Charging station site selection planning method and device fusing user behavior characteristics, electronic equipment and storage medium

Abstract

The invention discloses a charging station site selection planning method and device fusing user behavior characteristics, electronic equipment and a storage medium, and belongs to the field of charging station planning, and the method comprises the steps: obtaining electric vehicle user time sequence behavior characteristics of each grid in a to-be-decided region, extracting typical behavior characteristics through clustering, and carrying out the clustering of the typical behavior characteristics; and inputting the typical behavior characteristics into a charging demand prediction model, and generating time distribution and space distribution corresponding to each cluster. And based on the prediction result, constructing a multi-objective optimization function model aiming at maximizing the demand coverage rate, minimizing the total cost and the power grid load peak-valley difference, and setting high-demand area coverage, peak demand satisfaction and key class cluster guarantee constraint conditions. And solving the model under the constraints, generating an optimal solution set, and carrying out charging station site selection according to the optimal solution set. By implementing the method, the problem that the demand prediction precision is limited due to the lack of deep mining and modeling of the time sequence behavior characteristics of the electric vehicle user in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging station planning, and in particular to a charging station site selection and planning method, device, electronic device and storage medium integrating user behavior characteristics. Background Art

[0002] With the booming electric vehicle industry and its continued rise in market penetration, the development of supporting charging infrastructure has become a critical factor influencing its development. Charging station site selection and planning, as the top-level design of network layout, directly impacts user charging convenience, operator return on investment, and the stable operation of the regional power grid. Therefore, establishing a decision-making methodology that accurately and efficiently guides charging station planning and construction is of vital strategic importance for improving user experience, ensuring power grid security, and promoting the healthy and sustainable development of the electric vehicle industry.

[0003] However, existing charging station site selection methods are mostly based on static statistical data or simple traffic and population distribution information. They lack in-depth exploration and modeling of the temporal behavioral characteristics of electric vehicle users, resulting in limited demand forecasting accuracy. Furthermore, some methods fail to effectively integrate multiple optimization objectives and demand differentiation constraints, making it difficult to balance multi-dimensional indicators such as demand coverage, economic costs, and grid load balance, thus affecting the scientific nature and adaptability of the overall layout. Summary of the Invention

[0004] The embodiments of the present invention provide a charging station site selection and planning method, device, electronic device and storage medium that integrate user behavior characteristics, which can solve the problem in the prior art of lack of in-depth mining and modeling of the temporal behavior characteristics of electric vehicle users, resulting in limited demand forecasting accuracy.

[0005] An embodiment of the present invention provides a method for charging station site selection and planning that integrates user behavior characteristics, including:

[0006] Obtain the temporal behavior characteristics of electric vehicle users in each grid area in the decision-making area;

[0007] Clustering the temporal behavior characteristics of the electric vehicle user, and taking the centers of each cluster as the typical temporal behavior characteristics of the user, to generate several typical temporal behavior characteristics of the user;

[0008] Input each user's typical temporal behavior characteristics into a preset charging demand prediction model in sequence, so that the charging demand prediction model outputs the charging demand time distribution and charging demand spatial distribution of the corresponding cluster according to the current user's typical temporal behavior characteristics each time;

[0009] Based on the temporal and spatial distributions of charging demand for each cluster, a multi-objective optimization function model is constructed with the goals of maximizing demand coverage, minimizing total cost, and minimizing the peak-to-valley difference in grid load. The related constraints of the multi-objective optimization function model include high-demand area coverage constraints, peak demand satisfaction constraints, and key cluster protection constraints.

[0010] Under the conditions of the relevant constraints, the multi-objective optimization function model is solved to generate a set of optimal solutions for the decision of whether to build a charging station in each grid area;

[0011] The charging station site selection is performed in the area to be decided based on the optimal solution set.

[0012] Furthermore, the electric vehicle user's temporal behavior characteristics include the electric vehicle user's charging load sequence, charging times sequence, charging amount sequence, and charging start state of charge sequence within each preset time period.

[0013] Furthermore, the training of the charging demand prediction model includes:

[0014] Obtaining a user temporal behavior feature dataset; wherein the user temporal behavior feature dataset includes a plurality of electric vehicle user temporal behavior features and corresponding charging demand labels; the charging demand labels include a charging demand time distribution label and a charging demand spatial distribution label;

[0015] Randomly dividing the user temporal behavior feature dataset into several batches of training samples according to a preset batch size;

[0016] The training samples of each batch are input into the charging demand prediction model in sequence for iterative training until the preset training termination condition is reached; wherein, when the charging demand prediction model receives each batch of training samples, it outputs the corresponding predicted charging demand time distribution and predicted charging demand spatial distribution according to the temporal behavior characteristics of the electric vehicle users in the current batch; through a preset loss function, the loss function value is calculated based on the predicted charging demand time distribution and the corresponding charging demand time distribution label as well as the predicted charging demand spatial distribution and the charging demand spatial distribution label; and the parameters in the charging demand prediction model are updated according to the loss function value using a preset optimizer.

