A structural entropy-based method for site selection and capacity optimization of electric vehicle charging stations

Abstract

The application is suitable for the technical field of charging station site selection, and provides a method for electric vehicle charging station site selection and capacity optimization based on structural entropy, comprising: constructing an initial feature vector of a candidate parking lot based on business information and geographic location information of the candidate parking lot; constructing a weighted network graph and initializing the weighted network graph as a coding tree; calculating the structural entropy of the candidate parking lot based on the coding tree, generating an enhanced feature vector based on the structural entropy and the initial feature vector; clustering the candidate parking lot based on the enhanced feature vector to obtain multiple clusters, and defining each cluster as a demand area; constructing an energy utilization rate function based on the demand area, taking the energy utilization rate function as a fitness function of a genetic algorithm; and optimizing the electric vehicle charging station construction scheme by using the genetic algorithm to obtain an optimal electric vehicle charging station construction scheme which maximizes the fitness function and meets the construction constraint condition. The application can improve the energy utilization rate so that the energy can be fully utilized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of charging station site selection, and particularly relates to a method for electric vehicle charging station site selection and capacity optimization based on structural entropy. BACKGROUND

[0002] The current electric vehicle charging infrastructure site selection method has significant limitations. The existing technology tends to evaluate candidate parking lots in isolation, focusing on their inherent attributes, but completely ignores their position in the regional charging network topology, and cannot identify key nodes with network centrality.

[0003] In traditional charging station site selection, geographical coverage is often given priority over energy utilization rate as the core objective, and only rough estimates are used without accurately modeling the spatial and temporal dynamic differences in core operating indicators such as vehicle flow and user residence time in different parking lots, resulting in insufficient energy utilization rate. SUMMARY

[0004] The embodiment of the application provides a method for electric vehicle charging station site selection and capacity optimization based on structural entropy, which can solve the problem of insufficient energy utilization rate.

[0005] The embodiment of the application provides a method for electric vehicle charging station site selection and capacity optimization based on structural entropy, which comprises:

[0006] For each candidate parking lot in the plurality of candidate parking lots, an initial feature vector of the candidate parking lot is constructed based on business information and geographic location information of the candidate parking lot; the business information of the candidate parking lot is used to describe the parking situation in the candidate parking lot and the charging potential of the candidate parking lot during the off-peak period of the power grid;

[0007] A weighted network graph is constructed and initialized as a coding tree; the plurality of nodes in the weighted network graph correspond one-to-one to the plurality of candidate parking lots, and each edge in the weighted network graph represents the distance between the initial feature vectors corresponding to the two candidate parking lots connected by the edge; the plurality of nodes in the weighted network graph correspond one-to-one to the plurality of leaf nodes in the coding tree;

[0008] The structural entropy of each candidate parking lot is calculated based on the coding tree, and the enhanced feature vector of each candidate parking lot is generated based on the structural entropy and the initial feature vector of each candidate parking lot;

[0009] Based on the enhanced feature vector of each candidate parking lot, the plurality of candidate parking lots are clustered to obtain a plurality of clusters, and each cluster is defined as a demand area;

[0010] An energy utilization rate function is constructed based on the defined demand area, and the energy utilization rate function is used as the fitness function of the genetic algorithm;

[0011] The construction scheme of electric vehicle charging stations is optimized using a genetic algorithm to obtain the optimal electric vehicle charging station construction scheme that maximizes the fitness function and satisfies the preset construction constraints. The construction scheme of electric vehicle charging stations includes the construction locations of multiple electric vehicle charging stations and the number of charging piles configured in each electric vehicle charging station.

[0012] Optional business information includes average daily traffic volume, average parking time, proportion of vehicles parked for extended periods, and off-peak charging potential index;

[0013] Candidate parking lots initial feature vector for: , Indicates candidate parking lots longitude, Indicates candidate parking lots latitude, Indicates candidate parking lots The average daily traffic flow Indicates candidate parking lots Average parking time Indicates candidate parking lots The proportion of vehicles parked for extended periods Indicates candidate parking lots The low-end charging potential index;

[0014] , This indicates the number of times an electric vehicle stops during the observation period. This indicates the total number of days included in the observation period;

[0015] , Indicates the first observation within the observation period The time it takes for an electric vehicle to leave the parking lot. Indicates the first observation within the observation period The entry time for the next electric vehicle parking session;

[0016] , This indicates the number of times an electric vehicle stops during the observation period when its parking duration exceeds a preset time threshold.

[0017] , This indicates the number of times an electric vehicle stops during the off-peak hours of the power grid within the observation period when the parking duration exceeds a preset threshold.

