A method for assessing the charging load capacity of electric vehicles in power distribution networks, taking into account spatiotemporal profiles.

Abstract

This invention relates to a method for assessing the load carrying capacity of electric vehicle charging in distribution networks, considering spatiotemporal profiling. Belonging to the field of distribution network technology, this scheme combines key parameters such as charging pile type, battery capacity, and functional area attributes to construct multi-type charging load profiles and an improved K-means clustering basic load profile. An improved GA optimization algorithm is used to separate the load of existing charging stations, calculating the available capacity under the premise of transformer safety constraints. Kriging interpolation is used to fit the load curve of newly added charging stations, simultaneously calculating their available capacity. Based on the distribution network topology, capacity matching is verified level by level from 10kV to 110kV, and the allocation of mismatched capacity is optimized using an improved entropy weight-TOPSIS multi-criteria algorithm. This invention overcomes the problems of inaccurate profiling, imprecise load separation, and empirical capacity allocation in traditional methods, enabling precise adaptation of the distribution network to charging loads and improving its carrying capacity.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network technology, specifically relating to a method for assessing the load carrying capacity of electric vehicle charging in power distribution networks that considers spatiotemporal profiles. Background Technology

[0002] With the rapid increase in the penetration rate of new energy vehicles, the scale of charging pile construction continues to expand, and charging stations have become one of the core load nodes of the power distribution network. Charging loads exhibit significant multi-dimensional differences, with their size and distribution influenced by multiple factors such as charging pile type (DC fast charging / AC slow charging), vehicle battery capacity, functional area attributes (residential areas, commercial areas, highway service areas), and charging start times (peak / valley / flat periods). Under different scenarios, charging power, charging duration, and load time distribution vary significantly. For example, charging demand in residential areas is concentrated during off-peak hours at night, charging behavior in commercial areas is random and dispersed, and highway service areas primarily rely on short-term charging. Traditional power distribution network planning and design have not fully considered these dynamic and complex charging loads, and the existing capacity assessment system is ill-suited to new energy charging scenarios. This leads to safety risks such as transformer overload and insufficient line margin during power distribution network operation, or resource idleness due to inaccurate capacity assessment, seriously affecting the operational safety of the power distribution network and the efficiency of charging service supply.

[0003] Existing research still has significant shortcomings in the areas of load adaptation and capacity enhancement for distribution networks, and these shortcomings are closely related to core technical aspects such as load assessment and capacity optimization. At the load assessment level, traditional methods have obvious weaknesses: the static capacity margin method ignores the randomness of charging loads and the dynamic operating characteristics of the power grid, while the probabilistic method suffers from the "curse of dimensionality" due to computational complexity. Furthermore, neither method fully integrates key differentiated parameters such as charging pile type and functional area attributes, leading to a disconnect between assessment results and actual carrying capacity. Regarding data support, most studies rely solely on historical charging pile data for modeling, failing to effectively integrate multi-source information such as distributed photovoltaic power output and user travel chains, making it difficult to accurately characterize the essential load characteristics under various scenarios. Significant problems exist in capacity matching and scheduling: capacity allocation between upper and lower voltage levels relies heavily on empirical judgment, lacking a multi-criteria optimization model that covers load fluctuations and line margins; for special scenarios such as "tidal charging" in highway service areas, there are neither suitable dynamic capacity allocation schemes nor effective solutions for upgrading aging transformers to support high-power charging piles, making it difficult to balance power grid safety and charging service quality. Summary of the Invention

[0004] To address the aforementioned technical issues, this invention provides a method for assessing the load carrying capacity of electric vehicle charging in distribution networks that considers spatiotemporal profiles. By integrating multiple key parameters such as charging pile type and functional area attributes, it accurately constructs load profiles, separates charging loads, fits the load curves of newly added stations, and optimizes capacity allocation. This overcomes the shortcomings of traditional methods, such as inaccurate profiles, imprecise load separation, and experience-based capacity allocation, thus meeting the needs of safe operation of distribution networks and efficient supply of charging services.

[0005] The method for assessing the charging load capacity of electric vehicles in a power distribution network, which considers spatiotemporal profiles, as described in this invention includes the following steps:

[0006] S1. For charging stations within the target distribution network area, based on the correlation mechanism formula between charging power, duration, and start time, differentiated combinations of key influencing parameters such as charging pile type (fast charging / slow charging), battery capacity, functional area attributes, charging start time and duration are set to generate various types of charging load profiles; feature parameters such as electricity consumption period distribution in historical electricity consumption data of the area are extracted, and an improved K-means clustering algorithm is used to cluster according to the functional areas of residential areas, commercial areas, and highway service areas within the target distribution transformer area to generate a basic load profile for the target area.

[0007] S2. For existing charging stations in the target distribution network area, obtain the historical power consumption data of the transformers where each charging station is located in the area. Combine the generated charging load profile and basic load profile, and use the improved GA optimization algorithm to separate the charging load data. With the upper limit of the safe operating capacity of the transformer as a constraint, output the open capacity of the transformers in the existing charging station area.

[0008] S3. For newly added charging stations within the target distribution network area, using the charging load profile generated in S1 as a sample, and classifying them into functional areas such as residential areas, commercial areas, and highway service areas, the charging load characteristic parameters of existing charging stations in the same functional area are selected as interpolation sample points. The fitting parameters of the charging load curve of the newly added charging station are obtained by using the Kriging interpolation method. The charging load curve of the newly added charging station is generated by fitting, and the available capacity of the transformer in the area of ​​the newly added charging station is output by calculation.

[0009] S4. Based on the distribution network topology, and using the available capacity of each transformer calculated in S2 and S3 as a basis, the capacity is verified step by step from the 10kV voltage level to the 110kV voltage level. The matching degree between the total available capacity of the lower-level charging stations and the available capacity of the upper-level nodes is compared. When the available capacity of any upper-level voltage level node is lower than the total capacity of its lower-level nodes, the available capacity is optimized and allocated based on the functional area attributes of each charging station, load fluctuation characteristics, and the carrying capacity margin of the distribution network lines. The improved entropy weight-TOPSIS multi-criteria allocation algorithm is used.

