and the driver's optimal address set, solve the electric vehicle charging facility location model, and then obtain the electric vehicle charging facility planning scheme; the electric vehicle charging facility planning scheme includes the adjustment scheme of the number of charging piles in the current charging facility and the future charging facility location planning scheme.

[0015] Furthermore, the method for obtaining the optimal set of search results in S2 is as follows:

[0016] S2.1.1. Based on the selection habits of car users, assign values ​​to θ1, θ2, θ3, and θ4 in the user's best selection model. The four weight values ​​cannot all be the same, and the priority division must be reflected.

[0017] S2.1.2. Select all parking lots with charging facilities in the area where the demand point is located as candidate power stations;

[0018] S2.1.3 Calculate the objective function value minW for the i-th demand point and all candidate power plants using the user optimal selection model after weight assignment; denote the candidate power plant corresponding to the objective function value minW of the i-th demand point as F. (minW)i ;

[0019] S2.1.4. Select the two largest weights from the four weights, assign a value of 1 to one of the selected larger weights, and assign a value of 0 to the other three weights. Then substitute the reassigned weights into the user's optimal selection model; at this time, the user's optimal selection model outputs the candidate power station corresponding to the i-th demand point, denoted as F. (first)i ;

[0020] Assign a value of 1 to one of the selected larger weights, and assign 0 to the other three weights. Then substitute the reassigned weights into the user's optimal choice model. At this point, the user's optimal choice model outputs the candidate power station corresponding to the i-th demand point, denoted as F. (second)i ;

[0021] S2.1.5, the three candidate power plants obtained from S2.1.3 and S2.1.4 constitute the set of optimal search results corresponding to the i-th demand point, denoted as F. xyi =[F (minw)i F (first)i F (second)i ];

[0022] S2.1.6 Repeat the above steps to obtain the set of best optimization results corresponding to all requirement points.

[0023] Furthermore, the method for obtaining the optimal addressing set for drivers in S2 is as follows:

[0024] S2.2.1. Based on the selection habits of car users, assign values ​​to θ1, θ2, θ3, and θ4 in the user's best selection model. The four weight values ​​cannot all be the same, and the priority division must be reflected.

[0025] S2.2.2. All parking lots in the area where the demand point is located are considered as candidate power stations. The parking lots include those that have been equipped with charging facilities and those that have not been equipped with charging facilities.

[0026] S2.2.3. Using the weighted user optimal choice model, calculate the objective function value minW for the i-th demand point and all candidate power plants; denot the candidate power plant corresponding to the objective function value minW of the i-th demand point as F. (minW)i ';

[0027] S2.2.4. Select the two largest weights from the four weights; assign a value of 1 to one of the selected larger weights, and assign a value of 0 to the other three weights. Then substitute the reassigned weights into the user's optimal selection model. At this time, the user's optimal selection model outputs the candidate power station corresponding to the i-th demand point, denoted as F. (first)i ';

[0028] Assign a value of 1 to one of the selected larger weights and 0 to the other three weights. Then substitute the reassigned weights into the user's optimal choice model. Let F be the candidate power station corresponding to the i-th demand point output by the user's optimal choice model at this point. (second)i ';;

[0029] S2.2.5, the three candidate power stations obtained from S2.2.3 and S2.2.4 constitute the optimal addressing set for the driver corresponding to the i-th demand point, denoted as F. xzi =[F (minw)i’ F (first)i’ F (second)i’ ];

[0030] S2.2.6 Repeat the above steps to obtain the optimal addressing set for all demand points.

[0031] Furthermore, in S4, the method for obtaining the adjustment scheme for the number of charging piles in the current charging facility based on the best optimization result set is as follows:

[0032] S4.1.1 Generate an N*E initial matrix N0, which consists of E column vectors, each representing a scheme, for a total of E schemes;

[0033] A candidate power station is randomly selected from the set of best optimization results corresponding to N demand points in turn, thus forming an initial vector of length N.

[0034] S4.1.2, count the candidate power stations appearing in each column vector, and record the number of charging piles of different charging types in each candidate power station; let b be the number of fast charging piles in candidate power station j in the column vector. jkc The number of slow charging stations is b jmc Summarize the number of fast charging piles and slow charging piles for the same candidate power station j in each of the E column vectors, and thus obtain the range of the number of fast charging piles for candidate power station j, denoted as b. jkc ∈[B jkc1 B jkc2 ], B jkc1 B is the minimum number of fast charging piles for candidate power station j. jkc2 This represents the maximum number of fast charging piles for candidate power station j; similarly, the number of slow charging piles b for candidate power station j can be obtained. jmc ∈[B jmc1 B jmc2 ], B jmc1 B is the minimum number of slow charging piles for candidate power station j. jmc2 It is the maximum number of slow charging piles in candidate power station j;

[0035] S4.1.3, using the electric vehicle charging facility site selection model, calculate the objective function value minF of the E column vectors in the initial matrix N0, and the number b of fast charging piles of candidate power station j at each iteration. jkc and the number of slow charging stations b jmc Each is randomly selected from its corresponding range;

[0036] S4.1.4, Update the memory matrix

[0037] (1) Calculate the concentration and probability of selection for each option:

[0038]

[0039]

[0040] In the formula: A represents the reciprocal of minF, V ij V represents the similarity between column vectors; ij n represents the number of identical bits in vectors i and j; n represents the length of the vector.

[0041]

[0042] In the formula: N is the number of antibodies, T is the set threshold;

[0043]

[0044] In the formula: P i α is the selection probability; α is the proportionality coefficient; when the individual fitness is higher, the expected reproduction probability is greater; when the individual concentration is lower, the expected reproduction probability is greater. This method is beneficial for selecting individuals with high fitness, while inhibiting individuals with high concentration, thus ensuring individual diversity.

[0045] (2) Set the N*U memory matrix N U Sort the objective function values ​​of all schemes in ascending order, and put the column vectors corresponding to the first S smaller objective function values ​​into the memory matrix; then sort the data concentrations of the remaining ES column vectors in ascending order; and put the column vectors corresponding to the first US smaller concentrations into the memory matrix.

[0046] (3) Select, transform, and crossover the remaining EU individuals (i.e., column vectors) to generate matrix N. E-U :

[0047] Selection: Chromosomes with better fitness are selected using a roulette wheel selection method. The individual selection probability is the selection probability P obtained in (1). i ;

[0048] Crossover: Crossover is performed by randomly selecting crossover positions; each crossover is performed by randomly selecting two vectors, and two numbers within (0, N) are randomly selected as the crossover segment, and the numbers in the two vectors within this segment are swapped;

[0049] Mutation: Mutation is performed by randomly selecting mutation positions; each mutation randomly selects three numbers in (0, N) as mutation positions, and obtains the optimal addressing set for each demand point according to the above steps, and randomly replaces the number at the mutation position with one of the remaining two numbers in the set;

[0050] (4) Determine if the termination condition is met; if so, end the optimization. At this time, the column vector corresponding to the smallest objective function value in the matrix, i.e., the number of fast charging piles of candidate power station j in the scheme, is b. jkc The number of slow charging stations is b jmc A plan to adjust the number of charging piles in the current charging facilities;

[0051] Conversely, proceed to the next step; preset the number of iterations, and end the iteration when the preset number of iterations is reached;

[0052] (5) N U With N E-U Together, they form a new generation matrix N. n ;

[0053] (6) New generation matrix N n Recalculate the objective function value.