[0017] Furthermore, the multi-objective optimization function model is specifically:

[0018]

[0019] Among them, F(X) is the objective function vector; X is the vector of all location state variables y mThe site selection plan vector is constructed; f1(X) is the demand coverage function; f2(X) is the total cost function; f3(X) is the peak-to-valley difference function of the power grid load; M is the total number of geographical grids in the decision-making area; K is the total number of clusters; T is the total number of preset time periods; G m is the regional demand density of grid m; w i is the importance weight of cluster i; y m is the location state variable, when y m =1 means building a charging station in grid m. m =0 means no charging station is built in grid m; mi is the coverage efficiency of charging stations in grid m for cluster i; C con，m is the construction cost of building a charging station in grid m; C op，t is the unit operating cost of the charging station in period t; Q t，m is the predicted charging demand of grid m in time period t; Q max，m is the maximum service capacity of the charging stations planned and constructed in the grid m.

[0020] Furthermore, the coverage constraint of the high-demand area is specifically as follows:

[0021]

[0022] Where G th is the preset high demand threshold; α is the preset minimum coverage ratio of the high demand area; M high is the total number of high-demand grids;

[0023] The peak demand satisfies the constraints, specifically:

[0024]

[0025] The key cluster guarantee constraints are specifically:

[0026]

[0027] Where, β i is the minimum coverage set for key cluster j; K key is a collection of key clusters.

[0028] Based on the above method embodiments, the present invention provides corresponding device embodiments.

[0029] An embodiment of the present invention provides a charging station site selection and planning device that integrates user behavior characteristics, including: a user temporal behavior feature acquisition module, a feature clustering module, a charging demand prediction module, a multi-objective optimization function model construction module, a multi-objective optimization function model solution module, and a charging station site selection module;

[0030] The user temporal behavior feature acquisition module is used to acquire the temporal behavior features of electric vehicle users in each grid area in the area to be decided;

[0031] The feature clustering module is used to cluster the temporal behavior features of the electric vehicle users, and use the centers of each cluster as the typical temporal behavior features of the users to generate a number of typical temporal behavior features of the users;

[0032] The charging demand prediction module is used to sequentially input the typical temporal behavior characteristics of each user into a preset charging demand prediction model, so that the charging demand prediction model outputs the charging demand time distribution and charging demand spatial distribution of the corresponding cluster according to the current typical temporal behavior characteristics of the user;

[0033] The multi-objective optimization function model construction module is used to construct a multi-objective optimization function model with the goals of maximizing demand coverage, minimizing total cost, and minimizing peak-to-valley difference in grid load based on the temporal and spatial distributions of charging demand for each cluster, as well as related constraints for the multi-objective optimization function model; the related constraints include high-demand area coverage constraints, peak demand satisfaction constraints, and key cluster protection constraints;

[0034] The multi-objective optimization function model solving module is used to solve the multi-objective optimization function model under the conditions of the relevant constraints to generate a set of optimal solutions for the decision of whether to build a charging station in each grid area;

[0035] The charging station site selection module is used to select the site of the charging station in the area to be decided based on the optimal solution set.

[0036] Furthermore, the charging station site selection and planning device that integrates user behavior characteristics, the user time sequence behavior characteristics acquisition module, the electric vehicle user time sequence behavior characteristics include at least one of the electric vehicle user's charging load sequence, charging number sequence, charging amount sequence and charging starting state of charge sequence in each preset time period.

[0037] Furthermore, the charging station site selection and planning device integrating user behavior characteristics, the charging demand prediction module, and the training of the charging demand prediction model include:

[0038] Obtaining a user temporal behavior feature dataset; wherein the user temporal behavior feature dataset includes a plurality of electric vehicle user temporal behavior features and corresponding charging demand labels; the charging demand labels include a charging demand time distribution label and a charging demand spatial distribution label;

[0039] Randomly dividing the user temporal behavior feature dataset into several batches of training samples according to a preset batch size;

[0040] The training samples of each batch are input into the charging demand prediction model in sequence for iterative training until the preset training termination condition is reached; wherein, when the charging demand prediction model receives each batch of training samples, it outputs the corresponding predicted charging demand time distribution and predicted charging demand spatial distribution according to the temporal behavior characteristics of the electric vehicle users in the current batch; through a preset loss function, the loss function value is calculated based on the predicted charging demand time distribution and the corresponding charging demand time distribution label as well as the predicted charging demand spatial distribution and the charging demand spatial distribution label; and the parameters in the charging demand prediction model are updated according to the loss function value using a preset optimizer.

[0041] Based on the above method embodiment, the present invention provides a corresponding electronic device embodiment.

[0042] An embodiment of the present invention provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the charging station site selection planning method that integrates user behavior characteristics as described in any one of the above-mentioned method embodiments.

[0043] Based on the above method embodiment, the present invention provides a corresponding storage medium embodiment.

[0044] An embodiment of the present invention provides a storage medium having a computer program stored thereon, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the charging station site selection planning method that integrates user behavior characteristics as described in any one of the above method embodiments.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] An embodiment of the present invention provides a method, device, electronic device, and storage medium for charging station site selection planning that integrates user behavior characteristics. The method obtains the temporal behavior characteristics of electric vehicle users in each grid area within a pending decision area; performs cluster analysis on the temporal behavior characteristics, extracts the cluster centers of each cluster as the corresponding typical temporal behavior characteristics of the user; sequentially inputs the typical temporal behavior characteristics of each user into a preset charging demand prediction model to obtain the temporal and spatial distributions of charging demand for the corresponding cluster; based on the temporal and spatial distributions of charging demand for each cluster, constructs a multi-objective optimization function model with the goals of maximizing demand coverage, minimizing construction and operation costs, and minimizing peak-to-valley differences in grid load, and sets constraints including high-demand area coverage constraints, peak demand satisfaction constraints, and key cluster guarantee constraints; on the premise that the constraints are met, solves the multi-objective optimization function model to obtain a set of optimal solutions containing charging station construction decisions for each grid area; and based on the optimal solution set, completes the generation of a charging station site selection plan for the pending decision area.