[0010] Furthermore, in S1, based on the correlation formula between charging power, duration, and start time, differentiated combinations of key influencing parameters such as charging pile type, battery capacity, functional area attributes, and charging start time are set to generate initial profiles of various types of charging loads, specifically:

[0011] a. Standardize and quantify the core influencing parameters: Divide charging pile types into DC fast charging (T=1, DC fast charging) and AC slow charging (T=2, AC slow charging); classify battery capacity by vehicle type into small cars, medium cars, and large cars; classify functional area attributes into residential areas (A=1, charging demand concentrated at night), commercial areas (A=2, random charging during the day), and highway service areas (A=3, short-term charging); and classify the charging start time into valley periods (H=1, time period 0:) based on peak-valley average. The time period is divided into three categories: peak period (H=2, 8:00-17:00) and off-peak period (H=3, 17:00-24:00). The initial remaining power SOC0 is set to 0.3 (i.e., 30%) by default. The correction coefficient k(A,H) is determined according to the coupling of functional area and time period (residential area: off-peak 1.0, off-peak 0.95, peak 0.85; commercial area: off-peak 0.95, off-peak 1.0, peak 0.90; highway service area: off-peak 1.0, off-peak 1.0, peak 0.95).

[0012] b. Establish the core formula for the correlation mechanism between charging power, duration, and start time:

[0013] 1) Charging power quantification formula (based on charging pile type):

[0014] (1)

[0015] In the formula, Rated power of the charging pile (baseline value): kW (average DC fast charging power); kW (average AC slow charging); k(A,H) is a correction factor that reflects the actual available power in different scenarios.

[0016] 2) Formula for calculating charging time:

[0017] Charging time is determined by battery capacity, remaining charge level, actual charging power, and efficiency, as shown in the following formula:

[0018] (2)

[0019] In the formula, The target remaining battery capacity is 0.95 for residential areas, 0.7 for commercial areas, and 0.8 for highway service areas; C represents the battery capacity ( For small cars, It is a mid-size car. (For large vehicles) This is the initial battery capacity; For charging efficiency.

[0020] 3) Charging load time distribution formula:

[0021] The time-period distribution of charging load is determined by the starting time period and the charging duration, as shown in the following formula:

[0022] (3)

[0023] In the formula, h represents the current time, and its value ranges from 0 to 24; The reference time for the start time period: h (typical starting time of valley segment); h (typical starting time of the horizontal segment); h (typical start time of peak segment).

[0024] c. Set differentiated combination schemes and generate an initial profile library: Set combination schemes according to the dimensions of "charging pile type (2 types) × battery capacity (3 types) × functional area attributes (3 types) × charging start time (3 types)", and select 9 typical scenarios (DC fast charging + vehicles of different capacities + highway service area + flat section, AC slow charging + vehicles of different capacities + residential area + valley section, DC fast charging + vehicles of different capacities + commercial area + peak section); For each typical scenario, input the differentiated combination parameters of the scenario in sequence, calculate and extract the core features of the charging load profile based on the charging power quantification formula in step b, and form an initial profile library of multi-type charging loads covering multiple scenario labels.

[0025] The advantages of the above steps compared to existing technologies are as follows: standardizing and quantifying parameters such as charging pile type and functional area attributes to solve the problem of chaotic traditional parameters; constructing power, duration, and load distribution formulas that couple multiple factors to avoid rough empirical estimations; and establishing a multi-scenario profile library to make up for the deficiency of incomplete scenario coverage, provide accurate data for subsequent calculations, and solve the problem of the disconnect between traditional profiles and reality.

[0026] Furthermore, in S1, an improved K-means clustering algorithm is used to cluster residential areas, commercial areas, and highway service areas within the target distribution network area, generating a basic load profile for the target area, specifically:

[0027] a. Data preparation and feature extraction: Extract the daily electricity consumption of each power consumption unit from the historical basic load data of the target distribution network area. Peak load Valley load ratio Peak load ratio Load fluctuation coefficient (Standard deviation of daily electricity consumption) Average daily electricity consumption The ratio, i.e. These core features constitute the sample set. .

[0028] b. Data standardization:

[0029] (4)

[0030] in, These are the original eigenvalues; Let be the mean of the j-th feature; Let be the standard deviation of the j-th feature; The standardized sample set is ;

[0031] c. Based on the prior knowledge of the electricity consumption characteristics of the three functional areas, select the initial center:

[0032] 1) Select the sample with the highest proportion of valleys and the smallest fluctuation coefficient as the initial center of the residential area. ;

[0033] 2) Select the sample with the highest peak percentage and largest fluctuation coefficient as the initial center of the commercial area. ;

[0034] 3) Select samples with "high peak load and concentrated time period" as the initial center of the highway service area;

[0035] d. Iterative clustering:

[0036] 1) First calculate the Euclidean distance using equation (5), then assign each sample to the nearest class using equation (6), and obtain the result. ;

[0037] (5)

[0038] (6)

[0039] In the formula, These are the standardized eigenvalues; This is the standardized mean; Let be the standardized feature value of the i-th sample; Let be the standardized mean of the i-th sample.

[0040] 2) Cluster center update:

[0041] (7)

[0042] In the formula, For the set of samples of class k, .

[0043] 3) Convergence judgment

[0044] Stop when the change in the sum of squared errors (SSE) within a class is less than the threshold or the number of iterations is ≥30:

[0045] (8)

[0046] e. Generate a basic load profile

[0047] For each cluster category, denormalize the standardized features. Calculate the mean of the original features, label the functional areas, and generate a basic load profile.

[0048] The advantages of the above steps compared to existing technologies are as follows: by extracting multi-dimensional electricity consumption characteristics and combining them with the prior selection of initial centers for electricity consumption in functional areas, the problems of random centers and large clustering bias in traditional K-means are solved; by generating profiles through precise iteration and de-standardization, the load characteristics of different functional areas can be accurately distinguished, which greatly improves the adaptability of scenarios and clustering efficiency.