[0054] Furthermore, based on the optimal addressing set of drivers, the method for obtaining the future charging facility site selection planning scheme is as follows:

[0055] S4.2.1 Generate an N*E initial matrix N0, which consists of E column vectors, each representing a scheme, for a total of E schemes;

[0056] A candidate power station is randomly selected from the set of best optimization results corresponding to N demand points in turn, thus forming an initial vector of length N.

[0057] S4.2.2, count the candidate power stations appearing in each initial vector, and record the number of charging piles of different charging types in each candidate power station; let b be the number of fast charging piles in candidate power station j in the column vector. jkc The number of slow charging stations is b jmc Summarize the number of fast charging piles and slow charging piles of the same candidate power station j in the E initial vectors, and thus obtain the range of the number of fast charging piles of candidate power station j, denoted as b. jkc ∈[B jkc1 B jkc2 ], B jkc1B is the minimum number of fast charging piles for candidate power station j. jkc2 This represents the maximum number of fast charging piles for candidate power station j; similarly, the number of slow charging piles b for candidate power station j can be obtained. jmc ∈[B jmc1 B jmc2 ], B jmc1 B is the minimum number of slow charging piles for candidate power station j. jmc2 It is the maximum number of slow charging piles in candidate power station j;

[0058] S4.2.3, using the electric vehicle charging facility site selection model, calculate the objective function value minF of the E column vectors in the initial matrix N0, and the number b of fast charging piles of candidate power station j at each iteration. jkc and the number of slow charging stations b jmc Each is randomly selected from its corresponding range;

[0059] S4.2.4, Update the memory matrix

[0060] (1) Calculate the column vector concentration and selection probability of each scheme:

[0061]

[0062]

[0063] In the formula: A represents the reciprocal of minF, V ij V represents the similarity between column vectors; ij n represents the number of identical bits in vectors i and j; n represents the length of the vector.

[0064]

[0065] In the formula: N is the number of antibodies, T is the set threshold.

[0066]

[0067] In the formula: P i α is the selection probability; α is the proportionality coefficient. When the individual fitness is higher, the expected reproduction probability is greater; when the individual concentration is lower, the expected reproduction probability is greater. This method is beneficial for selecting individuals with high fitness, while inhibiting individuals with high concentration, thus ensuring individual diversity.

[0068] (2) Set the N*U memory matrix N U Sort the objective function values ​​of all schemes in ascending order, and put the column vectors corresponding to the first S smaller objective function values ​​into the memory matrix; then sort the data concentrations of the remaining ES column vectors in ascending order; and put the column vectors corresponding to the first US smaller concentrations into the memory matrix.

[0069] (3) Select, transform, and crossover the remaining EU individuals (i.e., column vectors) to generate matrix N. E-U :

[0070] Selection: Chromosomes with better fitness are selected using a roulette wheel selection method. The individual selection probability is the selection probability P obtained in (1). i .

[0071] Crossover: Crossover is performed by randomly selecting crossover positions. Each crossover involves randomly selecting two vectors and randomly selecting two numbers within (0, N) as the crossover segment. The numbers in the two vectors within this segment are then swapped.

[0072] Mutation: Mutation is performed by randomly selecting mutation positions. Each mutation randomly selects three numbers within (0, N) as mutation positions. Based on the above steps, the optimal addressing set for each demand point is obtained, and the number at the mutation position is randomly replaced with one of the remaining two numbers in the set.

[0073] (4) Determine if the termination condition is met; if yes, terminate; at this point, for parking lots already equipped with charging facilities, the column vector corresponding to the minimum objective function value in the matrix, i.e., the number of fast charging piles for candidate power station j in the scheme, is b. jkc The number of slow charging stations is b jmc This is a scheme for adjusting the number of charging piles in the current charging facilities; for parking lots without charging facilities, the column vector corresponding to the minimum objective function value in the matrix is ​​the output candidate power station j in the scheme, which is the site selection scheme.

[0074] Conversely, proceed to the next step; preset the number of iterations, and end the iteration when the preset number of iterations is reached;

[0075] (5) N U With N E-U Together, they form a new generation matrix N. n ;

[0076] (6) New generation matrix N n Recalculate the objective function value.

[0077] Furthermore, the values ​​of the weights θ1-θ4 in the user's optimal choice model are determined by the driver based on their own selection habits. If the driver prefers to choose a road with less congestion, then θ1 is increased, and the other three weights are adjusted accordingly.

[0078] Furthermore, the terms of the user's optimal choice model are expressed as follows:

[0079] 1) The degree of road congestion, W1, is represented as:

[0080]

[0081] In the formula: the location where the driver has a charging need is recorded as the demand point, and all parking lots within the area of ​​the demand point are recorded as candidate charging stations; therefore, L represents the road segment the driver takes from the demand point to the candidate charging station selected by the driver; L h This represents the set of all routes from the demand point to the candidate power station; σ L The degree of congestion of road segment L when a driver decides to go to the selected candidate power station is σ, which is based on the road conditions shown on the online map. The degree of congestion is represented by the values ​​1, 2, and 3, which indicate that the road segment is in a smooth state, relatively congested, and extremely congested, respectively.

[0082] 2) Travel time

[0083]

[0084] In the formula: W2 represents the driving time from the demand point to the selected charging facility; t L 0 This represents the free-roaming time of road segment L when traffic flow is zero, which is the segment length divided by the speed limit of the segment; x L c represents the traffic flow rate of road segment L; L For t L Take 1.15t L 0 Traffic volume at that time;

[0085] 3) Queuing time

[0086]

[0087] In the formula: n is the nth position of driver i in the charging queue when driver i arrives at the charging station; Z i n-1 The battery capacity of the vehicle preceding driver i in the charging queue, Q i n-1 P represents the battery level of driver i's vehicle when it arrives at the charging station. i W3 represents the power consumption per unit time of driver i's vehicle when it is not in motion. n-1 Let T be the queuing time for driver i's vehicle. When n = 1, meaning there are no vehicles ahead of driver i in the queue, the driver only needs to wait for the first vehicle to finish charging at the charging station. q The remaining charging time is the time until the first vehicle in the charging station to finish charging when driver i arrives at the charging station. The charging power of the vehicle preceding driver i when driver i is queuing at charging station j is determined by k different charging types, which can be divided into fast charging and slow charging.