[0018] Optionally, the structural entropy of each candidate parking lot is calculated based on the coding tree, including:

[0019] Candidate parking lots are calculated using the following formula. structural entropy of :

[0020] ;

[0021] wherein, denotes a candidate parking lot in the encoding tree corresponding to the path from the leaf node to the root node, is a subtree represented by a node on denotes the sum of the weights of all edges connecting the subtrees internal nodes and external nodes, denotes the sum of the weights of all edges in the weighted network graph, , denotes the set of all nodes in the weighted network graph, denotes the weighted degree of the candidate parking lot corresponding to the node , denotes the sum of the weighted degrees of all nodes in the subtree , denotes the subtree represented by the parent node of the subtree , denotes the sum of the weighted degrees of all nodes in the subtree .

[0022] Optionally, based on the structural entropy and the initial feature vector of each candidate parking lot, an enhanced feature vector of each candidate parking lot is generated, including:

[0023] For each candidate parking lot in the plurality of candidate parking lots, the structural entropy of the candidate parking lot is added to the initial feature vector of the candidate parking lot to obtain an enhanced feature vector of the candidate parking lot.

[0024] Optionally, based on the enhanced feature vector of each candidate parking lot, the plurality of candidate parking lots are clustered to obtain a plurality of clusters, including:

[0025] The enhanced feature vectors of the plurality of candidate parking lots are clustered using a weighted K-means clustering algorithm to obtain a plurality of vector clusters;

[0026] The candidate parking lot corresponding to each vector cluster is taken as a cluster group;

[0027] In the clustering process, the distance between the enhanced feature vector of the candidate parking lot and the cluster center is calculated using the following formula:

[0028] ​​;

[0029] wherein, denotes the total dimension of the enhanced feature vector, denotes a predefined dynamic weight vector corresponding to the i-th dimension of the enhanced feature vector , denotes the dynamic weight corresponding to the i-th dimension of the enhanced feature vector , denotes the value of the enhanced feature vector at the i-th dimension, denotes the value of the cluster center at the i-th dimension.

[0030] Optionally, the energy utilization rate function is:

[0031] ;

[0032] wherein, denotes the energy utilization rate, denotes the predicted total charging amount, denotes the potential total service capacity.

[0033] Optionally, the expression of the predicted total charging amount is:

[0034] ;

[0035] wherein, denotes the total number of candidate parking lots, denotes 1 if the candidate parking lot is selected to establish an electric vehicle charging station, and denotes 0 if the candidate parking lot is not selected to establish an electric vehicle charging station, denotes 1 if the candidate parking lot is selected to establish an electric vehicle charging station, and denotes 0 if the candidate parking lot is not selected to establish an electric vehicle charging station, denotes the total predicted annual demand amount that can be captured by the electric vehicle charging station at the candidate parking lot , denotes a preset valley charging value coefficient, denotes the valley charging potential index of the candidate parking lot

[0036] ;

[0037] wherein, denotes the total number of demand areas, denotes the charging demand amount flowing from the demand area and captured by the electric vehicle charging station ; ​

[0038] ;

[0039] wherein, represents the demand area total annual charging demand potential, represents the electric vehicle charging station attractiveness function value, , represents the electric vehicle charging station configured charging pile number, represents the electric vehicle charging station travel friction function value, , represents the preset distance attenuation coefficient, represents the geographical distance or the average driving time estimated by the navigation software between the center of the demand area and the electric vehicle charging station , represents a set consisting of a plurality of electric vehicle charging stations in the electric vehicle charging station construction scheme, represents the attractiveness function value of the electric vehicle charging station , represents the travel friction function value of the electric vehicle charging station .

[0040] Optionally, the expression of the total potential service capacity is:

[0041] ;

[0042] wherein, represents the configured charging pile number of the electric vehicle charging station , represents the average rated power of a single charging pile, represents the annual effective service hours.

[0043] Optionally, the preset construction constraint condition is:

[0044] ;

[0045] ;

[0046] ;

[0047] wherein, represents the preset number of constructions, represents the preset total number of charging piles, represents the maximum number of charging piles that the electric vehicle charging station can be configured.

[0048] Optionally, the plurality of chromosomes in the initial population of the genetic algorithm correspond one-to-one to the plurality of electric vehicle charging station stationing schemes.

[0049] The above scheme of the present application has the following beneficial effects:

[0050] In the embodiments of the present application, by generating an enhanced feature vector of the candidate parking lot based on multi-dimensional factors such as real business data, geographical position and network topology of the candidate parking lot, clustering each candidate parking lot based on the enhanced feature vector, defining candidate parking lots with similar characteristics as a demand area, then constructing an energy utilization rate function based on the demand area, and taking the energy utilization rate function as the fitness function of the genetic algorithm, finally optimizing the charging station stationing position and capacity configuration by using the genetic algorithm, the stationing position and capacity configuration under the maximum energy utilization rate can be obtained, so the present application can improve the energy utilization rate and make the energy fully utilized.

