Double-objective optimization method for electric vehicle charging scheduling based on incentive game

Abstract

The invention discloses a dual-objective optimization method for electric vehicle charging scheduling based on an incentive game, and relates to the technical field of electric vehicle charging scheduling, and the method comprises the following steps: S100, constructing a user behavior dynamic response recognition model, obtaining historical charging behavior data and incentive trigger time nodes, extracting response changes of a user under different incentive conditions, and obtaining a user behavior dynamic response recognition model; and generating a response feature vector. According to the method, dynamic allocation and differentiated scheduling of excitation resources are realized through construction of user response feature vectors and response hierarchy division; behavior feedback and load data are introduced to construct a response feedback matrix, strategy effect evaluation and adaptive optimization are supported, a recognition-guide-feedback-correction closed-loop control mechanism is formed, the excitation response efficiency, the power grid dispatching adaptability and the system intelligence level are improved, and the power grid operation safety and the user charging effectiveness are effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle charging scheduling, and in particular to a dual-objective optimization method for electric vehicle charging scheduling based on incentive game. Background Art

[0002] "Dual-objective optimization of electric vehicle charging scheduling based on incentive games" involves leveraging incentive mechanisms from game theory to design strategies that guide users to independently select appropriate charging times or power levels, ensuring their willingness to charge while alleviating peak load pressure on the power grid. This approach models electric vehicle charging scheduling as a multi-player game process with dual optimization objectives: maximizing electric vehicle user satisfaction or revenue, and minimizing grid-side load fluctuations or operating costs. By designing incentive strategies (such as dynamic pricing and subsidy mechanisms) and establishing game models (such as non-cooperative games and Stackelberg games), a coordinated optimization between regulating user behavior and improving system efficiency is achieved, ensuring that the electric vehicle charging process satisfies individual interests while also ensuring the safe and efficient operation of the overall power system.

[0003] The existing technology has the following deficiencies: In the charging scheduling process for large-scale electric vehicle users, existing incentive-based game-based regulation mechanisms generally rely on unified price signals or incentive strategies to guide users to concentrate charging during specific time periods to achieve load shaving. However, in actual operation, multiple users simultaneously perceive the same incentive information and make "game-optimal" responses based on the principle of maximizing individual benefits. This often leads to a sudden influx of charging demand during the incentive period, forming new concentrated load peaks. This group behavior lacks scheduling-level time-sharing guidance and load peak regulation mechanisms, which can easily induce secondary load peak concentration, leading to deviations in grid load forecasts and imbalances in scheduling resource allocation.

[0004] Furthermore, under secondary load peak conditions, substations and distribution branches within the regional power grid may exceed their designed capacity and become overloaded, triggering abnormal operating conditions such as voltage drops and current surges. In severe cases, this can cause the distribution system's protective circuits to trip, or even lead to equipment failures such as local transformer insulation breakdown or thermal burnout. Furthermore, system-level power regulation mechanisms struggle to achieve rapid adaptive control of supply and demand, which can easily lead to real-time frequency fluctuations, voltage disturbances, and misaligned energy storage system responses. This can disrupt the dynamic balance of the entire charging scheduling system, severely impacting both the operational safety of the power system and the charging reliability of electric vehicle users.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide a dual-objective optimization method for electric vehicle charging scheduling based on incentive game. By constructing user response feature vectors and response level division, dynamic allocation and differentiated scheduling of incentive resources are realized; behavioral feedback and load data are introduced to construct a response feedback matrix to support strategy effect evaluation and adaptive optimization, forming an identification-guidance-feedback-correction closed-loop control mechanism, thereby improving the incentive response efficiency, grid scheduling adaptability and system intelligence level, effectively ensuring the safety of grid operation and user charging utility, and solving the problems in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a dual-objective optimization method for electric vehicle charging scheduling based on incentive game, comprising the following steps: S100: Build a user behavior dynamic response recognition model, obtain historical charging behavior data and incentive triggering time nodes, extract user response changes under different incentive conditions, and generate a response feature vector; S200, grouping users based on response feature vectors to form multiple response level groups with different incentive sensitivity gradients, and setting corresponding incentive control priorities; S300, based on the response level groups and the incentive control priorities, combined with the response characteristics of each response level group and the current power grid operation status, generating a distributed incentive scheduling table; S400: When a user initiates a charging connection request, differentiated incentive information is pushed based on the user's response level group and the parameters in the distributed incentive scheduling table to guide the user to generate an access path sequence; S500, when a user performs a charging task according to an access path sequence, collecting the user's charging behavior data and grid load status data for the corresponding period, and constructing a feedback matrix between the user's path behavior and the grid response status; S600, based on the feedback matrix, identifies the degree of deviation between user behavior and recommended paths, as well as the difference between grid status and scheduling expectations. It dynamically corrects the incentive amplitude, access period, and control strategy parameters, updates the scheduling table, and simultaneously optimizes the user behavior identification model to achieve closed-loop adaptive optimization control.