[0088] Furthermore, the optimal choice model for users is based on the road congestion level and traffic flow on each road at the moment the driver generates a charging demand. Based on this, four options are obtained: travel time, road congestion level, queuing time, and electricity price. It is also set that once the driver decides to go to the selected charging facility, the choice will not be changed midway.

[0089] Furthermore, the electric vehicle charging facility site selection model consists of the construction cost of the charging facility operating for the target lifespan, the distance cost for users to travel to the charging facility, the queuing cost for users, and a penalty function, as follows:

[0090] 1) Construction cost of charging facilities until the target lifespan

[0091]

[0092] Where: r is the depreciation rate; m is the planned life of the charging facilities; S j The area occupied by the fast and slow traffic cones; G j S represents the unit land price for candidate power plants; jk G represents the number of fast-charging piles at candidate power stations; k Price per unit for fast charging stations; S jm G represents the number of slow-charging piles at candidate power stations; m The unit price for slow charging stations; F j For candidate power plant infrastructure costs and other expenses; h j M is the total number of candidate power plants, which is the decision variable.

[0093] 2) Cost of travel distance to charging facilities for users

[0094]

[0095] In the formula: L ij Q represents the distance from demand point i to candidate power station j. i Electricity consumption per kilometer for user i electric vehicle; D j Let x be the unit electricity price of candidate power plant j; ij Let x be the decision variable for distance cost. ij A value of 1 indicates that user i chooses to go to charging station j, and x ij A value of 0 indicates that user i did not choose to go to charging station j; N is the set of demand points;

[0096] 3) User queuing costs

[0097]

[0098] In the formula: T ij User queuing time; P i This refers to the power consumption per unit time for user i when not in driving mode; Gi X represents the average hourly earnings of the surveyed users. ij X is the decision variable for queuing cost. ij A value of 1 indicates that user i needs to queue at charging station j. ij A value of 0 indicates that user i does not need to queue at charging station j;

[0099] 4) Penalty function

[0100]

[0101] In the formula: Z i For the user's electric vehicle battery capacity; Q ij L represents the amount of electricity that user i generates at the time they arrive at candidate power station j; ij Let H be the distance between demand point i and candidate power station j, and let H be the decision variable. H is 10 when user i's electricity is insufficient to reach charging station j or when the total charging amount of all users exceeds the distribution quota of candidate power station j. 15 H is 0 when user i has enough electricity to reach charging station j and the total charging amount of all users is less than the distribution quota of candidate station j.

[0102] The beneficial effects of this invention are:

[0103] 1. This application constructs a user optimal choice model based on driving time, road congestion, queuing time, and electricity price, which can determine the user optimal choice set for each driver according to different drivers' choice habits.

[0104] 2. Based on the user's best choice model, the set of user's best choices is obtained. The set of user's best choices includes the set of best optimization results and the set of driver's best addressing.

[0105] 3. This application constructs a site selection model for electric vehicle charging facilities by comprehensively considering cost minimization and user satisfaction. It takes into account both total operating costs and construction costs, as well as user satisfaction, by converting parameters such as distance and queuing time into cost formulas: distance cost and queuing cost.

[0106] 4. Based on the electric vehicle charging facility site selection model, by utilizing the optimal search result set and the driver optimal site selection set, we can obtain the adjustment scheme for the number of charging piles in the current charging facilities and the future charging facility site selection planning scheme. This can efficiently and rationally determine the urban charging facility site selection scheme while meeting the driving habits of different drivers. At the same time, it can determine the number of fast and slow charging piles corresponding to each charging station, provide a scientific number of charging piles to be built for candidate sites that have not yet built charging facilities, and solve the problem of resource waste caused by the overflow of charging piles for candidate power stations that have already built charging facilities. Attached Figure Description

[0107] Figure 1 This is a general framework diagram of a method for optimizing and planning car charging facilities based on multi-source data, as described in this invention.

[0108] Figure 2 This is the process of solving the site selection model for electric vehicle charging facilities. Detailed Implementation

[0109] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0110] S1. Based on travel time, road congestion, queuing time, and electricity price, construct a user's optimal choice model, which is expressed as:

[0111] min W=θ1W1+θ2W2+θ3W3+θ4W4

[0112] In the formula: W1 is the degree of road congestion; W2 is the travel time; W3 is the queuing time; W4 is the electricity price; θ1, θ2, θ3, and θ4 are the weights corresponding to W1-W4, and θ1+θ2+θ3+θ4=1;

[0113] It should also be noted that the road congestion level and traffic flow of each road in the user optimal choice model constructed in this invention are based on the moment when the driver generates a charging demand. On this basis, four options are obtained: travel time, road congestion level, queuing time, and electricity price. It is also set that once the driver decides to go to the selected charging facility, the choice will not be changed midway.

[0114] In this application, the values ​​of each weight θ1-θ4 of the user's best choice model are determined by the driver according to their own selection habits. If the driver prefers to choose a road with less congestion, then θ1 can be increased. Similarly, the other three weights can be adjusted accordingly.

[0115] The following explains the components of the user's optimal choice model:

[0116] 1) The degree of road congestion, W1, is represented as:

[0117]

[0118] In the formula: the location where the driver has a charging need is recorded as the demand point, and all parking lots within the area of ​​the demand point are recorded as candidate charging stations; therefore, L represents the road segment that the driver takes from the demand point to the candidate charging station (i.e., charging facility) selected by the driver; L h This represents the set of all routes from the demand point to the candidate power station; σ LThe degree of congestion of road segment L when the driver decides to go to the selected candidate power station is σ, which is based on the road conditions shown on the online map. The degree of congestion is represented by the values ​​1, 2, and 3, which respectively indicate that the road segment is smooth, relatively congested, and extremely congested. Moreover, the present invention is designed so that once the driver decides to go to the selected candidate power station (i.e., charging facility), he will not change his choice midway.

[0119] 2) Travel time

[0120]

[0121] In the formula: W2 represents the driving time from the demand point to the selected charging facility; t L 0 This represents the free-roaming time of road segment L when traffic flow is zero, which is the segment length divided by the speed limit of the segment; x L c represents the traffic flow rate of road segment L; L For t L Take 1.15t L 0 Traffic volume at that time.

[0122] 3) Queuing time

[0123]

[0124] In the formula: n is the nth position of driver i in the charging queue when driver i arrives at the charging station; Z i n-1 The battery capacity of the vehicle preceding driver i in the charging queue, Q i +-1 P represents the battery level of driver i's vehicle when it arrives at the charging station. i W3 represents the power consumption per unit time of driver i's vehicle when it is not in motion. n-1 Let T be the queuing time for driver i's vehicle. When n = 1, meaning there are no vehicles ahead of driver i in the queue, the driver only needs to wait for the first vehicle to finish charging at the charging station. q The remaining charging time is the time until the first vehicle in the charging station to finish charging when driver i arrives at the charging station. The charging power of the vehicle preceding driver i when driver i is queuing at charging station j is determined by k different charging types, which can be divided into fast charging and slow charging.