[0047] The present invention obtains and clusters the temporal behavioral characteristics of electric vehicle users, extracts representative typical behavioral patterns, and improves the accuracy of charging demand prediction, thereby solving the problems of rough user modeling and inaccurate demand identification in the existing technology; further, the present invention constructs a multi-objective optimization model based on the prediction results, and sets constraints such as coverage, peak demand and key cluster protection, thereby effectively improving the comprehensive performance of site selection decisions in taking into account user demand differences, grid load balance and construction costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention provides a flowchart of a method for site selection and planning of charging stations that integrates user behavior characteristics, as provided in one embodiment of the present invention.

[0049] Figure 2 This is a structural diagram of a charging station site selection and planning device that integrates user behavior characteristics, provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0051] like Figure 1As shown, to address the problem in the prior art of limited demand forecasting accuracy due to a lack of in-depth exploration and modeling of electric vehicle user temporal behavior characteristics, an embodiment of the present invention provides a charging station site planning method that integrates user behavior characteristics, comprising at least the following steps:

[0052] Step S1, obtaining the temporal behavior characteristics of electric vehicle users in each grid area in the decision-making area;

[0053] In a preferred embodiment, the electric vehicle user's temporal behavior characteristics include the electric vehicle user's charging load sequence, charging times sequence, charging amount sequence, and charging start state of charge sequence within each preset time period.

[0054] Specifically, the step of obtaining the temporal behavior characteristics of electric vehicle users in each grid area in the area to be decided first requires collecting and organizing raw data that can reflect the historical charging activities of electric vehicle users. In order to construct the temporal characteristics, the analysis period such as a day or a week can be divided into multiple preset time periods, for example, divided into hours. Subsequently, for each user, the specific charging behavior parameters of the user in each preset time period are counted. For example, by summarizing the charging power and duration of the user in a certain time period, the charging load of the time period can be obtained; by counting the number of charging events, the number of charging times can be obtained; and the battery state of charge (SOC) at the beginning of the charging behavior in the time period is recorded. Finally, the above parameters calculated for each user in all preset time periods are arranged in chronological order to form the temporal behavior characteristics of the electric vehicle user that can reflect their dynamic behavior habits.

[0055] Specifically, the electric vehicle user's temporal behavior characteristics include the user's charging load sequence, charging frequency sequence, charging amount sequence, and charging initial state of charge sequence within each preset time period. The temporal behavior characteristics obtained in this way can express user behavior habits in the form of structured and quantifiable data sequences, laying a solid data foundation for subsequent clustering algorithms to accurately identify user groups and characterize differentiated needs.

[0056] Step S2: clustering the temporal behavior characteristics of the electric vehicle users, and taking the centers of the clusters as the typical temporal behavior characteristics of the users, to generate a number of typical temporal behavior characteristics of the users;

[0057] Specifically, this step aims to automatically identify and segment user groups with similar behavior patterns from massive amounts of unlabeled user behavior data. Specifically, a clustering algorithm from unsupervised learning can be used. This algorithm calculates the similarity or distance between different users' temporal behavior feature vectors and groups users with similar behavior patterns into the same cluster.

[0058] After clustering is complete, several distinct user clusters are generated. To quantify and characterize the behavior of each cluster, the mathematical average of the temporal behavior characteristics of all users within that cluster can be calculated. This average is the cluster center (center of gravity) for that cluster. This center of gravity serves as the typical temporal behavior characteristic of that user group, representing the most common and core behavioral patterns of the entire group. This step allows the complex and diverse behavior of individual users to be summarized into a limited number of representative typical behavioral patterns, providing clear and effective input for subsequent differentiated demand forecasting for different user groups.

[0059] Step S3: Input each user's typical temporal behavior characteristics into a preset charging demand prediction model in sequence, so that the charging demand prediction model outputs the charging demand time distribution and charging demand spatial distribution of the corresponding cluster according to the current user's typical temporal behavior characteristics each time;

[0060] After generating several typical temporal behavior features of users representing different user groups, a specific embodiment of the present invention uses these typical features as input for refined demand forecasting. The forecasting process is completed by a pre-trained charging demand forecasting model. This model is a data-driven model. It has mastered the mapping relationship between abstract behavior patterns and specific demand distributions by learning a large number of historical temporal behavior features and their corresponding real-time and spatial distributions of charging demand. During execution, each user's typical temporal behavior feature is input into the forecasting model in turn. The model will parse the typical behavior pattern of the current input based on the knowledge it has learned internally, and calculate and output the charging demand time distribution and charging demand spatial distribution corresponding to the pattern in the entire area to be decided.

[0061] The temporal distribution of charging demand can be represented as a sequence indexed by time periods, whose values ​​reflect the intensity of charging demand for users in that cluster at different times. The spatial distribution of charging demand can be represented as a data map indexed by geographic grids, whose values ​​reflect the intensity of charging demand for users in that cluster at different geographic locations. Through this process, the present invention can present the potential demand of different user clusters in a quantified and structured temporal and spatial distribution, providing refined data input for the subsequent construction of a multi-objective optimization model that accounts for user heterogeneity.