[0051] Other beneficial effects of the present application will be described in detail in the subsequent specific embodiments section. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0053] Figure 1 The flowchart of the electric vehicle charging station site selection and capacity optimization method based on structural entropy provided by an embodiment of the present application. DETAILED DESCRIPTION

[0054] In the following description, specific details are set forth in order to provide a thorough understanding of the embodiments of the present application. However, persons skilled in the art will understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits and methods are omitted so as not to obscure the description of the present application with unnecessary details.

[0055] It should be understood that when used in the specification and the appended claims of the present application, the term "comprising" indicates the presence of the described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or sets thereof.

[0056] It should also be understood that the term "and / or" as used herein refers to a conjunction, an association, one or more of any combination of associated listed terms, and all possible combinations, and includes these combinations.

[0057] As used in the description of the application and the appended claims, the term "if' can be interpreted to mean "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [a described condition or event] is detected" can be interpreted to mean "upon determining" or "in response to determining" or "upon detecting [the described condition or event]" or "in response to detecting [the described condition or event]," depending on the context.

[0058] In addition, in the description and the appended claims of the application, the terms "first", "second", "third", etc. are used only to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

[0059] Reference in the specification to "one embodiment" or "some embodiments" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. The appearances of the phrases "in one embodiment", "in some embodiments", "in other embodiments", "in additional embodiments", etc. in various places in the specification are not necessarily all referring to the same embodiment, although they can. The terms "comprising", "including", "having" and their variants, mean "including but not limited to", unless otherwise expressly specified.

[0060] In view of the problem of insufficient energy utilization, the embodiments of the application provide a method for site selection and capacity optimization of an electric vehicle charging station based on structural entropy. The method generates an enhanced feature vector of a candidate parking lot based on multi-dimensional factors such as real business data, geographic location and network topology of the candidate parking lot, and clusters the candidate parking lots based on the enhanced feature vector, defines candidate parking lots with similar characteristics as a demand area, then constructs an energy utilization rate function based on the demand area, and uses the energy utilization rate function as the fitness function of a genetic algorithm, and finally optimizes the site selection and capacity configuration of the charging station using the genetic algorithm, so as to obtain the site selection and capacity configuration under the maximum energy utilization rate. Therefore, the application can improve the energy utilization rate and make the energy fully utilized.

[0061] The method for site selection and capacity optimization of an electric vehicle charging station based on structural entropy provided by the application will be described below in conjunction with specific embodiments.

[0062] AsFigure 1 As shown, the method for site selection and capacity optimization of electric vehicle charging stations based on structural entropy provided by the embodiments of the present application includes the following steps:

[0063] Step 11, for each candidate parking lot in the plurality of candidate parking lots, an initial feature vector of the candidate parking lot is constructed based on the business information and geographic location information of the candidate parking lot.

[0064] The business information of the candidate parking lot is used to describe the parking situation in the candidate parking lot and the charging potential of the candidate parking lot during the low valley period of the power grid.

[0065] It should be noted that the present application focuses on the problem of using existing parking lots to transform charging stations, and the purpose is to use at least part of the above plurality of candidate parking lots as electric vehicle charging stations, so it is necessary to construct the initial feature vector of each candidate parking lot to quantify the operation situation and location information of the candidate parking lot.

[0066] In some embodiments of the present application, the business information of the above candidate parking lot specifically includes the daily vehicle flow, the average parking time, the long parking vehicle proportion, and the low valley charging potential index of the candidate parking lot.

[0067] Correspondingly, the initial feature vector of the candidate parking lot is:

[0068]

[0069] In the above formula, represents the longitude of the candidate parking lot ; represents the latitude of the candidate parking lot , and the longitude and latitude of the candidate parking lot can be obtained by global positioning system (GPS) technology; represents the daily vehicle flow of the candidate parking lot , which is used to measure the potential user base of the candidate parking lot ; represents the average parking time of the candidate parking lot , which is used to measure the possibility and potential charging time of the user to charge; represents the long parking vehicle proportion of the candidate parking lot , which is used to identify the size of the core charging user group; represents the low valley charging potential index of the candidate parking lot , which is used to evaluate the coupling strength between the long parking behavior of the candidate parking lot and the low valley period of the power grid.

[0070] , ​​This indicates the number of times an electric vehicle stops during the observation period. This indicates the total number of days included in the observation period. This observation period can be set according to actual conditions, such as the past three months or the past six months. For example, the number of electric vehicle parking visits within the observation period can be obtained through the candidate parking lot management system.

