Method, system and medium for identifying response potential of electric vehicle based on behavior backstepping

Abstract

The present application relates to the technical field of power demand response, and particularly relates to an electric vehicle response potential identification method and system based on behavior backstepping, and a medium, the method comprising: obtaining historical charging behavior data of electric vehicle users and associated electricity price data and environmental data, and constructing a demand response evaluation dataset; constructing a demand response potential evaluation model based on the demand response evaluation dataset; normalizing and weighting the fusion of each dimension representation result output by the demand response potential evaluation model to obtain a demand response potential comprehensive score of each electric vehicle user; and based on the demand response potential comprehensive score, identifying the electric vehicle users in layers, and determining target response users, aggregated resource sets or call priorities in a power network demand response event. Through the present application, the problem that the prior art cannot accurately identify the comprehensive call potential of electric vehicle users in demand response is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of power demand response technology, and in particular to a method, system, and medium for identifying the response potential of electric vehicles based on behavioral backpropagation. Background Technology

[0002] With the expansion of new energy grid connection and the rapid popularization of electric vehicles, the demand for flexible adjustment resources in the power system is constantly increasing. Demand response has become an important technical means to ensure the safe and stable operation of the power grid, improve peak shaving and valley filling capabilities, and promote the consumption of new energy. Electric vehicles have both controllable load and energy storage resource attributes, and can participate in grid regulation through orderly charging, load reduction, and even vehicle-grid interaction, thus having high application value in demand response.

[0003] Current technologies typically assess and screen electric vehicle (EV) users' responsiveness based on single indicators or simple statistical characteristics such as historical charging frequency, charging amount, and participation frequency. While these methods can reflect user behavior characteristics to some extent, they generally suffer from problems such as limited evaluation dimensions, insufficient mining of behavioral temporal patterns, and difficulty in comprehensively reflecting user stability, adjustable capacity, activity level, scheduling adaptability, and subjective participation tendencies. This makes it difficult to accurately identify users with truly high demand response potential and adapt to resource allocation needs in different demand response scenarios. Therefore, there is an urgent need for a method that can comprehensively and accurately identify the demand response potential of EV users based on historical charging behavior.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of this disclosure and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a method, system, and medium for identifying the response potential of electric vehicles based on behavior back-inference, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for identifying the response potential of electric vehicles based on behavioral back-inference, the method comprising: Acquire historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, and construct a demand response assessment dataset; Based on the aforementioned demand response assessment dataset, a demand response potential evaluation model is constructed by combining behavioral sequence modeling and multidimensional behavioral analysis to characterize the response stability, response capacity, response speed, response flexibility, and response willingness of electric vehicle users. The characteristic results of each dimension output by the demand response potential evaluation model are normalized and weighted and fused to obtain the comprehensive score of demand response potential for each electric vehicle user. Based on the comprehensive score of demand response potential, electric vehicle users are stratified and identified, and the target response users, aggregated resource sets, or call priorities in power network demand response events are determined.

[0007] Furthermore, a demand response potential evaluation model is constructed, including: Based on the aforementioned demand response assessment dataset, extract time-series features, capacity features, activity features, distribution features, and subjective control features that characterize the demand response behavior of electric vehicle users; The response stability, response capacity, response speed, response flexibility, and response willingness are constructed based on the time-series characteristics, capacity characteristics, activity characteristics, distribution characteristics, and subjective control characteristics, respectively. The response stability, response capacity, response speed, response flexibility, and response willingness are normalized so that the characterization results of each dimension fall within a preset scoring range. The response stability, response capacity, response speed, response flexibility, and response willingness are weighted and integrated according to preset weights to obtain a comprehensive score of the demand response potential, wherein the sum of the preset weights is 1.

[0008] Furthermore, constructing the response stability includes: A time-series sample is constructed based on the daily charging volume sequence in the historical charging behavior data, and the time-series sample is input into a bidirectional gated recurrent unit model for training. The user behavior regularity score is determined based on the error between the prediction results and the actual results of the bidirectional gated cyclic unit model, so as to characterize the fitting difficulty of the bidirectional gated cyclic unit model to the daily charging volume sequence. The average response duration normalization value is determined based on the single charging duration in the historical charging behavior data to characterize the user's ability to continuously provide response load. The ability to respond to severe weather is determined based on the temperature and precipitation data in the environmental data, so as to characterize the stability of the user's response under environmental fluctuation conditions. The participation rate during high-price periods is determined based on the electricity price during the initial charging period in the electricity price data, in order to characterize the user's robustness to price fluctuations; The response stability is obtained by weighting and fusing the behavioral regularity score, the normalized value of the average response duration, the severe weather response capability, and the participation rate during high electricity price periods.