[0062] In a preferred embodiment, the training of the charging demand prediction model includes:

[0063] Obtaining a user temporal behavior feature dataset; wherein the user temporal behavior feature dataset includes a plurality of electric vehicle user temporal behavior features and corresponding charging demand labels; the charging demand labels include a charging demand time distribution label and a charging demand spatial distribution label;

[0064] Randomly dividing the user temporal behavior feature dataset into several batches of training samples according to a preset batch size;

[0065] The training samples of each batch are input into the charging demand prediction model in sequence for iterative training until the preset training termination condition is reached; wherein, when the charging demand prediction model receives each batch of training samples, it outputs the corresponding predicted charging demand time distribution and predicted charging demand spatial distribution according to the temporal behavior characteristics of the electric vehicle users in the current batch; through a preset loss function, the loss function value is calculated based on the predicted charging demand time distribution and the corresponding charging demand time distribution label as well as the predicted charging demand spatial distribution and the charging demand spatial distribution label; and the parameters in the charging demand prediction model are updated according to the loss function value using a preset optimizer.

[0066] In order for the charging demand prediction model to have the ability to infer the specific demand distribution from the typical behavior patterns of users, it needs to be trained in advance. The training process first needs to construct a user time-series behavior feature dataset. This dataset contains a large number of historical samples, each of which consists of an "input" and a "label". Among them, the "input" is the time-series behavior characteristics of a user that actually occurred in history, and the "label" is the time distribution and spatial distribution of charging demand actually generated by this type of user at that time, corresponding to the behavior characteristics. This label provides the model with the "correct answer" for learning. In order to improve training efficiency and stability, this huge dataset is usually randomly divided into several batches with smaller data volumes for subsequent batch-by-batch iterative training.

[0067] After training begins, these batches of training samples are sequentially fed into the charging demand prediction model. Each time the model receives a batch of training samples, it makes a prediction based on the user's temporal behavior characteristics in the samples, outputting its own "guessed" temporal and spatial distributions of charging demand. Subsequently, a preset loss function calculates a loss function value that quantifies the "error" of the prediction by comparing the difference between the model's predicted distribution and the true distribution of the training labels. Next, a preset optimizer fine-tunes the model's internal parameters based on this loss function value, with the goal of reducing the error of the next prediction. This "prediction-error calculation-parameter adjustment" process is repeated iteratively until the preset training termination condition is reached. Through this training process, the prediction model is able to learn the deep connection between user behavior and charging demand, providing the ability to make accurate predictions in practical applications.

[0068] Step S4: Based on the temporal and spatial distributions of charging demand for each cluster, a multi-objective optimization function model is constructed with the goals of maximizing demand coverage, minimizing total cost, and minimizing the peak-to-valley difference in grid load, as well as related constraints for the multi-objective optimization function model; the related constraints include high-demand area coverage constraints, peak demand satisfaction constraints, and key cluster protection constraints;

[0069] After obtaining the refined charging demand distribution differentiated by cluster, the specific embodiment of the present invention will construct a multi-objective optimization function model based on this to guide the final site selection decision. The core idea of ​​this model is to simultaneously seek the optimal balance solution for three interrelated and even conflicting objectives. The first goal is to maximize the demand coverage rate, which aims to enable the constructed charging stations to serve the charging needs of users of different clusters to the greatest extent from the perspective of social benefits; the second goal is to minimize the total cost, which aims to comprehensively consider the construction cost and subsequent operating costs of the charging station from the perspective of economic benefits; the third goal is to minimize the peak-to-valley difference in grid load, which aims to balance the load impact of the charging network on the grid within a day from the perspective of grid benefits, and ensure the stable operation of the regional power grid.

[0070] In order to make the optimization results meet the engineering and service requirements of the real world, the model also sets several key related constraints. Among them, the high-demand area coverage constraint is used to ensure that the site selection plan must cover a certain proportion of the hot spots with the strongest demand; the peak demand satisfaction constraint ensures that the capacity of each planned and constructed charging station is sufficient to cope with the highest peak charging demand within its service range; and the key cluster protection constraint provides a minimum service coverage bottom line for those identified user clusters with special value or rigid demand to prevent them from being ignored in the global optimization. By constructing such a comprehensive optimization model that includes multi-dimensional objectives and multiple constraints, the present invention can transform a complex site selection problem into a mathematical problem that can be accurately solved, providing a model basis for generating a scientific and reasonable site selection plan that takes into account the interests of all parties.

[0071] In a preferred embodiment, the multi-objective optimization function model is specifically:

[0072]

[0073] Among them, F(X) is the objective function vector; X is the vector of all location state variables y m The site selection plan vector is constructed; f1(X) is the demand coverage function; f2(X) is the total cost function; f3(X) is the peak-to-valley difference function of the power grid load; M is the total number of geographical grids in the decision-making area; K is the total number of clusters; T is the total number of preset time periods; G m is the regional demand density of grid m; w iis the importance weight of cluster i; y m is the location state variable, when y m =1 means building a charging station in grid m. m =0 means no charging station is built in grid m; mi is the coverage efficiency of charging stations in grid m for cluster i; C con，m is the construction cost of building a charging station in grid m; C op，t is the unit operating cost of the charging station in period t; Q t，m is the predicted charging demand of grid m in time period t; Q max，m is the maximum service capacity of the charging stations planned and constructed in the grid m.