[0125] n = yb jk Let y be the y-th vehicle in the charging station when driver i's vehicle arrives, and b be the y-th vehicle in the charging station. jk Let j be the number of charging piles of type k in charging station j. Charging type k is divided into fast charging and slow charging.

[0126] In this invention, charging time refers to an estimated time required for an electric vehicle to fully charge from its remaining charge to 100% battery capacity after arriving at charging station j.

[0127]

[0128] In the formula: T cd For charging time; Z i Q represents the vehicle battery capacity for driver i; ks The remaining battery power when the vehicle starts charging; g j The charging power of the charging station.

[0129] S2. Based on the user optimal choice model, obtain the set of user optimal choices, and use this set as the initial parameters for solving the electric vehicle charging facility location model. The set of user optimal choices includes the set of optimal search results and the set of driver optimal addresses; the methods for obtaining these two sets are as follows:

[0130] S2.1 Obtain the set of best optimization results corresponding to all requirement points.

[0131] S2.1.1. Based on the selection habits of car users, assign values ​​to θ1, θ2, θ3, and θ4 in the user's optimal selection model. The four weight values ​​cannot all be the same and must reflect the priority division. For example, when the driver attaches more importance to the factors of road congestion and travel time, the values ​​of θ1 and θ2 will be increased accordingly. In this embodiment, θ1 and θ2 are 0.4 and 0.3 respectively, and θ3 and θ4 are 0.2 and 0.1 respectively.

[0132] S2.1.2. Select all parking lots with charging facilities in the area where the demand point is located as candidate power stations;

[0133] S2.1.3 Calculate the objective function value minW for the i-th demand point and all candidate power plants using the user optimal selection model after weight assignment; denote the candidate power plant corresponding to the objective function value minW of the i-th demand point as F. (minW)i ;

[0134] S2.1.4. Select the two largest weights from the four weights, namely θ1 and θ2 in this embodiment; assign a value of 1 to one of the selected larger weights, and assign a value of 0 to the other three weights; then substitute the reassigned weights into the user's optimal selection model; in this embodiment, θ1 = 1, θ2 = θ3 = θ4 = 0; denot F as the candidate power station corresponding to the i-th demand point output by the user's optimal selection model at this time. (first)i ;

[0135] One of the selected larger weights is assigned a value of 1, and the other three weights are assigned a value of 0. The reassigned weights are then substituted into the user's optimal selection model. In this embodiment, θ2 = 1, θ1 = θ3 = θ4 = 0. The candidate power station corresponding to the i-th demand point output by the user's optimal selection model at this point is denoted as F. (second)i ;

[0136] S2.1.5, the three candidate power plants obtained from S2.1.3 and S2.1.4 constitute the set of optimal search results corresponding to the i-th demand point, denoted as F. xyi =[F (minw)i F (first)i F (second)i ];

[0137] S2.1.6 Repeat the above steps to obtain the set of best optimization results corresponding to all requirement points.

[0138] S2.2 Obtain the optimal addressing set for all demand points corresponding to drivers.

[0139] S2.2.1. Based on the selection habits of car users, assign values ​​to θ1, θ2, θ3, and θ4 in the user's optimal selection model. The four weight values ​​cannot all be the same and must reflect the priority division. For example, when the driver attaches more importance to the factors of road congestion and travel time, the values ​​of θ1 and θ2 will be increased accordingly. In this embodiment, θ1 and θ2 are 0.4 and 0.3 respectively, and θ3 and θ4 are 0.2 and 0.1 respectively.

[0140] S2.2.2. All parking lots in the area where the demand point is located are considered as candidate power stations. The parking lots include those that have been equipped with charging facilities and those that have not been equipped with charging facilities.

[0141] S2.2.3. Using the weighted user optimal choice model, calculate the objective function value minW for the i-th demand point and all candidate power plants; denot the candidate power plant corresponding to the objective function value minW of the i-th demand point as F. (minW)i ';

[0142] S2.2.4. Select the two largest weights from the four weights; assign a value of 1 to one of the selected larger weights, and assign a value of 0 to the other three weights. Then substitute the reassigned weights into the user's optimal choice model, and denote the candidate power station corresponding to the i-th demand point output by the user's optimal choice model as F. (first)i ';

[0143] Assign a value of 1 to one of the selected larger weights and 0 to the other three weights. Then substitute the reassigned weights into the user's optimal choice model. Let F be the candidate power station corresponding to the i-th demand point output by the user's optimal choice model at this point.(second)i ';;

[0144] S2.2.5, the three candidate power stations obtained from S2.2.3 and S2.2.4 constitute the optimal addressing set for the driver corresponding to the i-th demand point, denoted as F. xzi =[F (min w)i’ F (first)i’ F (sec ond)i’ ];

[0145] S2.2.6 Repeat the above steps to obtain the optimal addressing set for all demand points.

[0146] S3. Construct an electric vehicle charging facility site selection model that comprehensively considers cost minimization and user satisfaction. The electric vehicle charging facility site selection model is expressed as:

[0147] min F=ω1F1+ω2F2+ω3F3+F4

[0148] In the formula: F represents the total operating cost of the charging facility to the target lifespan; F1 represents the construction cost of the charging facility to the target lifespan; F2 represents the distance cost for users to travel to the charging facility; F3 represents the queuing cost for users; F4 represents the penalty function; ω1, ω2, and ω3 are the weights corresponding to F1, F2, and F3, respectively.

[0149] The following explains the components of the electric vehicle charging facility site selection model:

[0150] 1) Construction cost of charging facilities until the target lifespan

[0151]

[0152] Where: r is the depreciation rate; m is the planned life of the charging facilities; S j The area occupied by the fast and slow traffic cones; G j S represents the unit land price for candidate power plants; jk G represents the number of fast-charging piles at candidate power stations; k Price per unit for fast charging stations; S jm G represents the number of slow-charging piles at candidate power stations; m The unit price for slow charging stations; F j For candidate power plant infrastructure costs and other expenses; h j M is the total number of candidate power plants, which is the decision variable.

[0153] 2) Cost of travel distance to charging facilities for users

[0154]

[0155] In the formula: L ijQ represents the distance from demand point i to candidate power station j. i Electricity consumption per kilometer for user i electric vehicle; D j Let x be the unit electricity price of candidate power plant j; ij Let x be the decision variable for distance cost. ij A value of 1 indicates that user i chooses to go to charging station j, and x ij A value of 0 indicates that user i did not choose to go to charging station j; N is the set of demand points.

[0156] 3) User queuing costs

[0157]

[0158] In the formula: T ij User queuing time; P i This refers to the power consumption per unit time for user i when not in driving mode; G i X represents the average hourly earnings of the surveyed users. ij X is the decision variable for queuing cost. ij A value of 1 indicates that user i needs to queue at charging station j. ij A value of 0 indicates that user i does not need to queue at charging station j.