[0071] , Indicates the first observation within the observation period The exit time of the next electric vehicle parking (which can be understood as the candidate parking lot) The Middle The electric vehicle that stopped the previous parking lot left the candidate parking lot. (time) Indicates the first observation within the observation period The entry time for the next electric vehicle parking (which can be understood as the candidate parking lot) The Middle The electric vehicle corresponding to the next electric vehicle parking should enter the candidate parking lot. (Time). For example, exit and entry times can be obtained through the management system of the candidate parking lot.

[0072] , This indicates the number of times an electric vehicle parked during the observation period when its parking duration exceeded a preset time threshold. The preset time threshold can be set according to actual conditions, such as 2 hours or 2.5 hours. For example, it can be obtained through the candidate parking lot management system. .

[0073] , This indicates the number of times electric vehicles parked during off-peak hours within the observation period when their parking duration exceeded a preset threshold. For example, this can be obtained through a candidate parking lot management system. .

[0074] It should be noted that, for candidate parking lots initial feature vector When calculating the average daily traffic flow, average parking duration, proportion of long-term parked vehicles, and off-peak charging potential index, this candidate parking lot was used. The corresponding relevant data (i.e., the data required for the calculation).

[0075] It needs to be further explained that, in order to facilitate subsequent data processing, after obtaining the daily average traffic flow, the average parking time, the long-stopping vehicle proportion and the valley charging potential index according to the above calculation method, the minimum-maximum normalization method can be used to linearly scale the numerical values of all features (i.e. longitude, latitude, daily average traffic flow, average parking time, long-stopping vehicle proportion and valley charging potential index) to the unified interval of [0, 1]. That is, the longitude, latitude, daily average traffic flow, average parking time, long-stopping vehicle proportion and valley charging potential index in the initial feature vector of the above candidate parking lot are all normalized data.

[0076] Step 12, construct a weighted network graph, and initialize the weighted network graph as the coding tree.

[0077] The plurality of nodes in the above weighted network graph correspond one-to-one to the plurality of candidate parking lots, and each edge in the weighted network graph represents the distance between the initial feature vectors corresponding to the two candidate parking lots connected by the edge. The plurality of nodes in the weighted network graph correspond one-to-one to the plurality of leaf nodes in the coding tree.

[0078] It needs to be noted that, in order to comprehensively capture the mutual relationship between all candidate parking lots, the weighted network graph is a complete graph, that is, there is an edge between any two different nodes. The weight of the edge in the weighted network graph can be determined by the Gaussian kernel function according to the Euclidean distance between the two nodes, that is, the weight of the edge is obtained by the Gaussian kernel function on the initial feature vectors corresponding to the two candidate parking lots connected by the edge.

[0079] In the weighted network graph, the weighted degree of the node corresponding to the candidate parking lot is defined as , which is equal to the sum of the weights of all edges connected to the node ; the definition is the sum of the weights of all edges in the weighted network graph, , and the set of all nodes in the weighted network graph is represented by

[0080] In some embodiments of the present application, after obtaining the weighted network graph, the weighted network graph can be converted into a coding tree by using the traditional method of generating a coding tree. In related technologies, a hierarchical coding tree can be constructed by maximizing entropy reduction iteration. This step constructs a coding tree that can reflect the hierarchical structure of the network through a bottom-up iterative process aimed at maximizing global entropy reduction.

[0081] ​For the convenience of understanding, the generation process of the coding tree is briefly described herein. Specifically, the generation process of the coding tree includes the following loop merging and coding tree structure normalization processing.

[0082] Loop merging: this process repeatedly finds and merges a pair of sub-trees that can make the total structure entropy of the system produce the maximum reduction , taking all original nodes as initial sub-trees. The formula for calculating the entropy reduction (i.e. the reduction amount) is as follows: wherein is the network partition before merging, is the new partition after merging and , and represents the total structure entropy corresponding to , and represents the total structure entropy corresponding to . By repeatedly performing this optimal merging operation, all nodes eventually form a binary coding tree.

[0083] Coding tree structure normalization processing: the height of the preliminarily constructed coding tree is highly normalized so that its final effective height is unified to the preset target tree height . This process includes:

[0084] If the actual height of the tree is greater than , pruning processing is performed, and then some internal node that can make the increase of the total structure entropy of the system the smallest is iteratively removed until the tree height meets the constraint.

[0085] If the depth of a leaf node in the tree is less than , height padding is performed, and the effective path length from the leaf node to the root node can be supplemented to by logically adding virtual nodes and the like.

[0086] Step 13: calculate the structure entropy of each candidate parking lot based on the coding tree, and generate an enhanced feature vector of each candidate parking lot based on the structure entropy of each candidate parking lot and the initial feature vector.