[0049] Furthermore, in S2, an improved GA optimization algorithm is used to separate the charging load data, specifically:

[0050] First, based on the functional area where the charging station is located (residential area / commercial area / highway service area), extract the charging load profile:

[0051] (9)

[0052] In the formula, Profiles representing different types of charging loads

[0053] Then, extract the baseline load profile:

[0054] (10)

[0055] In the formula, These represent different types of basic load profiles.

[0056] Secondly, a load separation optimization model should be constructed, which should satisfy power constraints:

[0057] (11)

[0058] In the formula, For other base loads; This is the charging load.

[0059] Define the parameter vector to be optimized Two load optimization models were established:

[0060] (12)

[0061] In the formula, The characteristic weighting coefficient for charging load is... The basic load characteristic weighting coefficient; and These are the profiles corresponding to charging load and base load, respectively.

[0062] Secondly, the GA optimization algorithm is improved to solve for the parameter vector to be optimized. First, initialize the population and set the population size. upper and lower bounds of parameters , N sets of parameter combinations are randomly generated as initial individuals to form the initial population; the fitness function is designed with the goal of minimizing the load separation error:

[0063] (13)

[0064] In the formula, This is the sum of squared errors between the total load and the decomposed load; a higher fitness value indicates a better separation effect.

[0065] Then, elite selection is used to ensure convergence; single-point crossover is used to achieve gene recombination; random mutation is used to avoid local optima and generate a new generation of population; when the number of iterations reaches a preset value (100~200 generations) or the difference in fitness between two adjacent generations is less than 100, the population is considered to be in a state of convergence. Stop iteration and output the optimal parameters. .

[0066] Finally, substituting the optimal parameters, the final charging load is calculated:

[0067] (14)

[0068] In the formula, The optimal charging load characteristic weighting coefficient; A profile corresponding to the charging load.

[0069] The advantages of the above steps compared to existing technologies are as follows: by accurately constructing a load separation model, designing a fitness function with the goal of minimizing error, and combining elite selection, crossover mutation, and strict convergence criteria, the shortcomings of traditional algorithms such as slow convergence and easy trapping in local optima are solved, which greatly improves the accuracy and efficiency of charging load separation and provides reliable data support for subsequent evaluation.

[0070] Furthermore, in S2, the available capacity of the existing charging station area transformers is output, specifically as follows:

[0071] Based on the premise of ensuring the safe operation of the transformer, the rated capacity of the transformer should be determined first. The total installed charging pile capacity of the substation is [not specified]. (Rated power per charging pile × number of charging piles), and then obtain the total charging power for a certain period of time through the actual operation data of the charging station. For newly added charging stations Next, set the safety upper limit factor for the rated capacity of the transformer. (%), remaining available capacity of the computing platform This refers to the final available capacity of the transformer.

[0072] Furthermore, S3 specifically refers to:

[0073] S3-1. Obtain typical charging load profiles from S1 and geographical information of the functional areas to which existing charging stations belong, and define the sample set:

[0074] Each existing charging station serves as one sample point. ,in: For the geographical coordinates of existing charging stations, The load percentage of image j in the i-th existing charging station satisfies .

[0075] S3-2. Perform Kriging interpolation on the image weight parameters to obtain the fitting parameters for the load curve of the newly added charging station:

[0076] a. Extract a single image sample set: using the weight of image 1 For example, sample set

[0077] b. Calculate spatial distance:

[0078] 1) Distance between sample points:

[0079] 2) Distance between sample points and newly added stations:

[0080] c. Fit a spherical variogram. The variogram needs to be fitted to a continuous theoretical model to calculate the correlation at any distance. The formula for the spherical model is:

[0081] (15)

[0082] In the formula, The value of the nugget reflects the measurement error; The sill value-null value reflects spatial structural variation; The range is variable, reflecting the spatial correlation range.

[0083] d. Solving for the weights :

[0084] (16)

[0085] It should satisfy the property of being unbiased, that is .

[0086] e. Calculate the image weight of the newly added site:

[0087] f. Interpolate images 2-9 respectively until the result is obtained. to ;

[0088] g. If the interpolated sum deviates from 1, perform normalization. Ultimately retained This is the final weight.

[0089] S3-3. Generating the charging load curve for newly added charging stations, the formula is as follows:

[0090] (17)

[0091] In the formula, The time-by-time load curve of the j-th typical portrait; The weight of the j-th image obtained by interpolation; This represents the total number of newly added station piles.

[0092] S3-4. Calculate the available capacity of the distribution transformer in the area of ​​the newly added charging station. The calculation formula is the same as... Consistent.

[0093] The advantages of the above steps compared to existing technologies are as follows: by combining geographic information with profile weights, and quantifying spatial correlation by fitting a spherical variogram, the profile weights of newly added stations are accurately solved and load curves are generated. This solves the problem of traditional methods ignoring spatial differences, significantly improves the load fitting accuracy of newly added stations, and provides reliable technical support for the calculation of open capacity of distribution transformers and site planning.

[0094] Furthermore, S4 specifically refers to:

[0095] S4-1. For existing charging stations, if the downstream nodes have the available capacity... ,in, If the upper-level node's safe operating limit is reached, capacity optimization allocation is triggered; otherwise, it is considered that the upper and lower levels are matched. For newly added charging stations, if the lower-level node can open up capacity... If the allocation is correct, then optimized allocation is triggered; otherwise, it is considered that the upper and lower levels are matched.

[0096] S4-2. Improved entropy weight-TOPSIS optimization allocation for nodes with mismatched hierarchical levels. The specific steps are as follows:

[0097] a. Suppose there are n charging stations to be allocated, and m evaluation indicators (such as functional area attributes, load fluctuation, line margin, etc.), the matrix is:

[0098] (18)

[0099] in This is the original value of the m-th indicator for the n-th charging station.