[0159] 4) Penalty function

[0160]

[0161] In the formula: Z i For the user's electric vehicle battery capacity; Q ij L represents the amount of electricity that user i generates at the time they arrive at candidate power station j; ij Let H be the distance between demand point i and candidate power station j, and let H be the decision variable. H is 10 when user i's electricity is insufficient to reach charging station j or when the total charging amount of all users exceeds the distribution quota of candidate power station j. 15 H is 0 when user i has enough electricity to reach charging station j and the total charging amount of all users is less than the distribution quota of candidate station j.

[0162] The following explains the specific constraints of each parameter in the electric vehicle charging facility site selection model:

[0163] ∑ i∈N x ij =1 (1)

[0164] ∑ j∈M h j ≤M m (2)

[0165] x ij ≤h j i∈N j∈M (3)

[0166] Σ i∈N Σ j∈M (Z i -Q ij )≤R j i∈N j∈M (4)

[0167] L ij ≤L max i∈N j∈M (5)

[0168] x ij h j X ij ∈{0,1} i∈N j∈M (6)

[0169] H∈{0, 10} 15} (7)

[0170] ω1, ω2, ω3∈{0,1}, ω1+ω2+ω3=1 (8)

[0171] Constraint (1) means that user i can only go to one candidate power station j for charging each time;

[0172] Constraint (2) means that only the charging station determined by the scheme can be selected from the candidate power stations each time; M m The total number of candidate power plants;

[0173] Constraint (3) means that user i can only charge their device if the candidate power station j has built charging facilities.

[0174] Constraint (4) represents the service capacity limit for selecting candidate power station j, R j Let j be the power distribution quota for candidate power station;

[0175] Constraint (5) represents the maximum distance limit from demand point i to candidate power station j, L max This represents the maximum driving distance that user i can currently travel with the battery level.

[0176] Constraints (6) and (7) represent constraints on decision variables;

[0177] Constraint (8) represents a weight constraint;

[0178] S4. Based on the optimal search result set and the driver's optimal address set, solve the electric vehicle charging facility location model, and then obtain the electric vehicle charging facility planning scheme. The electric vehicle charging facility planning scheme includes the adjustment scheme of the number of charging piles in the current charging facility and the future charging facility location planning scheme.

[0179] S4.1 Based on the optimal search result set, an adjustment plan for the number of charging piles in the current charging facility is obtained. The specific process is as follows:

[0180] S4.1.1 Generate an N*E initial matrix N0, which consists of E initial vectors (column vectors), each column vector representing a scheme, for a total of E schemes;

[0181] A candidate power plant is randomly selected from the set of best optimization results corresponding to N demand points, thus forming an initial vector of length N. If the first column vector is represented as F... xyi =[F (minw)i F (first)i F (second)i This means that the first item in the column vector is a candidate power station F randomly selected from the set of best optimization results corresponding to the first demand point. (minW)i .

[0182] S4.1.2, count the candidate power stations appearing in each column vector, and record the number of charging piles of different charging types in each candidate power station; for example, the number of fast charging piles in candidate power station j in the first initial vector is b. jkc The number of slow charging stations is b jmc Summarize the number of fast charging piles and slow charging piles for the same candidate power station j in each of the E column vectors, and thus obtain the range of the number of fast charging piles for candidate power station j, denoted as b. jkc ∈[B jkc1 B jkc2 ], B jkc1 B is the minimum number of fast charging piles for candidate power station j. jkc2 This represents the maximum number of fast charging piles for candidate power station j; similarly, the number of slow charging piles b for candidate power station j can be obtained. jmc ∈[B jmc1 B jmc2 ], B jmc1 B is the minimum number of slow charging piles for candidate power station j. jmc2 It is the maximum number of slow charging piles in candidate power station j;

[0183] S4.1.3, using the electric vehicle charging facility site selection model, calculate the objective function value minF of the E column vectors in the initial matrix N0, and the number b of fast charging piles of candidate power station j at each iteration. jkc and the number of slow charging stations b jmc Each is randomly selected from its corresponding range.

[0184] S4.1.4, Update the memory matrix

[0185] (1) Calculate the concentration of each scheme's column vector (for a certain scheme) and the probability of selection:

[0186]

[0187]

[0188] In the formula: A represents the reciprocal of minF, V ij V represents the similarity between column vectors; ij The number of identical bits in vectors i and j is represented by ; n represents the length of the vector.

[0189]

[0190] In the formula: N is the number of antibodies, T is the set threshold.

[0191]

[0192] In the formula: P i α is the selection probability; α is the proportionality coefficient. When the individual fitness is higher, the expected reproduction probability is greater; when the individual concentration is lower, the expected reproduction probability is greater. This method is beneficial for selecting individuals with high fitness, while inhibiting individuals with high concentration, thus ensuring individual diversity.

[0193] (2) Set the N*U memory matrix N U Sort the objective function values ​​of all schemes in ascending order, and put the column vectors corresponding to the first S smaller objective function values ​​into the memory matrix; then sort the data concentrations of the remaining ES column vectors in ascending order, and put the column vectors corresponding to the first US smaller concentrations into the memory matrix.

[0194] (3) Select, transform, and crossover the remaining EU individuals (i.e., column vectors) to generate matrix N. E-U :

[0195] Selection: Chromosomes with better fitness are selected using a roulette wheel selection method; the individual selection probability is the selection probability P obtained in (1). i .

[0196] Crossover: Crossover is performed by randomly selecting crossover positions. Each crossover involves randomly selecting two vectors and two numbers within the range (0, N) as the crossover segment. The numbers in the two vectors within this segment are then swapped.

[0197] Mutation: Mutation is performed by randomly selecting mutation positions. Each mutation randomly selects three numbers within (0, N) as mutation positions. Based on the above steps, the optimal addressing set for each demand point is obtained, and the number at the mutation position is randomly replaced with one of the remaining two numbers in the set.

[0198] (4) Determine if the termination condition is met; if so, end the optimization. At this time, the number of fast charging piles of candidate power station j in the column vector (i.e., the scheme) corresponding to the smallest objective function value in the matrix is ​​b.jk The number of slow charging stations is b jm A plan to adjust the number of charging piles in the current charging facilities;

[0199] Conversely, proceed to the next step. The termination condition for this application can be a preset number of iterations; the iteration ends when the set number of iterations is reached.

[0200] (5) N U With N E-U Together, they form a new generation matrix N. n .

[0201] (6) New generation matrix N n Recalculate the objective function value.

[0202] S4.2. Based on the optimal addressing set of drivers, a future charging facility site selection planning scheme is obtained. The specific process is as follows:

[0203] S4.2.1 Generate an N*E initial matrix N0, which consists of E initial vectors (column vectors), each column vector representing a scheme, for a total of E schemes;

[0204] A candidate power plant is randomly selected from the set of best optimization results corresponding to N demand points, thus forming an initial vector of length N. If the first column vector is represented as F... xzi =[F (minw)i’ F (first)i’ F (second)i′ This means that the first item in the column vector is a candidate power station F randomly selected from the set of best optimization results corresponding to the first demand point. (minW)i '.