[0087] In some embodiments of the present application, the structure entropy of the candidate parking lot can be calculated by the following formula:

[0088] ;

[0089] wherein represents the candidate parking lot in the coding tree a path from a leaf node to a root node, is a subtree represented by a node on denotes a set of subtrees the sum of weights of all edges between internal nodes and external nodes, denotes the sum of weights of all edges in a weighted network graph, , denotes a set of all nodes in a weighted network graph, denotes a set of candidate parking lots corresponding to a node weighted degree of denotes the sum of weighted degrees of all nodes in a subtree corresponding to a parent node of a subtree denotes a subtree represented by a parent node of a subtree denotes the sum of weighted degrees of all nodes in a subtree corresponding to a parent node of a subtree .

[0090] In some embodiments of the present application, the structural entropy of each candidate parking lot in the plurality of candidate parking lots can be added to the initial feature vector of the candidate parking lot, respectively, to obtain an enhanced feature vector of the candidate parking lot.

[0091] It should be noted that, in order to facilitate data processing, the calculated structural entropy needs to be normalized, and then the normalized structural entropy is added to the initial feature vector of the candidate parking lot to obtain the enhanced feature vector. For the candidate parking lot , the enhanced feature vector of the candidate parking lot may be represented as: It can be understood that all features in the enhanced feature vector are normalized values.

[0092] Step 14, based on the enhanced feature vector of each candidate parking lot, clustering the plurality of candidate parking lots to obtain a plurality of clusters, and defining each cluster as a demand area.

[0093] In some embodiments of the present application, before performing clustering, a dynamic weight vector corresponding to the dimension of the enhanced feature vector (also referred to as a predefined dynamic weight vector Each weight element in the dynamic weight vector corresponds to one dimension of the geographical information, the daily average traffic volume, the average parking duration, the long-stay vehicle proportion, the valley charging potential index, and the structural entropy in the feature vector. It should be noted that the dynamic weight vector The values of the elements in the dynamic weight vector can be preset and adjusted according to different development stages or targets, so as to guide the final result of clustering and improve the clustering accuracy.

[0094] In some embodiments of the present application, the weighted K-means clustering algorithm can be specifically used to cluster the enhanced feature vectors of the plurality of candidate parking lots to obtain a plurality of vector clusters; and then the candidate parking lot corresponding to each vector cluster is taken as a cluster group.

[0095] The core of the weighted K-means clustering algorithm is the distance measurement method, which uses the weighted Euclidean distance to calculate the distance between a data point and a cluster center. In combination with the dynamic weight vector In the clustering process, the distance between the enhanced feature vector of the candidate parking lot and the cluster center is calculated by using the following formula:

[0096]

[0097] wherein, d represents the total dimension of the enhanced feature vector, w represents the dynamic weight corresponding to the i th dimension of the enhanced feature vector in the dynamic weight vector (i.e., the weight element described above), xi represents the value of the enhanced feature vector in the i th dimension, xci represents the value of the cluster center in the i th dimension. It should be noted that the i th dimension of the enhanced feature vector is one of the geographical information, the daily average traffic volume, the average parking duration, the long-stay vehicle proportion, the valley charging potential index, and the structural entropy.

[0098] ​​​​​​​​After clustering the candidate parking lots, each cluster is defined as an independent demand region. The center (i.e., the geographic center) of the demand region is determined by the centroid of the coordinates of all candidate parking lots within that cluster. Specifically, the arithmetic mean of the longitude coordinates of all candidate parking lots within the cluster can be used as the longitude of the demand region's center, and the arithmetic mean of the latitude coordinates of all candidate parking lots can be used as the latitude of the demand region's center, thus serving as the center of the demand region.

[0099] Step 15: Construct an energy utilization function based on the defined demand region, and use the energy utilization function as the fitness function of the genetic algorithm.

[0100] The energy utilization rate function is as follows:

[0101] ;

[0102] in, Indicates energy efficiency. This indicates the predicted total charging amount. This indicates the total potential service capacity.

[0103] The following is a prediction of the total charging amount. The calculation method is illustrated by example.

[0104] In some embodiments of this application, the total charging amount is predicted. The expression is:

[0105] ;

[0106] in, This represents the total number of candidate parking lots. ,like This indicates that you are selecting a candidate parking lot. If electric vehicle charging stations are built, This means you do not choose a candidate parking lot. Establish electric vehicle charging stations, Indicates candidate parking lots As an electric vehicle charging station Capable total annual forecast demand This represents the preset off-peak charging value coefficient. It is generally set to a non-negative constant. Indicates candidate parking lots The low-end charging potential index. This ensures that only the selected candidate parking lots are used. Only the amount of charge generated is included in the total charge.