[0100] b. Indicator standardization and entropy weight calculation:

[0101] (19)

[0102] (20)

[0103] In the formula, For the probability index; For the index entropy; Entropy weight.

[0104] c. Weighted standardization and ideal solution:

[0105] (twenty one)

[0106] In the formula, It is a weighted matrix; , These are the positive and negative ideal solutions, respectively.

[0107] d. Distance and proximity (assignment priority):

[0108] (twenty two)

[0109] In the formula, , These represent the distances to the positive and negative ideal solutions, respectively. For proximity, a larger value indicates higher priority.

[0110] e. Capacity allocation:

[0111] (twenty three)

[0112] In the formula, This is a reduction in the amount of charge per charging station. This represents the final, optimized, and available capacity.

[0113] The advantages of the above steps compared to existing technologies are as follows: by objectively quantifying the weight of indicators through entropy weight, combining the proximity calculation of TOPSIS to determine the allocation priority, and considering indicators such as functional areas and load fluctuations from multiple dimensions, the shortcomings of traditional allocation that are subjective and have a single dimension are solved, thereby achieving scientific optimization of capacity allocation and improving the efficiency of distribution network resource utilization and the safety of power grid operation.

[0114] The beneficial effects of this invention are as follows:

[0115] 1) By constructing a charging load profile with differentiated parameters and a basic load profile with improved K-means clustering, the load characteristics of functional areas such as residential areas, commercial areas, and highway service areas, as well as fast charging / slow charging scenarios, can be accurately distinguished. Key information from historical electricity consumption data can be effectively extracted, reducing load type confusion errors and providing accurate data support for subsequent charging load separation and capacity calculation. This significantly improves the accuracy and scenario-specificity of distribution network load analysis.

[0116] 2) The improved GA optimization algorithm proposed in this invention achieves precise separation of charging load. It calculates the available capacity of existing charging station area transformers by combining transformer safety constraints, and uses Kriging interpolation to fit the load curve of new charging stations. This ensures that existing transformers do not exceed the safety limit, accurately assesses the capacity requirements of new charging stations, avoids blind construction or capacity waste, achieves precise matching between distribution network resources and charging demand, and improves operational safety and resource utilization efficiency.

[0117] 3) By verifying the capacity at each level from 10kV to 110kV and optimizing the allocation of entropy weight-TOPSIS, when there is a mismatch between the capacity of the upper and lower levels, the key scenario needs are prioritized based on multiple criteria such as functional area attributes and load fluctuations. This solves the capacity bottleneck problem of the distribution network level, avoids local overload limiting the carrying capacity, and improves the overall capacity of the distribution network to accept charging loads, thus adapting to the growth of charging demand for new energy vehicles. Attached Figure Description

[0118] Figure 1 This is a flowchart of the method described in this invention;

[0119] Figure 2 This is a diagram of the distribution network topology of the method described in this invention;

[0120] Figure 3 A diagram of the charging load;

[0121] Figure 4 A basic load profile;

[0122] Figure 5 A schematic diagram illustrating the step-by-step verification of the developable capacity of the distribution network area. Detailed Implementation

[0123] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0124] Example: Figure 1 A schematic flowchart of a method for assessing the charging load capacity of electric vehicles in a distribution network, considering spatiotemporal profiles, is provided as an example of this invention; Figure 1 As shown, the method of the present invention includes the following steps:

[0125] S1. For charging stations within the target distribution network area, based on the correlation mechanism formula between charging power, duration, and start time, differentiated combinations of key influencing parameters such as charging pile type (fast charging / slow charging), battery capacity, functional area attributes, and charging start time and duration are set to generate various types of charging load profiles; feature parameters such as electricity consumption time distribution are extracted from the historical electricity consumption data of the area, and an improved K-means clustering algorithm is used to cluster according to the functional areas of residential areas, commercial areas, and highway service areas within the target distribution network area to generate a basic load profile for the target area; specifically:

[0126] First, based on the correlation formula between charging power, duration, and start time, differentiated combinations of key influencing parameters such as charging pile type, battery capacity, functional area attributes, and charging start time are set to generate initial profiles of various types of charging loads. The steps are as follows:

[0127] a. Standardize and quantify the core influencing parameters: Divide charging pile types into DC fast charging (T=1, DC fast charging) and AC slow charging (T=2, AC slow charging); classify battery capacity by vehicle type into small cars, medium cars, and large cars; classify functional area attributes into residential areas (A=1, charging demand concentrated at night), commercial areas (A=2, random charging during the day), and highway service areas (A=3, short-term charging); and classify the charging start time into valley periods (H=1, time period 0:) based on peak-valley average. The time period is divided into three categories: peak period (H=2, 8:00-17:00) and off-peak period (H=3, 17:00-24:00). The initial remaining power SOC0 is set to 0.3 (i.e., 30%) by default. The correction coefficient k(A,H) is determined according to the coupling of functional area and time period (residential area: off-peak 1.0, off-peak 0.95, peak 0.85; commercial area: off-peak 0.95, off-peak 1.0, peak 0.90; highway service area: off-peak 1.0, off-peak 1.0, peak 0.95).

[0128] b. Establish the core formula for the correlation mechanism between charging power, duration, and start time:

[0129] 1) Charging power quantification formula (based on charging pile type):

[0130] (1)

[0131] In the formula, Rated power of the charging pile (baseline value): kW (average DC fast charging power); kW (average AC slow charging); k(A,H) is a correction factor that reflects the actual available power in different scenarios.

[0132] 2) Formula for calculating charging time:

[0133] Charging time is determined by battery capacity, remaining charge level, actual charging power, and efficiency, as shown in the following formula:

[0134] (2)

[0135] In the formula, The target remaining battery capacity is 0.95 for residential areas, 0.7 for commercial areas, and 0.8 for highway service areas; C represents the battery capacity ( For small cars, It is a mid-size car. (For large vehicles) This is the initial battery capacity; For charging efficiency.