[0009] Furthermore, constructing the response capacity includes: The average response power is determined based on the charging power of each charging session in the historical charging behavior data. The average response load capacity is determined based on the average response power consumption and the preset normalized upper limit, so as to characterize the relative response scale of the user among all users. The battery capacity is calculated by back-calculating the maximum daily charging amount from the historical charging behavior data, wherein the back-calculation process of the battery capacity adopts the assumption that the remaining power at the start of charging is a preset proportion. The battery capacity utilization rate is determined based on the average response power and the battery capacity to characterize the extent to which a user releases adjustable capacity during a typical response behavior. The average response load capacity and the battery capacity utilization rate are weighted and fused to obtain the response capacity, wherein the weighting parameters of the weighting and fusion are used to adjust the proportion of absolute response scale and relative utilization efficiency in the response capacity.

[0010] Furthermore, constructing the response speed includes: A preset evaluation period is set, and the preset evaluation period is divided into multiple continuous sliding windows; The number of active response days within each sliding window is counted to obtain the level of response activity of electric vehicle users in each sliding window; The average percentage of active response days within the preset evaluation period is determined based on the number of active response days within each sliding window. The response speed is derived from the average percentage of active response days to characterize the frequency and rhythm of user responses over a longer timescale.

[0011] Furthermore, building the aforementioned response flexibility includes: Based on the historical charging behavior data, the discrete behavioral distribution of electric vehicle users in the time, location and capacity dimensions are statistically analyzed. Calculate the corresponding Shannon entropy value based on the discrete behavior distributions in the time dimension, the location dimension, and the capacity dimension, respectively. The representation results of each dimension are normalized based on the Shannon entropy value and the maximum entropy value of the corresponding dimension to obtain the representation results of time flexibility, location flexibility and capacity flexibility. The time flexibility representation results, location flexibility representation results, and capacity flexibility representation results are fused according to preset weights to obtain the response flexibility, wherein the weight corresponding to the location flexibility representation results is higher than the weight corresponding to the time flexibility representation results and the capacity flexibility representation results.

[0012] Furthermore, constructing the response intention includes: Count the total number of charging records for electric vehicle users; The number of times the charging record was terminated by the user and not due to the battery being fully charged or by system control is counted. The response intention is determined based on the ratio between the number of times charging is actively terminated and the total number of charging records, so as to characterize the user's tendency to actively control the charging process and the degree of cooperation.

[0013] Furthermore, electric vehicle users are stratified and identified, including: Based on the target demand response scenario, configure the fusion weights corresponding to the response stability, response capacity, response speed, response flexibility, and response willingness; When the target demand response scenario focuses on response scale, the weighting of the average response load capacity in the response capacity is increased; When the target demand response scenario focuses on resource release efficiency or user participation tendency, the weight ratio corresponding to battery capacity utilization in the response capacity is increased. The comprehensive score of demand response potential is recalculated based on the configured fusion weights, and electric vehicle users are sorted, stratified, or aggregated according to the comprehensive score of demand response potential. Based on the results of sorting, stratification, or aggregation, determine the target response users, candidate aggregated resource sets, or call priorities in emergency peak shaving scheduling, daily load shifting, price response services, virtual power plant user aggregation, or grid flexibility testing.

[0014] A behavior-based electric vehicle response potential identification system, the system comprising: The data acquisition module acquires historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, and constructs a demand response assessment dataset. The model building module, based on the demand response assessment dataset, uses a combination of behavioral sequence modeling and multidimensional behavioral analysis to construct a demand response potential evaluation model to characterize the response stability, response capacity, response speed, response flexibility, and response willingness of electric vehicle users. The potential scoring module normalizes and weights the output of the demand response potential evaluation model for each dimension, and then integrates them to obtain a comprehensive demand response potential score for each electric vehicle user. The hierarchical identification module identifies electric vehicle users based on a comprehensive score of demand response potential and determines the target responding users, aggregated resource sets, or call priorities in power network demand response events.