[0107] ;

[0108] wherein, denotes the total number of demand areas, denotes the captured charging demand amount from demand areas flowing to and being charged by the electric vehicle charging station .

[0109] ;

[0110] The above formula can be understood as a charging demand gravity model for simulating market competition and user selection behavior.

[0111] wherein, denotes the demand area total annual charging demand potential (i.e., total annual charging demand potential) is generated by aggregating the core business indicators (i.e., daily vehicle flow, average parking duration, long-stay vehicle ratio) of all candidate parking lots within the corresponding cluster group. Specifically, it is the weighted sum of the daily vehicle flow, average parking duration, and long-stay vehicle ratio of all candidate parking lots within the cluster group, and the specific weights can be set according to actual conditions. In this way, the size of the total annual charging demand potential directly reflects the inherent charging potential of the cluster group.

[0112] denotes the attractiveness function value of the electric vehicle charging station , which is used to quantify the comprehensive attractiveness of the electric vehicle charging station to users. In this application, the attractiveness function value (i.e., attractiveness) is mainly determined by the service capacity of the electric vehicle charging station , which is related to the number of charging piles allocated to the electric vehicle charging station . Specifically, , denotes the number of charging piles configured by the electric vehicle charging station .

[0113] denotes the travel friction function value of the electric vehicle charging station , which is used to quantify the degree of attenuation of the user's willingness to travel from the center of the demand area to the electric vehicle charging station with distance or time. Specifically, , denotes a preset distance attenuation coefficient for describing the sensitivity of users in the area to travel costs, which can be calibrated through statistical analysis of existing user behavior data; denotes the demand area​​ geographical distance between the center and the electric vehicle charging station or the average driving time estimated by navigation software.

[0114] represents a set of multiple electric vehicle charging stations in the electric vehicle charging station construction scheme, represents the attractiveness function value of the electric vehicle charging station represents the travel friction function value of the electric vehicle charging station It should be noted that is determined in the same way as the foregoing , except that is the attractiveness function value of the electric vehicle charging station , which needs to be calculated using the relevant data of the electric vehicle charging station is determined in the same way as the foregoing , except that is the travel friction function value of the electric vehicle charging station , which needs to be calculated using the relevant data of the electric vehicle charging station .

[0115] The calculation method of the potential total service capacity is exemplarily described below.

[0116] In some embodiments of the present application, the expression of the aforementioned potential total service capacity is as follows:

[0117] ;

[0118] wherein, represents the number of charging piles configured in the electric vehicle charging station ; represents the average rated power of a single charging pile, which can be obtained by obtaining the rated power of each charging pile from the specification parameters of each charging pile and then taking the average of the rated powers; represents the annual effective service hours (i.e., the annual effective service hours of a single candidate parking lot), which is determined according to the actual operation time of the candidate parking lot.

[0119] Step 16: using a genetic algorithm to optimize the electric vehicle charging station construction scheme to obtain an optimal electric vehicle charging station construction scheme that maximizes the fitness function and meets the preset construction constraints.

[0120] The electric vehicle charging station construction scheme includes the construction locations of multiple electric vehicle charging stations and the number of charging piles configured in each electric vehicle charging station. ​​​

[0121] The preset station building constraint is:

[0122] ;

[0123] ;

[0124] ;

[0125] wherein, represents the preset number of stations, represents the preset total number of charging piles, represents the maximum number of charging piles that can be configured for each electric vehicle charging station. That is, the number of candidate parking lots selected as electric vehicle charging stations must be , and the total number of charging piles of the electric vehicle charging stations is , the number of charging piles of each electric vehicle charging station cannot exceed its physical upper limit, and only when , , can be greater than 0.

[0126] In the optimization process of the genetic algorithm described above, a plurality of chromosomes in the initial population of the genetic algorithm correspond one-to-one to a plurality of electric vehicle charging station building schemes. It should be noted that the present application uses a traditional genetic algorithm to optimize the electric vehicle charging station building scheme, the difference being that the fitness function of the present application is an energy utilization rate function, and the final optimization goal is to obtain an optimal electric vehicle charging station building scheme that maximizes the fitness function (i.e., maximizes the energy utilization rate) and satisfies the preset station building constraint. The optimal electric vehicle charging station building scheme includes the building locations of electric vehicle charging stations (actually the locations of candidate parking lots) and the number of charging piles configured in each of the electric vehicle charging stations. According to the optimal electric vehicle charging station building scheme, the energy utilization rate can be maximized, achieving the effect of improving the energy utilization rate and making the energy fully utilized.