[0136] 3) Charging load time distribution formula:

[0137] The time-period distribution of charging load is determined by the starting time period and the charging duration, as shown in the following formula:

[0138] (3)

[0139] In the formula, h represents the current time, and its value ranges from 0 to 24; The reference time for the start time period: h (typical starting time of valley segment); h (typical starting time of the horizontal segment); h (typical start time of peak segment).

[0140] c. Set differentiated combination schemes and generate an initial profile library: Set combination schemes according to the dimensions of "charging pile type (2 types) × battery capacity (3 types) × functional area attributes (3 types) × charging start time (3 types)", and select 9 typical scenarios (DC fast charging + vehicles of different capacities + highway service area + flat section, AC slow charging + vehicles of different capacities + residential area + valley section, DC fast charging + vehicles of different capacities + commercial area + peak section); For each typical scenario, input the differentiated combination parameters of the scenario in sequence, calculate and extract the core features of the charging load profile based on the charging power quantification formula in step b, and form an initial profile library of multi-type charging loads covering multiple scenario labels.

[0141] Subsequently, an improved K-means clustering algorithm was used to cluster the target distribution network area according to its functional zones: residential areas, commercial areas, and highway service areas, generating a basic load profile for the target area. The steps are as follows:

[0142] a. Data preparation and feature extraction: Extract the daily electricity consumption of each power consumption unit from the historical basic load data of the target distribution network area. Peak load Valley load ratio Peak load ratio Load fluctuation coefficient (Standard deviation of daily electricity consumption) Average daily electricity consumption The ratio, i.e. These core features constitute the sample set. .

[0143] b. Data standardization:

[0144] (4)

[0145] in, These are the original eigenvalues; Let be the mean of the j-th feature; Let be the standard deviation of the j-th feature; The standardized sample set is ;

[0146] c. Based on the prior knowledge of the electricity consumption characteristics of the three functional areas, select the initial center:

[0147] 1) Select the sample with the highest proportion of valleys and the smallest fluctuation coefficient as the initial center of the residential area. ;

[0148] 2) Select the sample with the highest peak percentage and largest fluctuation coefficient as the initial center of the commercial area. ;

[0149] 3) Select samples with "high peak load and concentrated time period" as the initial center of the highway service area;

[0150] d. Iterative clustering:

[0151] 1) First calculate the Euclidean distance using equation (5), then assign each sample to the nearest class using equation (6), and obtain the result. ;

[0152] (5)

[0153] (6)

[0154] In the formula, These are the standardized eigenvalues; This is the standardized mean; Let be the standardized feature value of the i-th sample; Let be the standardized mean of the i-th sample.

[0155] 2) Cluster center update:

[0156] (7)

[0157] In the formula, For the set of samples of class k, .

[0158] 3) Convergence judgment

[0159] Stop when the change in the sum of squared errors (SSE) within a class is less than the threshold or the number of iterations is ≥30:

[0160] (8)

[0161] e. Generate a basic load profile

[0162] For each cluster category, denormalize the standardized features. Calculate the mean of the original features, label the functional areas, and generate a basic load profile.

[0163] S2. For existing charging stations within the target distribution network area, obtain the historical electricity consumption data of the transformers where each charging station is located. Combine this data with the generated charging load profile and basic load profile, and use an improved GA optimization algorithm to separate the charging load data. Using the upper limit of the transformer's safe operating capacity as a constraint, output the available capacity of the transformers in the existing charging station area. Specifically:

[0164] S2-1. The improved GA optimization algorithm is used to separate the charging load data. The steps are as follows:

[0165] First, based on the functional area where the charging station is located (residential area / commercial area / highway service area), extract the charging load profile:

[0166] (9)

[0167] In the formula, Profiles representing different types of charging loads

[0168] Then, extract the baseline load profile:

[0169] (10)

[0170] In the formula, These represent different types of basic load profiles.

[0171] Secondly, a load separation optimization model should be constructed, which should satisfy power constraints:

[0172] (11)

[0173] In the formula, For other base loads; This is the charging load.

[0174] Define the parameter vector to be optimized Two load optimization models were established:

[0175] (12)

[0176] In the formula, The characteristic weighting coefficient for charging load is... The basic load characteristic weighting coefficient; and These are the profiles corresponding to charging load and base load, respectively.

[0177] Secondly, the GA optimization algorithm is improved to solve for the parameter vector to be optimized. First, initialize the population and set the population size. upper and lower bounds of parameters , N sets of parameter combinations are randomly generated as initial individuals to form the initial population; the fitness function is designed with the goal of minimizing the load separation error:

[0178] (13)

[0179] In the formula, This is the sum of squared errors between the total load and the decomposed load; a higher fitness value indicates a better separation effect.

[0180] Then, elite selection is used to ensure convergence; single-point crossover is used to achieve gene recombination; random mutation is used to avoid local optima and generate a new generation of population; when the number of iterations reaches a preset value (100~200 generations) or the difference in fitness between two adjacent generations is less than 100, the population is considered to be in a state of convergence. Stop iteration and output the optimal parameters. .

[0181] Finally, substituting the optimal parameters, the final charging load is calculated:

[0182] (14)

[0183] In the formula, The optimal charging load characteristic weighting coefficient; A profile corresponding to the charging load.

[0184] S2-2. Based on the premise of safe operation constraints of the transformer, first determine the rated capacity of the transformer. The total installed charging pile capacity of the substation is [not specified]. (Rated power per charging pile × number of charging piles), and then obtain the total charging power for a certain period of time through the actual operation data of the charging station. For newly added charging stations Next, set the safety upper limit factor for the rated capacity of the transformer. (%), remaining available capacity of the computing platform This is the final available capacity of the transformer.