[0074] Specifically, to achieve scientific site selection, it is necessary to construct a mathematical model that can be solved by computer based on the time distribution and spatial distribution of charging demand of various clusters obtained in the previous step. This model is a multi-objective optimization function model. Its core is to define the unknown quantities that need to be solved, namely the decision variables, and the known quantities that serve as model input parameters. In this solution, the decision variable is the site selection plan vector X representing "which grid to build a station or not", which is composed of a series of site selection state variables y m The composition is the answer that the optimization algorithm ultimately needs to find. The known quantities in the model include the demand data for each spatiotemporal unit predicted in the previous steps, the weights of various clusters obtained through analysis, and parameters such as the preset cost and service radius.

[0075] It should be noted that in this model, some parameters that are known quantities need to be determined through further calculations. For example, Q t，m represents the predicted charging demand of grid m at time period t, which is obtained by summing the demands of all clusters predicted in the previous step in this space-time unit. It is calculated as follows:

[0076]

[0077] Where Q t，m，i is the predicted charging demand of cluster i in time period t and grid m.

[0078] Then, the total demand of the grid in all time periods is summed up again to obtain the final regional demand density G of the grid. m , and its final calculation method is:

[0079]

[0080] This G m The higher the value, the greater the comprehensive charging demand in the grid area, and the higher the priority in site selection decision-making.

[0081] Similarly, parameter δ mi It represents the coverage efficiency of grid m for cluster i. It is not directly input, but is calculated based on the preset service radius R of the charging station and the user's travel distance d. mi The calculation method is: when d mi When ≤R, δ mi =1, indicating effective coverage; otherwise, δ mi = 0. By clearly defining and calculating the decision variables and known quantities, the present invention rigorously transforms the complex, multi-dimensional site selection problem into a standard multi-objective optimization problem, making it possible to automatically solve the subsequent algorithm.

[0082] In the multi-objective optimization function model, the parameter d mi Represents the distance between the representative activity center of a certain type of user (i.e., cluster i) and a certain geographic grid (grid m). The calculation of this parameter is entirely dependent on the data obtained and analyzed in the early stage of the present invention. The specific process is as follows:

[0083] In the embodiment of the present invention, we first obtain the temporal behavior characteristics of electric vehicle users including the geographical location where charging occurs. After identifying different user "clusters i" through clustering algorithms, we can analyze all historical charging events belonging to the cluster and find the centroid of the geographical location where these events occurred. Therefore, d mi The value of is defined as the geographic distance between the representative geographic activity center of cluster i and the center of grid m where the candidate charging station is located. In this way, we objectively and quantitatively define the geographic relationship between user groups and potential charging station locations using only charging behavior data, providing effective input for subsequent coverage efficiency calculations.

[0084] The importance weight w of cluster i i It aims to characterize the priority or importance of different user clusters in the overall site selection plan. Its setting can be based on the contribution of each user cluster to the total regional demand, or combined with specific strategies (for example, giving higher weights to clusters represented by high-frequency charging users, users of specific vehicle types, or active users at key urban transportation nodes). In practical applications, w i Preliminary quantification can be carried out based on the proportion of each type of user in the total charging demand, or adjustments and optimization can be made through expert experience and policy guidance to ensure that the charging needs of key user groups are fully guaranteed.

[0085] For each candidate grid m in the area to be decided, a fixed cost value for building a charging station there is preset during model construction. This value is derived from a comprehensive evaluation of various external factors, such as the land lease or purchase costs for the grid, the estimated costs of infrastructure construction (such as charging stations, distribution facilities, civil engineering, etc.), and related approval and licensing fees. These costs are determined before the site selection decision model is run and remain unchanged throughout the optimization process. Therefore, when evaluating a site selection plan, the model directly references these preset construction cost values ​​for calculation.

[0086] Similarly, for each preset time period t, the unit operating cost of the charging station is a predetermined parameter. This value reflects the variable cost required to provide each unit of charging service at the charging station in the grid during a specific period of time, such as the cost per kilowatt-hour or the operating cost per hour. The components of operating costs may include electricity costs (taking into account time-of-use electricity prices), labor, equipment maintenance, and site management fees. Since factors such as electricity prices may vary with time periods, they are defined as time-related parameters, but these values ​​that vary with time periods are also predetermined before the model is run. When evaluating the operating costs of a site selection plan, the model will directly use the corresponding preset unit operating cost values ​​for cumulative calculation based on the selected charging station location and operating period.

[0087] The maximum service capacity Q of the charging stations planned and constructed within the grid m max，m It represents the maximum charging service capacity that the planned charging stations can provide simultaneously or in a specific period of time within the grid m. This depends on the number of charging piles in the charging station, the power of each pile, and the total power supply capacity of the charging station. max，m The setting needs to comprehensively consider factors such as the predicted charging demand peak, land availability, and grid access capabilities of the grid area to ensure that the charging station can meet the maximum demand load in the service area after it is built, while avoiding waste of resources.

[0088] In a preferred embodiment, the high-demand area coverage constraint is specifically:

[0089]

[0090] Where G th is the preset high demand threshold; α is the preset minimum coverage ratio of the high demand area; M high is the total number of high-demand grids;

[0091] The high-demand area coverage constraint is designed to ensure that the final charging station site selection plan can provide sufficient service coverage for the hotspot areas with the strongest demand, ensuring that users in high-demand areas can easily find charging facilities. this a critical value used to define which grid areas are considered “high demand areas”. Its setting is usually based on statistical analysis of historical charging demand data or forecast demand data. For example, the regional demand density G th Grids that rank in the top 20% or reach a certain demand volume are defined as high-demand grids. This threshold is an empirical value or a strategic value determined through preliminary data analysis.