[0127] In some embodiments of the present application, the optimization process of the genetic algorithm described above includes the following steps:

[0128] Step 16.1, chromosome encoding: encode each potential electric vehicle charging station building scheme as a chromosome : , , if , it indicates that an electric vehicle charging station is selected to be built at candidate parking lot , if , indicates not to select the candidate parking lot establishing an electric vehicle charging station, indicates an electric vehicle charging station the number of charging piles configured. That is, in the optimization process, the energy utilization rate function in , are known quantities;

[0129] Step 16.2, fitness function: the constructed energy utilization rate function is taken as the fitness function;

[0130] Step 16.3, iterative evolution: the population is iteratively optimized through genetic operators such as selection, crossover and mutation, and higher fitness solutions are constantly sought;

[0131] Step 16.4, the algorithm stops after reaching the preset termination condition (such as the number of iterations), and outputs the chromosome with the highest fitness value found in the entire evolution process. After decoding, the chromosome is the optimal electric vehicle charging station construction scheme solved by the present application, which clearly indicates the optimal construction location and the number of charging piles that should be configured for each station.

[0132] The effect of the electric vehicle charging station site selection and capacity optimization method of the present application is exemplarily illustrated below with a specific example.

[0133] Taking 50 parking lots (i.e. candidate parking lots) in a city and a total of 451575 entry and exit records in a month as an example for ablation experiment.

[0134] Electric vehicle charging station site selection and capacity optimization based on structural entropy. After feature extraction and min-max normalization, in the structural entropy calculation, a weighted network graph is constructed, an entropy reduction iteration is constructed to build a coding tree, and the target height is normalized, the final structural entropy of each node is calculated , and is appended to the feature vector. Weighted K-means, K=5, using weighted Euclidean distance. Optimization algorithm, gravity model demand allocation, friction , =0.5. The objective function is maximized . Genetic algorithm (population 50, number of generations 100, crossover 0.8, mutation 0.2) is used for solution.

[0135] Under the above environmental data, multiple runs are taken for average, and it is found that the average energy utilization rate of the model using structural entropy is 0.4190, and the average energy utilization rate of the model not using structural entropy is 0.2738. The model using structural entropy is significantly better than the model not using structural entropy in terms of energy utilization rate, with an increase of about 53.2%. Therefore, the site selection and capacity optimization method of the present application is reasonable and feasible.

[0136] To sum up, the electric vehicle charging station site selection and capacity optimization method provided by the embodiments of the present application defines the demand area based on the real business data, geographic location, network topology and other multi-dimensional factors of the candidate parking lot, then constructs an energy utilization rate function based on the demand area, and takes the energy utilization rate function as the fitness function of the genetic algorithm, finally optimizes the charging station site and capacity configuration by using the genetic algorithm, obtains the site and capacity configuration under the maximum energy utilization rate, and achieves the effect of improving the energy utilization rate and fully utilizing the energy.

[0137] The above is the preferred embodiment of the present application. It should be noted that for those skilled in the art, without departing from the principles described in the present application, a number of improvements and refinements can be made, which should also be considered as the protection scope of the present application.