[0185] S3. For newly added charging stations within the target distribution network area, using the charging load profile generated in S1 as a sample, and categorizing them by functional areas (residential, commercial, and highway service areas), select the charging load characteristic parameters of existing charging stations in the same functional area as interpolation sample points. Use Kriging interpolation to obtain the fitting parameters for the charging load curve of the newly added charging stations; fit the charging load curve of the newly added charging stations; and calculate and output the available capacity of the transformers in the area of ​​the newly added charging stations. Specifically:

[0186] S3-1. Obtain typical charging load profiles from S1 and geographical information of the functional areas to which existing charging stations belong, and define the sample set:

[0187] Each existing charging station serves as one sample point. ,in: For the geographical coordinates of existing charging stations, The load percentage of image j in the i-th existing charging station satisfies .

[0188] S3-2. Perform Kriging interpolation on the image weight parameters to obtain the fitting parameters for the load curve of the newly added charging station:

[0189] a. Extract a single image sample set: using the weight of image 1 For example, sample set

[0190] b. Calculate spatial distance:

[0191] 1) Distance between sample points:

[0192] 2) Distance between sample points and newly added stations:

[0193] c. Fit a spherical variogram. The variogram needs to be fitted to a continuous theoretical model to calculate the correlation at any distance. The formula for the spherical model is:

[0194] (15)

[0195] In the formula, The value of the nugget reflects the measurement error; The sill value-null value reflects spatial structural variation; The range is variable, reflecting the spatial correlation range.

[0196] d. Solving for the weights :

[0197] (16)

[0198] It should satisfy the property of being unbiased, that is .

[0199] e. Calculate the image weight of the newly added site:

[0200] f. Interpolate images 2-9 respectively until the result is obtained. to ;

[0201] g. If the interpolated sum deviates from 1, perform normalization. Ultimately retained This is the final weight.

[0202] S3-3. Generating the charging load curve for newly added charging stations, the formula is as follows:

[0203] (17)

[0204] In the formula, The time-by-time load curve of the j-th typical portrait; The weight of the j-th image obtained by interpolation; This represents the total number of newly added station piles.

[0205] S3-4. Calculate the available capacity of the distribution transformer in the area of ​​the newly added charging station. The calculation formula is as follows: To calculate.

[0206] S4. Based on the distribution network topology, and using the available capacity of each transformer calculated in S2 and S3 as a basis, the capacity is verified step by step from the 10kV voltage level upwards to the 110kV voltage level. The matching degree between the total available capacity of the lower-level charging stations and the available capacity of the upper-level nodes is compared. When the available capacity of any upper-level voltage level node is lower than the total capacity of its lower-level nodes, the available capacity is optimized and allocated based on the functional area attributes of each charging station, load fluctuation characteristics, and the carrying capacity margin of the distribution network lines. Specifically, the improved entropy weight-TOPSIS multi-criteria allocation algorithm is used for the following:

[0207] S4-1. For existing charging stations, if the downstream nodes have the available capacity... ,in, If the upper-level node's safe operating limit is reached, capacity optimization allocation is triggered; otherwise, it is considered that the upper and lower levels are matched. For newly added charging stations, if the lower-level node can open up capacity... If the allocation is correct, then optimized allocation is triggered; otherwise, it is considered that the upper and lower levels are matched.

[0208] S4-2. Improved entropy weight-TOPSIS optimization allocation for nodes with mismatched hierarchical levels. The specific steps are as follows:

[0209] a. Suppose there are n charging stations to be allocated, and m evaluation indicators (such as functional area attributes, load fluctuation, line margin, etc.), the matrix is:

[0210] (18)

[0211] in This is the original value of the m-th indicator for the n-th charging station.

[0212] b. Indicator standardization and entropy weight calculation:

[0213] (19)

[0214] (20)

[0215] In the formula, For the probability index; For the index entropy; Entropy weight.

[0216] c. Weighted standardization and ideal solution:

[0217] (twenty one)

[0218] In the formula, It is a weighted matrix; , These are the positive and negative ideal solutions, respectively.

[0219] d. Distance and proximity (assignment priority):

[0220] (twenty two)

[0221] In the formula, , These represent the distances to the positive and negative ideal solutions, respectively. For proximity, a larger value indicates higher priority.

[0222] e. Capacity allocation:

[0223] (twenty three)

[0224] In the formula, This is a reduction in the amount of charge per charging station. This represents the final, optimized, and available capacity.

[0225] Experimental verification

[0226] To enable those skilled in the art to better understand the present invention, examples are given below.

[0227] A radial power grid was selected as the test area, and its power grid topology is as follows: Figure 2 As shown, Figure 2 Node 1 operates at 110kV, nodes 2 and 3 at 35kV, nodes 4, 5, and 6 at 10kV, and the remaining nodes at 380V. The lower-level nodes in this topology are divided into three functional areas: a residential area, a commercial area, and a highway service area. Each functional area is further divided into newly added charging stations and existing charging stations. The simulation environment is a 64-bit Windows operating system, an Intel(R) Core(TM) i7-7700 CPU @3.6GHz, and 64GB of RAM.

[0228] Charging load profile and base load profile as follows Figure 3 As shown in Table 4; based on these two profiles, the available capacity of the distribution transformer area for existing charging stations and newly added charging stations is calculated according to S2 and S3 respectively, as shown in Table 1.

[0229] Table 1 Available Capacity for Each Site

[0230]

[0231] Subsequently, S4 was used to verify the open capacity of the upper and lower level nodes considering the topology. The verification results are shown in Tables 2 and 3.

[0232] Table 2 Adjustment Amounts for Each Station

[0233]

[0234]

[0235] The results of the multi-level voltage level verification intuitively demonstrate the control effect of the method. Table 3 shows that the initial total open capacity of 35kV node 2 was 3800kW, exceeding the budgeted capacity by 800kW, and 110kV node 1 exceeded the budget by 200kW, posing a significant safety risk. After step-by-step verification from 10kV to 110kV and improved entropy weight-TOPSIS multi-criteria allocation, the total capacity after verification at both levels of nodes strictly matched the budget upper limit, resolving the over-limit issue. From the site level, the capacity of each site was dynamically adjusted based on the constraints of the upper-level nodes and its own characteristics. For example, after verification at 35kV and 110kV, the capacity of commercial site 9 was optimized from the initial 391.4kW to 77.4kW. Other sites underwent similar adjustments, and the results fully aligned with the load fluctuation characteristics of different functional areas and the line carrying capacity margin.