[0092] The minimum high-demand area coverage ratio, α, represents the minimum percentage of high-demand grid areas that must be covered in the final site selection plan. For example, if α = 0.8, this means that at least 80% of high-demand grid areas must be covered by charging stations. This ratio is typically determined by planning objectives, resource constraints, and quality of service requirements, and is a pre-set policy or planning indicator.

[0093] The total number of high-demand grids M high Refers to the total number of grids identified as high demand in the decision area. It is used to determine the high demand threshold G th Afterwards, by traversing all grids G m The value is obtained by counting the number of grids that meet the conditions.

[0094] The peak demand satisfies the constraints, specifically:

[0095]

[0096] The key cluster guarantee constraints are specifically:

[0097]

[0098] Where, β i is the minimum coverage set for key cluster j; K key is a collection of key clusters.

[0099] The set of key clusters K key This set includes all user clusters identified as "critical." The definition of critical clusters is typically determined based on external policies. For example, in one embodiment, clusters corresponding to user groups with rigid charging requirements and high importance, such as ride-hailing drivers, logistics fleets, and emergency vehicles, can be included in this set.

[0100] In another embodiment, for each cluster i, its total predicted charging demand in all grids and all time periods is calculated, and clusters whose total demand reaches a preset threshold or ranks in the top N are defined as key clusters.

[0101] In another embodiment, the time distribution of charging demand for each cluster is analyzed to identify clusters with high demand during daily or weekly peak charging periods. These clusters have a significant impact on the peak-to-valley variation in grid load, and defining them as key clusters helps optimize grid load balancing.

[0102] It can also be combined with the regional demand density G of the grid m , analyze the spatial distribution of charging demand of each cluster. If the user behavior of a cluster is mainly concentrated in G m The hotspot grid area with a higher value is more important for meeting the coverage target of high-demand areas and can be defined as a key cluster.

[0103] This set is predetermined after clustering and user behavior analysis, combined with actual operational needs or policy orientations.

[0104] Step S5: Solve the multi-objective optimization function model under the relevant constraints to generate a set of optimal solutions for the decision of whether to build a charging station in each grid area;

[0105] After constructing an optimization model that includes multiple objective functions and a series of constraints, the next key step is to solve the model. This solution process aims to find a set of optimal site selection options, that is, to determine which grid areas should be used to build charging stations. Specifically, this step solves the multi-objective optimization problem by applying various optimization algorithms. Because the model constructed in the present invention involves multiple conflicting objectives (for example, maximizing demand coverage may result in increased total cost, while minimizing total cost may sacrifice some demand coverage), it is necessary to use algorithms specifically suitable for multi-objective optimization (such as the NSGA-II algorithm). These algorithms iteratively search and weigh the relationships between various objectives while satisfying all preset constraints, thereby generating a set of non-inferior solutions (Pareto optimal solutions). Each non-inferior solution represents a feasible site selection option that cannot be improved on any objective without compromising the others. Ultimately, from this set of optimal solutions, the charging station site selection option that best meets actual needs and planning strategies is selected. By solving this model, it is possible to efficiently and scientifically screen out the optimal or near-optimal charging station layout option that takes into account multiple factors from a large number of possible site selection combinations.

[0106] Step S6: Select the location of the charging station in the area to be decided based on the optimal solution set.

[0107] Specifically, after obtaining the optimal solution set encompassing the decision on whether to build charging stations in each grid area, the final step is to select charging station sites within the decision-making area based on this solution set. Because multi-objective optimization typically generates a set of non-inferior solutions (Pareto optimal solutions) rather than a single optimal solution, a final site selection scheme must be selected from this set of optimal solutions. This selection process can be based on the decision-maker's pre-set preferences or additional evaluation criteria. For example, while ensuring basic needs coverage, options with lower costs or less impact on grid load can be prioritized. Alternatively, the final decision can be made based on non-modeled factors such as actual urban planning, land availability, and grid access conditions. Once the final site selection scheme is determined, it is clear which grids within the decision-making area will host charging stations, as well as the expected service range and capacity of each charging station. This step transforms the abstract mathematical model results into concrete engineering implementation plans, providing clear guidance for the construction of charging infrastructure. Ultimately, a scientific and reasonable charging station layout plan can be generated that balances user needs, economic costs, and grid load.

[0108] Based on the above method embodiments, the present invention provides corresponding device embodiments.