[0205] S4.2.2, count the candidate power stations appearing in each initial vector, and record the number of charging piles of different charging types in each candidate power station; for example, the number of fast charging piles in candidate power station j in the first initial vector is b. jkc The number of slow charging stations is b jmc Summarize the number of fast charging piles and slow charging piles of the same candidate power station j in the E initial vectors, and thus obtain the range of the number of fast charging piles of candidate power station j, denoted as b. jkc ∈[B jkc1 B jkc2 ], B jkc1 B is the minimum number of fast charging piles for candidate power station j. jkc2 This represents the maximum number of fast charging piles for candidate power station j; similarly, the number of slow charging piles b for candidate power station j can be obtained. jmc ∈[B jmc1 B jmc2 ], B jmc1B is the minimum number of slow charging piles for candidate power station j. jmc2 It is the maximum number of slow charging piles in candidate power station j;

[0206] S4.2.3, using the electric vehicle charging facility site selection model, calculate the objective function value minF of the E column vectors in the initial matrix N0, and the number b of fast charging piles of candidate power station j at each iteration. jkc and the number of slow charging stations b jmc Each is randomly selected from its corresponding range.

[0207] S4.2.4, Update the memory matrix

[0208] (1) Calculate the concentration of each scheme's column vector (for a certain scheme) and the probability of selection:

[0209]

[0210]

[0211] In the formula: A represents the reciprocal of minF, V ij V represents the similarity between column vectors; ij The number of identical bits in vectors i and j is represented by ; n represents the length of the vector.

[0212]

[0213] In the formula: N is the number of antibodies, T is the set threshold.

[0214]

[0215] In the formula: P i α is the selection probability; α is the proportionality coefficient. When the individual fitness is higher, the expected reproduction probability is greater; when the individual concentration is lower, the expected reproduction probability is greater. This method is beneficial for selecting individuals with high fitness, while inhibiting individuals with high concentration, thus ensuring individual diversity.

[0216] (2) Set the N*U memory matrix N U Sort the objective function values ​​of all schemes in ascending order, and put the column vectors corresponding to the first S smaller objective function values ​​into the memory matrix; then sort the data concentrations of the remaining ES column vectors in ascending order, and put the column vectors corresponding to the first US smaller concentrations into the memory matrix.

[0217] (3) Select, transform, and crossover the remaining EU individuals (i.e., column vectors) to generate matrix N. E-U :

[0218] Selection: Chromosomes with better fitness are selected using a roulette wheel selection method. The individual selection probability is the selection probability P obtained in (1). i .

[0219] Crossover: Crossover is performed by randomly selecting a crossover position. Each crossover involves randomly selecting two vectors and randomly selecting two numbers within (0, N) as the crossover segment. The numbers in the two vectors within this segment are then swapped.

[0220] Mutation: Mutation is performed by randomly selecting mutation positions. Each mutation randomly selects three numbers within (0, N) as mutation positions. Based on the above steps, the optimal addressing set for each demand point is obtained, and the number at the mutation position is randomly replaced with one of the remaining two numbers in the set.

[0221] (4) Determine if the termination condition is met; if yes, terminate; at this point, for parking lots already equipped with charging facilities, the number of fast charging piles for candidate power station j in the column vector (i.e., the scheme) corresponding to the minimum objective function value in the matrix is ​​b. jk The number of slow charging stations is b jm This is a scheme for adjusting the number of charging piles in the current charging facilities; for parking lots without charging facilities, the candidate power station j is output from the column vector (i.e., the scheme) corresponding to the minimum objective function value in the matrix, which is the site selection scheme.

[0222] Conversely, proceed to the next step. The termination condition for this application can be a preset number of iterations; the iteration ends when the set number of iterations is reached.

[0223] (5) N U With N E-U Together, they form a new generation matrix N. n .

[0224] (6) New generation matrix N n Recalculate the objective function value.

[0225] The multi-source data-based vehicle charging facility selection and planning method proposed in this application can be implemented modularly, for example, by setting up an information acquisition module, a human-computer interaction module, a decision-making module, and an information output module. Specifically, the information acquisition module uses information such as remaining battery power, charging stations, battery capacity, current road traffic conditions, charging station unit price, surrounding parking lots, unit land price, and power grid capacity to make judgments based on the multi-source data-based vehicle charging facility selection and planning method proposed in this application. The information output module is used to output the results of the decision-making module. The human-computer interaction module simultaneously outputs the optimal path corresponding to three candidate charging stations to the user, who then selects one of the candidate charging stations and follows its corresponding optimal path to charge.