[0236] Meanwhile, the bar charts of initial and final station capacities show that the reduction ratios were allocated systematically according to functional area attributes and load characteristics, with no indiscriminate capacity restrictions. Overall, this method, through a comprehensive design encompassing precise load profiling, modeled capacity calculation, and intelligent allocation optimization, maximizes the carrying capacity potential of the existing power grid while ensuring the safe operation of the distribution network. It provides a scientific basis for the planning of new charging stations and the operation of existing stations, demonstrating strong engineering application value.

[0237] The above description is merely a preferred embodiment of the present invention and is not intended to further limit the present invention. All equivalent changes made based on the description and drawings of the present invention are within the protection scope of the present invention.

Claims

1. A method for assessing the charging load carrying capacity of electric vehicles in power distribution networks considering spatiotemporal profiles, characterized in that, Includes the following steps: S1. For charging stations within the target distribution network area, based on the correlation mechanism formula between charging power, duration and start time, set differentiated combinations of key influencing parameters such as charging pile type, battery capacity, functional area attributes, charging start time and duration to generate various types of charging load profiles. Extract the period distribution characteristic parameters of electricity consumption data from the historical electricity consumption data of the region, and use a clustering algorithm to cluster the target distribution transformer area according to the functional areas of residential areas, commercial areas and highway service areas to generate a basic load profile of the target area. S2. For existing charging stations within the target distribution network area, obtain the historical electricity consumption data of the transformers where each charging station is located within the area. Combine the generated charging load profile and basic load profile, and use the GA optimization algorithm to separate the charging load data. Based on the upper limit of the safe operating capacity of the transformer, output the available capacity of the transformers in the existing charging station area; S3. For newly added charging stations within the target distribution network area, using the charging load profile generated in S1 as a sample, and classifying them into functional areas such as residential areas, commercial areas, and highway service areas, the charging load characteristic parameters of existing charging stations in the same functional area are selected as interpolation sample points. The fitting parameters of the charging load curve of the newly added charging station are obtained by using the Kriging interpolation method. The charging load curve of the newly added charging station is generated by fitting, and the available capacity of the transformer in the area of ​​the newly added charging station is output by calculation. S4. Based on the distribution network topology, and using the available capacity of each transformer calculated in S2 and S3 as a basis, the capacity is verified step by step from the 10kV voltage level to the 110kV voltage level. The matching degree between the total available capacity of the lower-level charging stations and the available capacity of the upper-level nodes is compared. When the available capacity of any upper-level voltage level node is lower than the total capacity of its lower-level nodes, the available capacity is optimized and allocated based on the functional area attributes of each charging station, load fluctuation characteristics, and the carrying capacity margin of the distribution network lines. The improved entropy weight-TOPSIS multi-criteria allocation algorithm is used to optimize the allocation of available capacity. S1 specifically refers to: a. Standardize and quantify the core influencing parameters: classify charging pile types into DC fast charging and AC slow charging; classify battery capacity into three categories according to vehicle type: small car, medium car, and large car; classify functional area attributes into three categories: residential area, commercial area, and highway service area; classify charging start time into three categories according to peak, valley, and average time: valley, average, and peak time; and determine the correction coefficient k(A,H) according to the coupling of functional area and time period. b. Establish the core formula for the correlation mechanism between charging power, duration, and start time: 1) Charging power quantification formula, based on charging pile type: (1) In the formula, Rated power of the charging station: kW, which is the average DC fast charging power; kW represents the average AC slow charging power; k(A,H) is a correction factor that reflects the actual usable power in different scenarios. 2) Formula for calculating charging time: Charging time is determined by battery capacity, remaining charge level, actual charging power, and efficiency, as shown in the following formula: (2) In the formula, C represents the target remaining battery power; C represents the battery capacity. This is the initial battery capacity; For charging efficiency, 3) Charging load time distribution formula: The time-period distribution of charging load is determined by the starting time period and the charging duration, as shown in the following formula: (3) In the formula, h represents the current time; The reference time for the start time period; c. Set up differentiated combination schemes and generate an initial profile library: Set up combination schemes according to the dimensions of "charging pile type × battery capacity × functional area attribute × charging start time", and select 9 typical scenarios; for each typical scenario, input the differentiated combination parameters of the scenario in sequence, calculate and extract the core features of the charging load profile based on the charging power quantification formula in step b, and form an initial profile library of multiple types of charging loads covering multiple scenario tags. S4 specifically refers to: S4-1. For existing charging stations, if downstream nodes can open up capacity... ,in, If the upper-level node's safe operating limit is reached, capacity optimization allocation is triggered; otherwise, it is considered that the upper and lower levels are matched. For newly added charging stations, if the lower-level node can open up capacity... If the allocation is correct, then optimized allocation is triggered; otherwise, it is considered that the upper and lower levels are matched. S4-2. Improved entropy weight-TOPSIS optimization allocation for nodes with mismatched hierarchical levels. The specific steps are as follows: Let there be n charging stations to be allocated and m evaluation indicators, with the matrix as follows: （18） in This represents the original value of the m-th indicator for the n-th charging station. b. Standardization of the index and calculation of entropy weight: （19） （20） In the formula, For the probability index; For the index entropy; For entropy weight, c-weighted standardization and ideal solution: （21） In the formula, It is a weighted matrix; , These are the positive and negative ideal solutions, respectively. d. Distance and proximity: （22） In the formula, , These represent the distances to the positive and negative ideal solutions, respectively. For closer relevance, a higher value indicates higher priority. e-capacity allocation: （23） In the formula, The reduction is for a single charging station; This refers to the final optimized allocation of available capacity. The available capacity for newly added charging stations.