[0109] like Figure 2 As shown, an embodiment of the present invention provides a charging station site selection and planning device that integrates user behavior characteristics, including: a user time series behavior feature acquisition module, a feature clustering module, a charging demand prediction module, a multi-objective optimization function model construction module, a multi-objective optimization function model solution module, and a charging station site selection module;

[0110] The user temporal behavior feature acquisition module is used to acquire the temporal behavior features of electric vehicle users in each grid area in the area to be decided;

[0111] The feature clustering module is used to cluster the temporal behavior features of the electric vehicle users, and use the centers of each cluster as the typical temporal behavior features of the users to generate a number of typical temporal behavior features of the users;

[0112] The charging demand prediction module is used to sequentially input the typical temporal behavior characteristics of each user into a preset charging demand prediction model, so that the charging demand prediction model outputs the charging demand time distribution and charging demand spatial distribution of the corresponding cluster according to the current typical temporal behavior characteristics of the user;

[0113] The multi-objective optimization function model construction module is configured to construct a multi-objective optimization function model and related constraints of the multi-objective optimization function model according to the charging demand time distribution and the charging demand space distribution of each cluster, the multi-objective optimization function model aiming to maximize demand coverage, minimize total cost, and minimize peak-valley difference of power grid load; the related constraints include high-demand area coverage rate constraint, peak demand satisfaction constraint, and key cluster guarantee constraint;

[0114] The multi-objective optimization function model solving module is configured to solve the multi-objective optimization function model under the condition of the related constraints, and generate an optimal solution set containing decisions of whether to build a charging station in each grid area;

[0115] The charging station site selection module is configured to perform charging station site selection in a to-be-decided area according to the optimal solution set.

[0116] In a preferred embodiment, the charging station site selection planning device fusing user behavior features, the user time sequence behavior feature acquisition module, and the electric vehicle user time sequence behavior features include charging load sequence, charging frequency sequence, charging amount sequence, and charging start state of charge sequence of the electric vehicle user in each preset time period.

[0117] In a preferred embodiment, the charging station site selection planning device fusing user behavior features, the charging demand prediction module, and the training of the charging demand prediction model includes:

[0118] An user time sequence behavior feature dataset is obtained; the user time sequence behavior feature dataset includes a plurality of electric vehicle user time sequence behavior features and corresponding charging demand labels; the charging demand labels include charging demand time distribution labels and charging demand space distribution labels;

[0119] The user time sequence behavior feature dataset is randomly divided into a plurality of batches of training samples according to a preset batch size;

[0120] Each batch of training samples is sequentially input into the charging demand prediction model for iterative training until a preset training termination condition is reached; when receiving each batch of training samples, the charging demand prediction model outputs corresponding predicted charging demand time distribution and predicted charging demand space distribution according to the electric vehicle user time sequence behavior features in the current batch; a loss function value is calculated by a preset loss function according to the predicted charging demand time distribution and the corresponding charging demand time distribution label and the predicted charging demand space distribution and the charging demand space distribution label; and the parameters in the charging demand prediction model are updated according to the loss function value using a preset optimizer.

[0121] It should be noted that the above-described embodiments of the device correspond to the above-described embodiments of the application, and can realize any one of the above-described methods for planning the location of a charging station by fusing user behavior features. In addition, the above-described embodiments of the device are merely illustrative, and the modules described as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. In addition, the connection between the modules in the device embodiments provided by the present application indicates that there is a communication connection between them, which can be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0122] On the basis of the above-described method embodiments of the application, an electronic device embodiment is provided.

[0123] An embodiment of the present application provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, when the processor executes the computer program, realizing the method for planning the location of a charging station by fusing user behavior features according to any one of the embodiments of the present application, or when the processor executes the computer program, realizing the functions of the modules in the above-described device embodiments.

[0124] For example, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the terminal device.

[0125] The terminal device can be a desktop computer, a notebook computer, a palm computer, a cloud server, and other computing devices. The terminal device can include, but is not limited to, a processor and a memory.

[0126] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0127] The memory can be used to store the computer programs and / or modules, and the processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created based on the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0128] Based on the above method embodiment, the present invention provides a corresponding storage medium embodiment;

[0129] Another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute any of the above-mentioned charging station site selection planning methods that integrate user behavior characteristics of the present invention.

[0130] The storage medium is a computer-readable storage medium, and the computer program includes computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunications signal, and a software distribution medium.

[0131] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0132] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A charging station site selection and planning method integrating user behavior characteristics, characterized in that: include: Obtain the temporal behavior characteristics of electric vehicle users in each grid area in the decision-making area; Clustering the temporal behavior characteristics of the electric vehicle user, and taking the centers of each cluster as the typical temporal behavior characteristics of the user, to generate several typical temporal behavior characteristics of the user; Input each user's typical temporal behavior characteristics into a preset charging demand prediction model in sequence, so that the charging demand prediction model outputs the charging demand time distribution and charging demand spatial distribution of the corresponding cluster according to the current user's typical temporal behavior characteristics each time; Based on the temporal and spatial distributions of charging demand for each cluster, a multi-objective optimization function model is constructed with the goals of maximizing demand coverage, minimizing total cost, and minimizing the peak-to-valley difference in grid load. The related constraints of the multi-objective optimization function model include high-demand area coverage constraints, peak demand satisfaction constraints, and key cluster protection constraints. Under the conditions of the relevant constraints, the multi-objective optimization function model is solved to generate a set of optimal solutions for the decision of whether to build a charging station in each grid area; The charging station site selection is performed in the area to be decided based on the optimal solution set.

2. The method for site selection and planning of charging stations integrating user behavior characteristics according to claim 1, characterized in that: The electric vehicle user's temporal behavior characteristics include the electric vehicle user's charging load sequence, charging times sequence, charging amount sequence, and charging start state of charge sequence within each preset time period.