[0226] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method for vehicle charging infrastructure selection and planning based on multi-source data, characterized in that, Comprising the following steps: S1, based on travel time, road congestion, queuing time, electricity price to build user best selection model; expressed as: min W = θ1W1 + θ2W2 + θ3W3 + θ4W4 In the formula: W1 is the road congestion; W2 is the travel time; W3 is the queuing time; W4 is the electricity price; θ1, θ2, θ3, θ4 are the weights corresponding to W1-W4 respectively, θ1+θ2+θ3+θ4=1; Each item of the user best selection model is expressed as: 1) The road congestion W1 is expressed as: In the formula, the position where the driver generates the charging demand is recorded as a demand point, all parking lots in the area where the demand point is located are recorded as candidate power stations, L represents the road section from the demand point to the candidate power station selected by the driver, and L h represents the collection of all road sections between the demand point and the candidate power station. σ L σ is a parameter representing the congestion degree of the road segment L when the driver decides to go to the selected candidate power station. The road condition displayed on the online map is taken as the standard. σ represents the congestion degree by numerical values 1, 2 and 3, which respectively represent that the road segment is in free flow, relatively congested and particularly congested. 2) Travel time: In the formula: W2 represents the travel time of the driver from the demand point to the selected charging facility; t L 0 denotes the free travel time of the road segment L at zero traffic flow, i.e. the length of the road segment divided by the speed limit of the road segment; x L is the traffic flow of the road segment L; c L is the free travel time of the road segment L at zero traffic flow, i.e. the length of the road segment divided by the speed limit of the road segment; x L is the traffic flow of the road segment L; c L 0 is the traffic flow of the road segment L; c S2, based on the user best selection model, the user best selection set is obtained, and the user best selection set is used as the initial parameter for solving the electric vehicle charging facility site selection model; the user best selection set includes a best optimization result set and a driver best addressing set; S3, the electric vehicle charging facility site selection model is constructed and expressed as: min F = ω1F1 + ω2F2 + ω3F3 + F4 In the formula: F represents the total operating cost of the charging facility operating to the target year limit; F1 represents the construction cost of the charging facility operating to the target year limit; F2 represents the distance cost of the user traveling to the charging facility; F3 represents the queuing cost of the user; F4 represents the penalty function; ω1, ω2, ω3 are the weights corresponding to F1, F2, F3 respectively; the electric vehicle charging facility site selection model is composed of the construction cost of the charging facility operating to the target year limit, the distance cost of the user traveling to the charging facility, the queuing cost of the user and the penalty function, which are as follows: 1) The construction cost of the charging facility operating to the target year limit Where: r is the depreciation rate; m is the planning life of charging facilities; S j is the area of fast and slow pile regions; G j is the unit land price of j candidate power stations; S jk is the number of fast charging piles of j candidate power stations; G k is the unit price of fast charging piles; S jm is the number of slow charging piles of j candidate power stations; G m is the unit price of slow charging piles; F j is the infrastructure cost and other costs of j candidate power stations; h j is the decision variable, and M is the total number of candidate power stations; 2) The distance cost of the user traveling to the charging facility wherein: L ij is the distance from demand point i to candidate power station j; Q i is the power consumption per kilometer of the electric vehicle of user i; D j is the unit price of j candidate power station; x ij is the decision variable of distance cost, x ij is 1 when user i chooses to go to j charging station, x ij is 0 when user i does not choose to go to j charging station; N is the demand point set; 3) The queuing cost of the user where: T ij is the queuing time for user i; P i is the power consumption of user i per unit time in non-driving state; G i is the average hourly income of the surveyed user; X ij is the decision variable of the queuing cost, X ij is 1 when user i needs to queue at j charging station, X ij is 0 when user i does not need to queue at j charging station; 4) The penalty function wherein: Z i Qi is the electric vehicle battery capacity of user i; Q ij Qi is the electric vehicle battery capacity of user i; Q ij is the distance between demand point i and candidate charging station j, H is a decision variable, H is 10 when the electric quantity of user i is not enough to reach the charging station j or the total sum of all user charging quantities is higher than the distribution capacity of candidate charging station j 15 H is 0 when the electric quantity of user i is enough to reach the charging station j and the total sum of all user charging quantities is lower than the distribution capacity of candidate charging station j. S4, based on the best optimization result set and the driver best addressing set, the electric vehicle charging facility site selection model is solved, and then the automobile charging facility planning scheme is obtained; the automobile charging facility planning scheme includes an adjustment scheme for the number of charging piles in the current charging facility and a future charging facility site selection planning scheme.

2. The method for multi-source data based selection and planning of electric vehicle charging infrastructure as claimed in claim 1, wherein, The method for obtaining the best optimization result set in S2 is: S2.1.1, according to the selection habit of automobile users, θ1, θ2, θ3, θ4 in the user best selection model are valued, and the values of the four weights cannot all be the same, which needs to reflect the priority division; S2.1.2, all parking lots with charging facilities in the region of the demand point are taken as candidate power stations; S2.1.3, calculate the target function value minW of the i-th demand point and all candidate power stations using the user's best selection model after weight assignment; record the candidate power station corresponding to the target function value minW of the i-th demand point as F (minW)i ; S2.1.4, select the larger two weights from the four weights, assign one of the selected larger weights as 1, and assign the remaining three weights as 0, and then substitute the re-assigned weights into the user best selection model; at this time, the user best selection model outputs the candidate power station corresponding to the i-th demand point as F (first)i ; Another weight of the selected larger weights is assigned a value of 1, and the remaining three weights are assigned a value of 0, and the re-assigned weights are substituted into the user best choice model; at this time, the user best choice model outputs the candidate power station corresponding to the i th demand point as F (second)i ; S2.1.5, the three candidate power stations obtained from S2.1.3 and S2.1.4 constitute the optimal searching result set corresponding to the i-th demand point, denoted as F xyi = [F (minw)i , F (first)i , F (second)i ] ; S2.1.6, repeat the above steps to obtain the best optimization result set corresponding to all demand points.

3. The automobile charging facility selection and planning method based on multi-source data according to claim 1, characterized in that, The method for obtaining the driver best addressing set in S2 is: S2.2.1, according to the selection habit of automobile users, θ1, θ2, θ3, θ4 in the user best selection model are valued, and the values of the four weights cannot all be the same, which needs to reflect the priority division; S2.2.2, all parking lots in the area where the demand point is located are taken as candidate power stations, including parking lots with charging facilities and parking lots without charging facilities; S2.2.3, using the user's best choice model after weight assignment to calculate the target function value minW of the i-th demand point and all candidate power stations; record the candidate power station corresponding to the target function value minW of the i-th demand point as F (minW)i ’; S2.2.4, select two larger weights from the four weights; assign one of the selected larger weights as 1, and the remaining three weights as 0, and then substitute the re-assigned weights into the user best selection model, and the candidate power station corresponding to the i-th demand point output by the user best selection model at this time is recorded as F (first)i ’; Another weight in the selected larger weights is assigned a value of 1, and the remaining three weights are assigned a value of 0, and the re-assigned weights are substituted into the user best choice model; the candidate power plant corresponding to the i-th demand point output by the user best choice model at this time is recorded as F (second)i ’; S2.2.5, the three candidate power stations obtained from S2.2.3 and S2.2.4 constitute a driver best addressing set corresponding to the ith demand point, denoted as F xzi = [F (min w)i’ ,F (first)i’ ,F (second)i’ ]; S2.2.6, repeat the above steps to obtain the driver's best addressing collection corresponding to all demand points.