[0015] A computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, enable a behavior-based method for identifying the response potential of electric vehicles.

[0016] The technical solution of this invention can achieve the following technical effects: By collecting historical charging behavior data of electric vehicle users and related data such as electricity prices and environment, a comprehensive scoring model is constructed that integrates response stability, capacity, speed, flexibility and willingness. This model can quantitatively assess the user demand response potential and identify high-potential users, effectively solving the problem that existing technologies cannot accurately identify the comprehensive utilization potential of electric vehicle users in demand response.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the method for identifying the response potential of electric vehicles based on behavior back-inference. Figure 2 Heatmap of scoring weights; Figure 3 Diagram of the Bi-GRU structure; Figure 4 For weight Recommended heatmap. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1; like Figure 1 As shown, this application provides a method for identifying the response potential of electric vehicles based on behavior back-inference, the method including: S10: Obtain historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, and construct a demand response assessment dataset; S20: Based on the demand response assessment dataset, a demand response potential evaluation model is constructed by combining behavioral sequence modeling and multidimensional behavioral analysis to characterize the response stability, response capacity, response speed, response flexibility and response willingness of electric vehicle users. S30: Normalize and weight the representation results of each dimension output by the demand response potential evaluation model to obtain the comprehensive score of demand response potential for each electric vehicle user. S40: Based on the comprehensive score of demand response potential, electric vehicle users are stratified and identified, and the target response users, aggregated resource sets, or call priorities are determined in the power network demand response events.

[0023] Specifically, in one embodiment, a method for identifying potential electric vehicle users for power grid demand response is provided. This method can be deployed in a grid-side demand response management platform, an aggregator platform, or a vehicle-to-grid interaction management system. First, historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, are obtained. The historical charging behavior data includes at least the charging start time, charging end time, charging amount, charging location, number of charging sessions, and whether the user actively terminated charging. The electricity price data includes at least the time-of-use electricity price information corresponding to each charging period. The environmental data includes at least temperature and precipitation information. The historical charging behavior data, electricity price data, and environmental data are then correlated and matched according to timestamps to construct a demand response assessment dataset. Based on a demand response assessment dataset, a demand response potential evaluation model is constructed using a combination of behavioral sequence modeling and multidimensional behavioral analysis to characterize the response stability, response capacity, response speed, response flexibility, and response willingness of electric vehicle users. Response stability is obtained by modeling the daily charging volume sequence of users. In this embodiment, a bidirectional gated cyclic unit model is used to train the daily charging volume sequence, and the model prediction error reflects the regularity of user charging behavior. Simultaneously, the average response duration, response capability under severe weather conditions, and participation rate during high electricity price periods are combined to comprehensively characterize response stability. Response capacity is determined based on the average response amount and battery capacity utilization rate in the user's historical charging behavior, where the user's battery capacity can be estimated by back-calculating from the maximum daily charging volume. Response speed is obtained by using a sliding window statistical analysis of response activity within a preset assessment period to characterize the frequency and activity level of user responses over a longer time scale. Response flexibility is determined by statistically analyzing the distribution of user behavior across time, location, and capacity dimensions, and using normalized Shannon entropy for calculation to characterize the user's adaptability under different scheduling scenarios. Response willingness is determined based on the proportion of user-initiated charging termination behavior in all charging behaviors, reflecting the user's tendency to actively cooperate with demand response. After obtaining the response stability, response capacity, response speed, response flexibility, and response willingness, the representation results of each dimension are normalized to fall into a unified scoring range. Weight parameters for each dimension are set according to the target demand response scenario, and the representation results of each dimension are weighted and fused to obtain a comprehensive score for the demand response potential of each electric vehicle user. The weight parameters can be adjusted according to different scheduling objectives. When the demand response scenario focuses more on the scale of load release, the weight related to response capacity is increased. When the demand response scenario focuses more on user cooperation, resource release efficiency, or multi-scenario adaptability, the weights corresponding to response willingness, response flexibility, or related utilization indicators are increased, thus enabling the demand response potential evaluation model to have scenario adaptability. Furthermore, electric vehicle users are ranked and stratified based on a comprehensive demand response potential score. Users with higher comprehensive scores are identified as high-potential demand response users, and the target response users, candidate aggregated resource sets, or call priorities are determined accordingly in power network demand response events. Specifically, in application scenarios such as emergency peak shaving, daily load shifting, price response services, virtual power plant user aggregation, and grid flexibility testing, user resources can be screened, aggregated, and called based on the comprehensive demand response potential score results to improve the accuracy of demand response object identification and the effectiveness of resource scheduling.