3. The method for site selection and planning of charging stations integrating user behavior characteristics according to claim 2, characterized in that: The training of the charging demand prediction model includes: Obtaining a user temporal behavior feature dataset; wherein the user temporal behavior feature dataset includes a plurality of electric vehicle user temporal behavior features and corresponding charging demand labels; the charging demand labels include a charging demand time distribution label and a charging demand spatial distribution label; Randomly dividing the user temporal behavior feature dataset into several batches of training samples according to a preset batch size; The training samples of each batch are input into the charging demand prediction model in sequence for iterative training until the preset training termination condition is reached; wherein, when the charging demand prediction model receives each batch of training samples, it outputs the corresponding predicted charging demand time distribution and predicted charging demand spatial distribution according to the temporal behavior characteristics of the electric vehicle users in the current batch; through a preset loss function, the loss function value is calculated based on the predicted charging demand time distribution and the corresponding charging demand time distribution label as well as the predicted charging demand spatial distribution and the charging demand spatial distribution label; and the parameters in the charging demand prediction model are updated according to the loss function value using a preset optimizer.

4. The method for site selection and planning of charging stations integrating user behavior characteristics according to claim 3, characterized in that: The multi-objective optimization function model is specifically: Among them, F(X) is the objective function vector; X is the vector of all location state variables y m The site selection plan vector is constructed; f1(X) is the demand coverage function; f2(X) is the total cost function; f3(X) is the peak-to-valley difference function of the power grid load; M is the total number of geographical grids in the decision-making area; K is the total number of clusters; T is the total number of preset time periods; G m is the regional demand density of grid m; w i is the importance weight of cluster i; y m is the location state variable, when y m =1 means building a charging station in grid m. m =0 means no charging station is built in grid m; mi is the coverage efficiency of charging stations in grid m for cluster i; C con，m is the construction cost of building a charging station in grid m; C op，t is the unit operating cost of the charging station in period t; Q t，m is the predicted charging demand of grid m in time period t; C max，m is the maximum service capacity of the charging stations planned and constructed in the grid m.

5. The method for site selection and planning of charging stations integrating user behavior characteristics according to claim 4, characterized in that: The high-demand area coverage constraint is specifically: Where G th is the preset high demand threshold; α is the preset minimum coverage ratio of the high demand area; M high is the total number of high-demand grids; The peak demand satisfies the constraints, specifically: The key cluster guarantee constraints are specifically: Where, β i is the minimum coverage set for key cluster j; K key is a collection of key clusters.

6. A charging station site selection and planning device integrating user behavior characteristics, characterized in that: include: User temporal behavior feature acquisition module, feature clustering module, charging demand prediction module, multi-objective optimization function model construction module, multi-objective optimization function model solution module and charging station location selection module; The user temporal behavior feature acquisition module is used to acquire the temporal behavior features of electric vehicle users in each grid area in the area to be decided; The feature clustering module is used to cluster the temporal behavior features of the electric vehicle users, and use the centers of each cluster as the typical temporal behavior features of the users to generate a number of typical temporal behavior features of the users; The charging demand prediction module is used to sequentially input the typical temporal behavior characteristics of each user into a preset charging demand prediction model, so that the charging demand prediction model outputs the charging demand time distribution and charging demand spatial distribution of the corresponding cluster according to the current typical temporal behavior characteristics of the user; The multi-objective optimization function model construction module is used to construct a multi-objective optimization function model with the goals of maximizing demand coverage, minimizing total cost, and minimizing peak-to-valley difference in grid load based on the temporal and spatial distributions of charging demand for each cluster, as well as related constraints for the multi-objective optimization function model; the related constraints include high-demand area coverage constraints, peak demand satisfaction constraints, and key cluster protection constraints; The multi-objective optimization function model solving module is used to solve the multi-objective optimization function model under the conditions of the relevant constraints to generate a set of optimal solutions for the decision of whether to build a charging station in each grid area; The charging station site selection module is used to select the site of the charging station in the area to be decided based on the optimal solution set.

7. The charging station site selection and planning device integrating user behavior characteristics according to claim 6, characterized in that: The user time sequence behavior feature acquisition module includes the electric vehicle user's time sequence behavior feature, including the electric vehicle user's charging load sequence, charging times sequence, charging amount sequence and charging start state of charge sequence within each preset time period.

8. The charging station site selection and planning device integrating user behavior characteristics according to claim 7, characterized in that: The charging demand prediction module, the training of the charging demand prediction model, includes: Obtaining a user temporal behavior feature dataset; wherein the user temporal behavior feature dataset includes a plurality of electric vehicle user temporal behavior features and corresponding charging demand labels; the charging demand labels include a charging demand time distribution label and a charging demand spatial distribution label; Randomly dividing the user temporal behavior feature dataset into several batches of training samples according to a preset batch size; The training samples of each batch are input into the charging demand prediction model in sequence for iterative training until the preset training termination condition is reached; wherein, when the charging demand prediction model receives each batch of training samples, it outputs the corresponding predicted charging demand time distribution and predicted charging demand spatial distribution according to the temporal behavior characteristics of the electric vehicle users in the current batch; through a preset loss function, the loss function value is calculated based on the predicted charging demand time distribution and the corresponding charging demand time distribution label as well as the predicted charging demand spatial distribution and the charging demand spatial distribution label; and the parameters in the charging demand prediction model are updated according to the loss function value using a preset optimizer.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for charging station site selection and planning integrating user behavior characteristics as described in any one of claims 1 to 5 is implemented.

10. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the charging station site selection and planning method integrating user behavior characteristics as described in any one of claims 1 to 5.

Images