4. The method for multi-source data based selection and planning of electric vehicle charging infrastructure as claimed in claim 2, wherein, The method for obtaining the adjustment scheme of the number of charging piles in the current charging facility based on the best optimization result collection in S4 is as follows: S4.1.1, generate an initial matrix N0 of N*E, the initial matrix N0 is composed of E column vectors, and each column vector represents a scheme, a total of E schemes; An initial vector of length N is formed by randomly selecting a candidate power station from the best optimization result collection corresponding to each of the N demand points in turn; S4.1.2, count the candidate power stations in each column vector and record the number of different charging types of power piles in each candidate power station respectively; let the number of fast charging piles in candidate power station j in the column vector be b jkc , and the number of slow charging piles be b jmc ; respectively aggregate the number of fast charging piles and the number of slow charging piles of the same candidate power station j in the E column vectors, thereby obtaining the range of the number of fast charging piles of the candidate power station j, denoted as b jkc ∈ [B jkc1 , B jkc2 ], B jkc1 is the minimum value of the number of fast charging piles of candidate power station j, and B jkc2 is the maximum value of the number of fast charging piles of candidate power station j; similarly, the number of slow charging piles of candidate power station j is b jmc ∈ [B jmc1 , B jmc2 ], B jmc1 is the minimum value of the number of slow charging piles of candidate power station j, and B jmc2 is the maximum value of the number of slow charging piles of candidate power station j; S4.1.3, calculate the objective function value minF of each column vector in the initial matrix N0 by using the electric vehicle charging facility site selection model, and select the number of fast charging piles b of the power station j at each iteration jkc and the number of slow charging piles b jmc are randomly selected in their respective ranges, respectively; S4.1.4, update the memory matrix (1) Calculate the concentration of each scheme and the selection probability: where: A represents the inverse of minF, V ij represents the similarity between column vectors; V ij represents the number of bits that are the same in vector i and vector j; n represents the vector length; wherein: N is the number of antibodies, T is a set threshold value; In the formula: P i α is the selection probability; α is the proportionality coefficient; when the individual fitness is higher, the expected reproduction probability is greater; when the individual concentration is lower, the expected reproduction probability is greater. This method is beneficial for selecting individuals with high fitness, while inhibiting individuals with high concentration, thus ensuring individual diversity. (2) Set the memory matrix N of N*U U Sort the objective function values of all schemes from small to large, and put the column vectors corresponding to the first S smaller objective function values into the memory matrix. Then, sort the remaining E-S column vectors according to the concentration from small to large, and put the column vectors corresponding to the first U-S smaller concentrations into the memory matrix. (3) Select the remaining E-U individuals, i.e. column vectors, and perform mutation, crossover operations to generate a matrix N E-U : Selection: Roulette wheel selection is used to select chromosomes with better fitness, and the selection probability P of individual selection is the selection probability obtained in (1) i ; Crossing: random selection of crossing position is adopted for crossing; two vectors are randomly selected each time, and two numbers in (0, N) are randomly selected as the crossing segment, and the numbers in the two vectors in the segment are exchanged; Mutation: random selection of mutation position is adopted for mutation; three numbers in (0, N) are randomly selected as the mutation position each time, and the number in the mutation position is randomly replaced with one of the remaining two numbers in the collection according to the above steps to obtain the best addressing collection of each demand point; (4) judging whether the ending condition is met; if yes, ending the optimization, at this time the column vector corresponding to the minimum objective function value in the matrix is the number of fast charging piles b of the candidate power station j in the scheme jk and the number of slow charging piles b jm is the adjustment scheme of the number of charging piles in the current charging facility; Otherwise, the next operation is performed; the preset iteration number is reached, and the iteration is ended when the set number of times is reached; (5) will be N U combined with N E-U together constitute a new generation of matrix N n ; (6) New generation of matrix N n The objective function value is recalculated.

5. The method for multi-source data based selection and planning of electric vehicle charging infrastructure as claimed in claim 3, wherein, Based on the driver's best addressing collection, the method for obtaining the future charging facility site planning scheme is as follows: S4.2.1, generate an initial matrix N0 of N*E, the initial matrix N0 is composed of E column vectors, and each column vector represents a scheme, a total of E schemes; An initial vector of length N is formed by randomly selecting a candidate power station from the best optimization result collection corresponding to each of the N demand points in turn; S4.2.2, count the number of candidate power stations in each initial vector and record the number of different charging types of power piles in each candidate power station respectively; let the number of fast charging piles in candidate power station j in column vector be b jkc , and the number of slow charging piles be b jmc ; respectively aggregate the number of fast charging piles and the number of slow charging piles of the same candidate power station j in E initial vectors, thereby obtaining the range of the number of fast charging piles of candidate power station j, denoted as b jkc ∈ [B jkc1 , B jkc2 ], B jkc1 is the minimum value of the number of fast charging piles of candidate power station j, and B jkc2 is the maximum value of the number of fast charging piles of candidate power station j; similarly, the number of slow charging piles of candidate power station j is b jmc ∈ [B jmc1 , B jmc2 ], B jmc1 is the minimum value of the number of slow charging piles of candidate power station j, and B jmc2 is the maximum value of the number of slow charging piles of candidate power station j; S4.2.3, calculate the objective function value minF of each column vector in the initial matrix N0 by using the electric vehicle charging facility site selection model, and select the number of fast charging piles b of the power station j at each iteration jkc and the number of slow charging piles b jmc are randomly selected in their respective ranges, respectively; S4.2.4, update the memory matrix (1) Calculate the concentration of each scheme column vector and the selection probability: where: A represents the inverse of minF, V ij represents the similarity between column vectors; V ij represents the number of bits that are the same in vector i and vector j; n represents the vector length; wherein: N is the number of antibodies, T is a set threshold value In the formula: P i α is the selection probability; α is the proportionality coefficient; when the individual fitness is higher, the expected reproduction probability is greater; when the individual concentration is lower, the expected reproduction probability is greater. This method is beneficial for selecting individuals with high fitness, while inhibiting individuals with high concentration, thus ensuring individual diversity. (2) Set the memory matrix N of N*U U Sort the objective function values of all schemes from small to large, and put the column vectors corresponding to the first S smaller objective function values into the memory matrix. Then, sort the remaining E-S column vectors according to the concentration from small to large, and put the column vectors corresponding to the first U-S smaller concentrations into the memory matrix. (3) Select the remaining E-U individuals, i.e. column vectors, and perform mutation, crossover operations to generate a matrix N E-U : Selection: Roulette wheel selection is used to select chromosomes with better fitness, and the selection probability P of individual selection is the selection probability obtained in (1) i ; Crossing: random selection of crossing position is adopted for crossing; two vectors are randomly selected each time, and two numbers in (0, N) are randomly selected as the crossing segment, and the numbers in the two vectors in the segment are exchanged; Mutation: random selection of mutation position is adopted for mutation; three numbers in (0, N) are randomly selected as the mutation position each time, and the number in the mutation position is randomly replaced with one of the remaining two numbers in the collection according to the above steps to obtain the best addressing collection of each demand point; (4) determine whether the end condition is met; if yes, end; at this time, for the parking lot equipped with charging facilities, the column vector corresponding to the minimum objective function value in the matrix is the number of fast charging piles of candidate power stations j in the scheme b jk and the number of slow charging piles is b jm is the adjustment scheme of the number of charging piles in the current charging facility; for the parking lot without charging facilities, the column vector corresponding to the minimum objective function value in the matrix is the output candidate power station j in the scheme, that is, the site selection scheme is obtained; Otherwise, the next operation is performed; the preset iteration number is reached, and the iteration is ended when the set number of times is reached; (5) will be combined with N U and N E-U together constitute a new generation of matrix N n ; (6) New generation of matrix N n The objective function value is recalculated.

6. The method for multi-source data based selection and planning of electric vehicle charging infrastructure as claimed in any one of the claims 1-5, wherein, The values of the weights θ1-θ4 of the user's best selection model are determined by the driver according to his selection habit, assuming that the driver prefers to select a lower road congestion level, then θ1 is increased, and the other three weights are adjusted accordingly.

7. The method for multi-source data based selection and planning of electric vehicle charging infrastructure as claimed in claim 1, wherein, The road congestion level and the traffic flow of each road of the user's best selection model are based on the moment when the driver generates the charging demand, and on this basis, the driving time, the road congestion level, the queuing time and the electricity price are obtained, and it is assumed that the driver will not change the selection midway after determining to go to the selected charging facility.

Images