[0024] The technical solution of this invention collects historical charging behavior data of electric vehicle users and related data such as electricity prices and environment, and constructs a comprehensive scoring model that integrates response stability, capacity, speed, flexibility and willingness. This model quantitatively evaluates the user's demand response potential and identifies high-potential users, effectively solving the problem that existing technologies cannot accurately identify the comprehensive call-up potential of electric vehicle users in demand response.

[0025] Furthermore, constructing a demand response potential evaluation model includes: Based on the demand response assessment dataset, time-series features, capacity features, activity features, distribution features, and subjective control features characterizing the demand response behavior of electric vehicle users are extracted; Based on time-series characteristics, capacity characteristics, activity characteristics, distribution characteristics, and subjective control characteristics, response stability, response capacity, response speed, response flexibility, and response willingness are constructed respectively. The response stability, response capacity, response speed, response flexibility, and response willingness are normalized so that the characterization results of each dimension fall within the preset scoring range. The response stability, response capacity, response speed, response flexibility, and response willingness are weighted and integrated according to preset weights to obtain a comprehensive score of demand response potential, where the sum of the preset weights is 1.

[0026] As a preferred embodiment of the above, a total score is constructed. Current user response activity rating It consists of the following five parts: ; in, Indicates response stability. Indicates response capacity. Indicates response speed. Indicates responsiveness. Indicates willingness to respond, each item and ; in, These are the corresponding weighting coefficients for response stability, response capacity, response speed, response flexibility, and response willingness, respectively, and are referenced. Figure 2 The meanings of the five scoring dimensions are shown in Table 1, based on the weighting of different strategy scenarios: .

[0027] Furthermore, building response stability includes: Time-series samples are constructed based on the daily charging volume sequence in historical charging behavior data, and the time-series samples are input into a bidirectional gated recurrent unit model for training. The error between the prediction results and the actual results of the bidirectional gated cyclic unit model is used to determine the regularity score of user behavior, so as to characterize the difficulty of fitting the bidirectional gated cyclic unit model to the daily charging volume sequence. The average response duration is normalized based on the duration of a single charge in historical charging behavior data to characterize the user’s ability to continuously provide response load. Determine severe weather response capability based on temperature and precipitation data in environmental data to characterize the stability of user response under fluctuating environmental conditions; The participation rate during high-price periods is determined based on the electricity price at the start of charging in the electricity price data, in order to characterize the user's robustness to price fluctuations; The response stability is obtained by weighting and integrating the behavioral regularity score, the normalized value of the average response duration, the response capability in severe weather, and the participation rate during high electricity price periods.

[0028] As a preferred embodiment of the above, calculation : ; in, Indicates the score for behavioral regularity. The score represents the duration of the response. The score indicates the ability to respond to severe weather. This indicates that periods with high electricity prices participate in the scoring, and the weights of each parameter satisfy the following: ; The default values ​​can be set to 0.4, 0.2, 0.2, and 0.2. (1) Determine The Bi-GRU structure diagram is as follows: Figure 3 As shown; Given user Daily charging sequence Samples are constructed using a sliding window: The Bi-GRU model outputs predicted values. The training objective is to minimize the mean squared error. ; In the formula, Let represent the loss value for user u, T represent the total time length or total sample length of the user's charging behavior sequence, L represent the sliding window length used when constructing the time series samples, and t represent the time index during the sliding window movement. This represents the true value corresponding to the (t+L)th time. After model training, the final root mean square error (RMSE) is extracted as the fitting difficulty of the user sequence. ; This error reflects whether the model can capture user behavior patterns; Considering that the smaller the RMSE, the more regular the behavior, and the easier the model is to learn, the following mapping is constructed: ; in It is a configurable maximum error threshold that ensures the final score falls within the [0, 1] range and has good discriminative power and interpretability. (2) Determine the average response duration: ; Let be the minimum time for the i-th charging action, and be the minimum time for the i-th charging action. Total number of charging cycles The normalized reference upper limit time is truncated to within 1. The higher the index, the longer the user can maintain the response in each response, and the stronger the ability to continuously supply load. (3) Determine the ability to respond to severe weather ; in Total number of charging cycles This is an indicator function that takes the value 1 when the condition is met. and Let be the temperature and precipitation during the i-th charge, respectively; To determine the threshold for severe weather, The maximum reference ratio (normalized upper limit) indicates that the user has a stronger ability to resist environmental fluctuations and better stability.