Claims

1. A structural entropy-based method for site selection and capacity optimization of electric vehicle charging stations, characterized in that, The method comprises the following steps: For each candidate parking lot in the plurality of candidate parking lots, an initial feature vector of the candidate parking lot is constructed based on business information and geographical position information of the candidate parking lot; The business information of the candidate parking lot is used to describe the parking situation in the candidate parking lot and the charging potential of the candidate parking lot in the low valley period of the power grid; A weighted network graph is constructed and initialized as a coding tree; the plurality of nodes in the weighted network graph correspond to the plurality of candidate parking lots one by one, each edge in the weighted network graph represents the distance between the initial feature vectors of the two candidate parking lots connected by the edge, and the plurality of nodes in the weighted network graph correspond to the plurality of leaf nodes in the coding tree one by one; Based on the coding tree, the structural entropy of each candidate parking lot is calculated, and based on the structural entropy and the initial feature vector of each candidate parking lot, an enhanced feature vector of each candidate parking lot is generated; Based on the enhanced feature vector of each candidate parking lot, the plurality of candidate parking lots are clustered to obtain a plurality of clusters, and each cluster is defined as a demand area; Based on the defined demand area, an energy utilization rate function is constructed, and the energy utilization rate function is used as the fitness function of a genetic algorithm; The genetic algorithm is used to optimize the electric vehicle charging station construction scheme to obtain an optimal electric vehicle charging station construction scheme that maximizes the fitness function and meets the preset construction constraints; the electric vehicle charging station construction scheme includes the construction locations of a plurality of electric vehicle charging stations and the number of charging piles configured for each electric vehicle charging station. 2.The method of claim 1, wherein, The business information includes daily average vehicle flow, average parking duration, long-stay vehicle proportion, and low valley charging potential index; candidate parking lot initial feature vector is: , longitude of the candidate parking lot , latitude of the candidate parking lot , average daily traffic volume of the candidate parking lot , average parking duration of the candidate parking lot , proportion of long-stay vehicles of the candidate parking lot , trough charging potential index of the candidate parking lot ; , represents the number of times of parking of the electric vehicle in the observation period, represents the total number of days included in the observation period; , denotes the exit time of the nth electric vehicle parking within the observation period, denotes the exit time of the nth electric vehicle parking within the observation period, denotes the exit time of the nth electric vehicle parking within the observation period, denotes the exit time of the nth electric vehicle parking within the observation period, , represents the number of times of parking of the electric vehicle when the parking duration is greater than the preset duration threshold in the observation period. , represents the number of times of parking of the electric vehicle in the low valley period of the power grid contained in the observation period, and the parking duration is greater than the preset duration threshold. 3.The method of claim 2, wherein, The structural entropy of each candidate parking lot is calculated based on the coding tree, which comprises: The structural entropy of the candidate parking lot is calculated by the following formula :​ ； in, Represents candidate parking lots in the coding tree The corresponding path from the leaf node to the root node. yes The subtree represented by a node on the tree, Represents the connection subtree The sum of the weights of all edges between an internal node and an external node. This represents the sum of the weights of all edges in the weighted network graph. , This represents the set of all nodes in the weighted network graph. Candidate parking lots in the weighted network graph corresponding nodes The weighting degree, Subtree The sum of the weighted degrees of all nodes in the array. Subtree The subtree represented by the parent node, Subtree The sum of the weighted degrees of all nodes in the equation. 4.The method of claim 1, wherein, The enhanced feature vector of each candidate parking lot is generated based on the structural entropy and the initial feature vector of each candidate parking lot, which comprises: For each candidate parking lot in the plurality of candidate parking lots, the structural entropy of the candidate parking lot is added to the initial feature vector of the candidate parking lot to obtain the enhanced feature vector of the candidate parking lot.

5. The method of claim 1, wherein, The plurality of candidate parking lots are clustered based on the enhanced feature vector of each candidate parking lot to obtain a plurality of clusters, which comprises: The weighted K-means clustering algorithm is used to cluster the enhanced feature vectors of the plurality of candidate parking lots to obtain a plurality of vector clusters; Each candidate parking lot corresponding to each vector cluster is taken as a cluster; In the clustering process, the distance between the enhanced feature vector of the candidate parking lot and the cluster center is calculated by the following formula :​​​ ； wherein, denotes the total dimension of the enhanced feature vector, denotes a predefined dynamic weight vector corresponding to the first dimension of the enhanced feature vector , denotes the value of the enhanced feature vector at the first dimension, denotes the value of the cluster center at the first dimension.

6. The method of claim 1, wherein, The energy utilization rate function is: ； wherein, represents the energy utilization rate, represents the predicted total charging amount, represents the potential total service capacity.

7. The method of claim 6, wherein, Predicted total charge The expression for the total charge is: ； wherein, represents the total number of candidate parking lots, , if represents the selection of the candidate parking lot to establish an electric vehicle charging station, if represents the non-selection of the candidate parking lot to establish an electric vehicle charging station, represents the candidate parking lot as an electric vehicle charging station the total predicted demand amount per year that can be captured, represents a preset valley charging value coefficient, represents the valley charging potential index of the candidate parking lot ; ； wherein, denotes the total number of demand zones, denotes the captured amount of charging demand from demand zones flowing to and being served by the electric vehicle charging station flowing to and being served by the electric vehicle charging station ； wherein, denotes the demand zone total annual charging demand potential, denotes the electric vehicle charging station the attractiveness function value, , denotes the electric vehicle charging station the number of charging piles configured, denotes the electric vehicle charging station the travel friction function value, , denotes the preset distance decay coefficient, denotes the demand zone the geographical distance or the average driving time estimated by a navigation software between the center and the electric vehicle charging station , denotes a set composed of a plurality of electric vehicle charging stations in the electric vehicle charging station construction scheme, denotes the electric vehicle charging station the attractiveness function value, denotes the electric vehicle charging station the travel friction function value.

8. The method of claim 7, wherein, Potential total service capacity The expression for the potential total service capacity is: ； wherein, representing electric vehicle charging stations the number of charging poles configured, representing the average rated power of a single charging pole, representing the annual effective service hours. 9.The method of claim 8, wherein, The preset construction constraints are: ； ； ； wherein, represents a preset number of stations, represents a preset total number of charging piles, represents an electric vehicle charging station the maximum number of charging piles that can be configured.

10. The method of claim 1, wherein, The plurality of chromosomes in the initial population of the genetic algorithm correspond to the plurality of electric vehicle charging station construction schemes one by one.

Images