2. The method for assessing the charging load carrying capacity of electric vehicles in a distribution network considering spatiotemporal profiles according to claim 1, characterized in that, In S1, a clustering algorithm is used to cluster residential areas, commercial areas, and highway service areas within the target distribution network area, generating a basic load profile for the target area. Specifically: a. Data preparation and feature extraction: Extract the daily electricity consumption of each power consumption unit from the historical basic load data of the target distribution network area. Peak load Valley load ratio Peak load ratio Load fluctuation coefficient These core features constitute the sample set ,in Standard deviation of daily electricity consumption Average daily electricity consumption The ratio, b. Data standardization: (4) in, These are the original eigenvalues; Let be the mean of the j-th feature; Let be the standard deviation of the j-th feature; The standardized sample set is ; c. Based on the prior knowledge of the electricity consumption characteristics of the three functional areas, the initial center is selected: 1) Select the sample with the highest proportion of valleys and the smallest fluctuation coefficient as the initial center of the residential area. ; 2) Select the sample with the "highest peak percentage and largest fluctuation coefficient" as the initial center of the commercial area. ; 3) Select samples with "high peak load and concentrated time period" as the initial center of the highway service area; d-Iterative Clustering: 1) First calculate the Euclidean distance using equation (5), then assign each sample to the nearest class using equation (6), and obtain the result. ; （5） （6） In the formula, These are the standardized eigenvalues; This is the standardized mean. The standardized feature value of the i-th sample; Let be the standardized mean of the i-th sample. 2) Cluster center update: （7） In the formula, For the set of samples of class k, ; 3) Convergence judgment Stop when the change in the sum of squared errors (SSE) within a class is less than the threshold or the number of iterations is ≥30: （8） e generates a baseline load profile. For each cluster category, denormalize the standardized features. Calculate the mean of the original features, label the functional areas, and generate a basic load profile.

3. The method for assessing the charging load carrying capacity of electric vehicles in a distribution network considering spatiotemporal profiles according to claim 2, characterized in that, In S2, the GA optimization algorithm is used to separate the charging load data, specifically as follows: First, based on the functional area where the charging station is located, extract the charging load profile: （9） In the formula, A profile representing different types of charging loads. Then, extract the baseline load profile: （10） In the formula, Representing different types of basic load profiles. Secondly, a load separation optimization model should be constructed, which should satisfy power constraints: （11） In the formula, For other base loads; For charging load, Define the parameter vector to be optimized Two load optimization models were established: (12) In the formula, The characteristic weighting coefficient for charging load is... It is the basic load characteristic weighting coefficient; and These are the profiles corresponding to charging load and base load, respectively. Secondly, the GA optimization algorithm is improved to solve for the parameter vector to be optimized. First, initialize the population and set the population size. upper and lower bounds of parameters , N sets of parameter combinations are randomly generated as initial individuals to form the initial population; the fitness function is designed with the goal of minimizing the load separation error: (13) In the formula, The fitness value is the sum of squared errors between the total load and the decomposed load. A higher fitness value indicates a better separation effect. Then, elite selection is used to ensure convergence; single-point crossover is used to achieve gene recombination; random mutation is used to avoid local optima and generate a new generation of population; when the number of iterations reaches a preset value or the difference in fitness between two adjacent generations is less than a certain value, the population is considered to be successfully integrated. Stop iteration and output the optimal parameters. , Finally, substituting the optimal parameters, the final charging load is calculated: (14) In the formula, The optimal charging load characteristic weighting coefficient; A profile corresponding to the charging load.

4. The method for assessing the charging load carrying capacity of electric vehicles in a distribution network considering spatiotemporal profiles according to claim 3, characterized in that, In S2, the available capacity of the existing charging station area transformers is output, specifically as follows: Based on the premise of ensuring the safe operation of the transformer, the rated capacity of the transformer should be determined first. The total installed charging pile capacity of the substation is [not specified]. Then, the total charging power for a certain period of time is obtained through the actual operating data of the charging station. For newly added charging stations Next, set the safety upper limit factor for the rated capacity of the transformer. The remaining available capacity of the computing platform This refers to the final available capacity of the transformer.

5. The method for assessing the charging load carrying capacity of electric vehicles in a distribution network considering spatiotemporal profiles according to claim 4, characterized in that, Step S3 is as follows: S3-1. Obtain typical charging load profiles from S1 and geographical information of the functional areas to which existing charging stations belong, and define the sample set: Each existing charging station serves as one sample point. ,in: For the geographical coordinates of existing charging stations, The load percentage of image j in the i-th existing charging station satisfies , S3-2. Perform Kriging interpolation on the image weight parameters to obtain the fitting parameters for the load curve of the newly added charging station: Extracting a single image sample set: Weight of image 1 Sample set b. Calculate spatial distance: 1) Distance between sample points: 2) Distance between sample points and newly added stations: c. Fit a spherical variogram. The variogram needs to be fitted to a continuous theoretical model to calculate the correlation at any distance. The formula for the spherical model is: （15） In the formula, The value of the nugget reflects the measurement error; The sill value-null value reflects spatial structural variation; The range is variable, reflecting the extent of spatial correlation. d Solving for weights : （16） It should satisfy the property of being unbiased, that is , e-calculates the profile weight of newly added sites: f interpolates images 2-9 respectively until it yields... to ; If the sum after interpolation deviates from 1, normalization is performed. Ultimately retained For the final weight, S3-3. Generating the charging load curve for newly added charging stations, the formula is as follows: （17） In the formula, The time-by-time load curve of the j-th typical portrait; The weight of the j-th image obtained by interpolation; This represents the total number of newly added station piles. S3-4. Calculate the available capacity of newly added charging stations, using the formula according to... To calculate.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for assessing the charging load capacity of electric vehicles in a distribution network, taking into account spatiotemporal profiles, as described in any one of claims 1 to 5.

7. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the computer instruction is executed by the processor, it implements the method for assessing the charging load capacity of electric vehicles in a distribution network, taking into account spatiotemporal profiles, as described in any one of claims 1-5.

Images