[0029] (4) Determine the participation rate during periods of high electricity prices: ; in Total number of charging cycles Let $i$ be the electricity price corresponding to the starting time period of the $i$-th charging. The high price threshold for electricity. This is a normalized reference value; the fact that users still participate in the response during peak electricity price periods indicates that their response is less affected by price drivers and has stronger time stability.

[0030] Furthermore, building response capacity includes: The average response power is determined based on the power of each charge in historical charging behavior data. The average response load capacity is determined based on the average response power consumption and the preset normalized upper limit, so as to characterize the relative response scale of the user among all users. Battery capacity is inferred from the maximum daily charging amount in historical charging behavior data. The inference process assumes that the remaining power at the start of charging is a preset proportion. Battery capacity utilization is determined based on average response charge and battery capacity to characterize the extent to which a user releases adjustable capacity during a typical response behavior. The average response load capacity and battery capacity utilization rate are weighted and fused to obtain the response capacity. The weight parameters of the weighted fusion are used to adjust the proportion of absolute response scale and relative utilization efficiency in the response capacity.

[0031] As a preferred embodiment of the above, calculation : Calculate the average response charge: ; in This represents the amount of electricity charged by the user during the i-th charge, and N is the total number of charges by the user. To reflect its relative response size among all users, a normalization process is used to construct the score: ; in This represents the user's average response capacity score. It is the upper limit of normalization; Since the user's battery capacity is difficult to obtain directly, we will use the maximum daily charging amount to estimate the battery capacity. Assuming that each charge starts when the battery is at 20% charge, the estimated battery capacity is: ; in This represents the user's maximum daily charging volume during the statistical period, and is used to define the response utilization score: ; This metric reflects whether a user fully utilizes their adjustable capacity during a typical response; ultimately, the two metrics are combined to construct a comprehensive capacity score. ; Among them, weight parameters Weighting parameters are used to control the weighting ratio between absolute capacity and relative utilization. Selection reference Figure 4 , Figure 4 The interpretation of the rows and columns of the heatmap is shown in Tables 2 and 3: ; The capacity score S-capacity comprehensively considers the user's average response power and battery capacity utilization rate, and the weighting parameter β controls the relative importance of the two in the final score. Its selection should be adjusted in combination with the specific grid dispatching objectives and service scenarios to achieve scenario adaptability and resource optimization capability of the scoring strategy. Generally, when the target scenario focuses on maximizing the response scale, such as emergency peak shaving and load suppression, high-load users should be given priority, and a larger weight is recommended, such as β=0.7~0.9. If the scenario focuses more on user participation or resource release, such as daily frequency regulation or price-driven demand response services, emphasizing full utilization of batteries rather than absolute capacity, the utilization rate can be increased, and β=0.3~0.5 is recommended. If the scheduling goal needs to balance scale and initiative, such as user stratification and aggregation when building a virtual power plant, a balanced setting of β=0.5 can be adopted. Furthermore, in actual system deployment, β can be used as a scheduling strategy parameter, which can be dynamically set by the power operator or learned and optimized through feedback from historical scheduling effects, thereby improving the flexibility and accuracy of the capacity scoring system.

[0032] Furthermore, building responsiveness includes: Set a preset evaluation period and divide the preset evaluation period into multiple continuous sliding windows; The number of active response days within each sliding window is counted to obtain the level of activity of electric vehicle users in each sliding window; The average percentage of active response days within a preset evaluation period is determined based on the number of active response days in each sliding window. Response speed is derived from the percentage of average active days to characterize the frequency and rhythm of user responses over a longer timescale.

[0033] As a preferred embodiment of the above, calculation : Because users' daily behavior may fluctuate occasionally, using only short-term behavioral data (such as the number of response days in the past 30 days) may lead to scoring bias and fail to reflect users' true behavioral rhythm and response habits. Therefore, a sliding window statistical method is introduced to characterize user response activity over a longer time scale. Let the study period be T = 360 days, the sliding window length be W = 30 days, and the number of sliding windows be K = T / W = 12. The number of active response days in the k-th window is... If the maximum value is 30, then the response speed score is defined as: ; The formula represents the percentage of active response days per month for users over the past year prior to the current evaluation date. Since the denominator is fixed at 360, the score is naturally normalized to [0,1] without the need for additional scaling.

[0034] Furthermore, building responsiveness includes: Based on historical charging behavior data, the discrete behavioral distribution of electric vehicle users in the time, location and capacity dimensions were statistically analyzed. Calculate the corresponding Shannon entropy value based on the discrete behavioral distributions in the time, location, and capacity dimensions, respectively. The representation results of each dimension are normalized based on the Shannon entropy value and the maximum entropy value of the corresponding dimension to obtain the representation results of time flexibility, location flexibility and capacity flexibility. The time flexibility characterization results, location flexibility characterization results, and capacity flexibility characterization results are fused according to preset weights to obtain response flexibility, where the weight corresponding to the location flexibility characterization results is higher than the weight corresponding to the time flexibility characterization results and the capacity flexibility characterization results.

[0035] As a preferred embodiment of the above, S is calculated. 灵活性 : Normalized Shannon entropy is used to measure the uniformity of behavior distribution across each dimension. For discrete behavior distributions in a given dimension... Its entropy is defined as: ; Where N represents the total number of categories or states in the discrete behavior distribution. The higher the entropy, the more dispersed and balanced the user behavior. To unify the standard, normalization is performed: ; This makes the final value normalized to Between. The final flexibility score is defined as: ; in, The normalized entropy for users at each charging station, Normalized entropy for users in different capacity ranges (divided into 5 segments based on maximum charging capacity); The normalized entropy for the user over 5 TOU time periods; the default weights are set as follows: This weight configuration emphasizes the dominant role of spatial distribution in scheduling elasticity, while also taking into account the ability of power and time dimensions to support multi-scenario scheduling. If S 灵活性 A score close to 1 indicates that the user has participated in multiple sites, multiple capacity ranges, and multiple time periods, with a balanced distribution of behavior and strong scheduling adaptability. If the score is close to 0, it indicates that the user's behavior is highly concentrated, and the scheduling conditions for such users are limited, so their callability needs to be carefully evaluated.

[0036] Furthermore, building a willingness to respond includes: Count the total number of charging records for electric vehicle users; The number of times a charging session was terminated by the user and not due to the battery being fully charged or by system control is recorded in the charging history. The willingness to respond is determined by the ratio between the number of times charging is actively terminated and the total number of charging records, in order to characterize the user's tendency to actively control the charging process and the degree of cooperation.

[0037] As a preferred embodiment of the above, calculation : Assume the user's charging records total The number of times the event was terminated by the user (not due to the battery being fully charged or by system control) was [number missing]. The response willingness score is defined as follows: ; The rating range is: A higher score indicates that the user actively controls the charging process more frequently, and their behavior shows a greater willingness to respond. 意愿 This indicates that the user is proactive, behaves controllably, and has a clear tendency to cooperate, making them a high-value scheduling and response resource with low S 意愿 This suggests that users rely more on the system's automatic behavior, and their cooperation and adaptability may be relatively low.

[0038] Furthermore, segmenting and identifying electric vehicle users includes: Configure fusion weights corresponding to response stability, response capacity, response speed, response flexibility, and response willingness based on the target demand response scenario; When the focus is on response scale in the target demand response scenario, increase the weight ratio of the average response load capacity in the response capacity; When the target demand response scenario focuses on resource release efficiency or user participation tendency, the weight ratio of battery capacity utilization in the response capacity should be increased. The comprehensive score of demand response potential is recalculated based on the configured fusion weights, and electric vehicle users are sorted, stratified, or aggregated according to the comprehensive score of demand response potential. Based on the results of sorting, stratification, or aggregation, determine the target response users, candidate aggregated resource sets, or call priorities in emergency peak shaving scheduling, daily load shifting, price response services, virtual power plant user aggregation, or grid flexibility testing.

[0039] Example 2; Based on the same inventive concept as the electric vehicle response potential identification method based on behavior back-reasoning in the foregoing embodiments, the present invention also provides an electric vehicle response potential identification system based on behavior back-reasoning, the system comprising: The data acquisition module acquires historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, and constructs a demand response assessment dataset. The model building module, based on the demand response assessment dataset, uses a combination of behavioral sequence modeling and multidimensional behavioral analysis to construct a demand response potential evaluation model to characterize the response stability, response capacity, response speed, response flexibility, and response willingness of electric vehicle users. The potential scoring module normalizes and weights the output of the demand response potential evaluation model for each dimension, and then integrates them to obtain a comprehensive demand response potential score for each electric vehicle user. The hierarchical identification module identifies electric vehicle users based on a comprehensive score of demand response potential and determines the target responding users, aggregated resource sets, or call priorities in power network demand response events.

[0040] The system described above in this invention can effectively realize the electric vehicle response potential identification method based on behavior back-inference, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0041] Example 3; Based on the same inventive concept as the behavior-back-engineering-based electric vehicle response potential identification method in the foregoing embodiments, the present invention also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, can implement the behavior-back-engineering-based electric vehicle response potential identification method.

[0042] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for identifying the response potential of electric vehicles based on behavioral back-inference, characterized in that, The method includes: Acquire historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, and construct a demand response assessment dataset; Based on the aforementioned demand response assessment dataset, a demand response potential evaluation model is constructed by combining behavioral sequence modeling and multidimensional behavioral analysis to characterize the response stability, response capacity, response speed, response flexibility, and response willingness of electric vehicle users. The characteristic results of each dimension output by the demand response potential evaluation model are normalized and weighted and fused to obtain the comprehensive score of demand response potential for each electric vehicle user. Based on the comprehensive score of demand response potential, electric vehicle users are stratified and identified, and the target response users, aggregated resource sets, or call priorities in power network demand response events are determined.

2. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to claim 1, characterized in that, Construct a demand response potential assessment model, including: Based on the aforementioned demand response assessment dataset, extract time-series features, capacity features, activity features, distribution features, and subjective control features that characterize the demand response behavior of electric vehicle users; The response stability, response capacity, response speed, response flexibility, and response willingness are constructed based on the time-series characteristics, capacity characteristics, activity characteristics, distribution characteristics, and subjective control characteristics, respectively. The response stability, response capacity, response speed, response flexibility, and response willingness are normalized so that the characterization results of each dimension fall within a preset scoring range. The response stability, response capacity, response speed, response flexibility, and response willingness are weighted and integrated according to preset weights to obtain a comprehensive score of the demand response potential, wherein the sum of the preset weights is 1.

3. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to claim 2, characterized in that, Constructing the response stability includes: A time-series sample is constructed based on the daily charging volume sequence in the historical charging behavior data, and the time-series sample is input into a bidirectional gated recurrent unit model for training. The user behavior regularity score is determined based on the error between the prediction results and the actual results of the bidirectional gated cyclic unit model, so as to characterize the difficulty of fitting the bidirectional gated cyclic unit model to the daily charging volume sequence. The average response duration normalization value is determined based on the single charging duration in the historical charging behavior data to characterize the user's ability to continuously provide response load. The ability to respond to severe weather is determined based on the temperature and precipitation data in the environmental data, so as to characterize the stability of the user's response under environmental fluctuation conditions. The participation rate during high-price periods is determined based on the electricity price during the charging start period in the electricity price data to characterize the user's robustness to price fluctuations; The response stability is obtained by weighting and fusing the behavioral regularity score, the normalized value of the average response duration, the severe weather response capability, and the participation rate during high electricity price periods.

4. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to claim 2, characterized in that, Constructing the response capacity includes: The average response power is determined based on the charging power of each charging session in the historical charging behavior data. The average response load capacity is determined based on the average response power consumption and the preset normalized upper limit, so as to characterize the relative response scale of the user among all users. The battery capacity is calculated by back-calculating the maximum daily charging amount from the historical charging behavior data, wherein the back-calculation process of the battery capacity adopts the assumption that the remaining power at the start of charging is a preset proportion. The battery capacity utilization rate is determined based on the average response power and the battery capacity to characterize the extent to which a user releases adjustable capacity during a typical response behavior. The average response load capacity and the battery capacity utilization rate are weighted and fused to obtain the response capacity, wherein the weighting parameters of the weighting and fusion are used to adjust the proportion of absolute response scale and relative utilization efficiency in the response capacity.

5. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to claim 2, characterized in that, The aforementioned response speed is constructed by: A preset evaluation period is set, and the preset evaluation period is divided into multiple continuous sliding windows; The number of active response days within each sliding window is counted to obtain the level of response activity of electric vehicle users in each sliding window; The average percentage of active response days within the preset evaluation period is determined based on the number of active response days within each sliding window. The response speed is derived from the average percentage of active response days to characterize the frequency and rhythm of user responses over a longer timescale.

6. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to claim 2, characterized in that, Building the aforementioned response flexibility includes: Based on the historical charging behavior data, the discrete behavioral distribution of electric vehicle users in the time, location and capacity dimensions are statistically analyzed. Calculate the corresponding Shannon entropy value based on the discrete behavior distributions in the time dimension, the location dimension, and the capacity dimension, respectively. The representation results of each dimension are normalized based on the Shannon entropy value and the maximum entropy value of the corresponding dimension to obtain the representation results of time flexibility, location flexibility and capacity flexibility. The time flexibility representation results, location flexibility representation results, and capacity flexibility representation results are fused according to preset weights to obtain the response flexibility, wherein the weight corresponding to the location flexibility representation results is higher than the weight corresponding to the time flexibility representation results and the capacity flexibility representation results.

7. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to claim 2, characterized in that, Constructing the stated response intention includes: Count the total number of charging records for electric vehicle users; The number of times the charging record was terminated by the user and not due to the battery being fully charged or by system control is counted. The response intention is determined based on the ratio between the number of times charging is actively terminated and the total number of charging records, so as to characterize the user's tendency to actively control the charging process and the degree of cooperation.

8. The method for identifying the response potential of electric vehicles based on behavioral back-inference according to any one of claims 2 to 7, characterized in that, Segmenting and identifying electric vehicle users, including: Based on the target demand response scenario, configure the fusion weights corresponding to the response stability, response capacity, response speed, response flexibility, and response willingness; When the target demand response scenario focuses on response scale, the weighting of the average response load capacity in the response capacity is increased; When the target demand response scenario focuses on resource release efficiency or user participation tendency, the weight ratio corresponding to battery capacity utilization in the response capacity is increased. The comprehensive score of demand response potential is recalculated based on the configured fusion weights, and electric vehicle users are sorted, stratified, or aggregated according to the comprehensive score of demand response potential. Based on the results of sorting, stratification, or aggregation, determine the target response users, candidate aggregated resource sets, or call priorities in emergency peak shaving scheduling, daily load shifting, price response services, virtual power plant user aggregation, or grid flexibility testing.

9. A system for identifying the response potential of electric vehicles based on behavioral back-inference, characterized in that, The system includes: The data acquisition module acquires historical charging behavior data of electric vehicle users, as well as electricity price data and environmental data associated with the historical charging behavior data, and constructs a demand response assessment dataset. The model building module, based on the demand response assessment dataset, uses a combination of behavioral sequence modeling and multidimensional behavioral analysis to construct a demand response potential evaluation model to characterize the response stability, response capacity, response speed, response flexibility, and response willingness of electric vehicle users. The potential scoring module normalizes and weights the output of the demand response potential evaluation model for each dimension, and then integrates them to obtain a comprehensive demand response potential score for each electric vehicle user. The hierarchical identification module identifies electric vehicle users based on a comprehensive score of demand response potential and determines the target responding users, aggregated resource sets, or call priorities in power network demand response events.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, can implement the electric vehicle response potential identification method based on behavior backpropagation as described in any one of claims 1-7.

Images