[0008] Preferably, step S100 includes: Obtain historical charging behavior data and incentive triggering time nodes of electric vehicle users, and construct a behavior-incentive interaction data sequence; Extract indicator data on user behavior changes before and after incentives, including behavioral fluctuation indicators, relative difference ratios, and incentive window delays; Based on the extracted indicator data, the response feature vector is constructed, normalized coding is adopted and a covariance weight adjustment mechanism is introduced to enhance feature discrimination; During subsequent charging operations, the values ​​of each dimension of the response feature vector are continuously updated to achieve dynamic correction and adaptive learning of the user behavior model.

[0009] Preferably, the covariance weight adjustment mechanism includes: Calculate the covariance matrix between the dimensions in multiple user response feature vectors to determine the degree of linear correlation between the feature dimensions; High weights are assigned to feature dimensions whose covariance values ​​are lower than a set threshold to enhance their contribution to distinguishing user behavior patterns. Assign low weights to feature dimensions whose covariance values ​​are higher than a set threshold to suppress their interference in identifying user behavior features; The feature vector after weight adjustment is reconstructed to improve the clustering clarity and grouping effectiveness of user response features in the feature space.

[0010] Preferably, step S200 includes: Perform unsupervised cluster analysis on the weighted user response feature vectors to form multiple behavioral response categories based on the feature space distribution; The average response eigenvector of each behavioral response category was calculated to determine its incentive sensitivity index; Divided into multiple response level groups based on the incentive sensitivity index; Set incentive control priorities for each response level group for subsequent incentive resource allocation and policy push timing configuration.

[0011] Preferably, step S300 includes: Based on the historical response characteristics of the response level group and the current power grid operation status, a dispatch scenario recognition dataset is constructed; Based on the scheduling scenario identification dataset, combined with the user's historical response function and target scheduling requirements, the incentive amplitude parameters of each response level group are determined; Analyze the load fluctuation curve of the target area power grid and set the charging access period parameters for each response level group; Configure the corresponding control intervention strategy parameters to form a distributed incentive scheduling table including incentive amplitude, access period and intervention strategy.

[0012] Preferably, step S400 includes: When an electric vehicle user initiates a charging connection request, it identifies the response level group to which it belongs and extracts the corresponding distributed incentive scheduling table parameters; Generate a differentiated incentive strategy with the lowest incentive cost and the largest behavioral adjustment range based on the distributed incentive schedule parameters and the current grid load status; Push incentive information including incentive amplitude, access time window, access location and validity period to user terminals; After the user selects, his access path sequence is registered, and the current path is used as a behavior execution record for subsequent behavior feedback and policy iteration analysis.

[0013] Preferably, step S500 includes: Obtain the charging behavior data of electric vehicle users during the execution of the access path sequence and bind the user ID and path number; Collect the operating status data of the corresponding grid node and its adjacent area during the user's charging period, and perform time alignment and anomaly elimination processing; Construct a response feedback matrix with user path behavior as row vector and grid operation status as column vector; The response feedback matrix is ​​input into the data feedback engine for incentive strategy evaluation and control parameter update analysis.

[0014] Preferably, step S600 includes: Periodically analyze the response feedback matrix to identify the degree of deviation between user behavior responses and recommended access paths, as well as the difference between grid load status and dispatch expectations; Determine the incentive strategy correction conditions based on the identified deviation degree and difference information; When the correction conditions are met, the incentive amplitude parameters, access period parameters and control intervention strategy parameters are dynamically adjusted, and the distributed incentive scheduling table is updated; Integrate the newly added user behavior data with the original training data, optimize the user behavior recognition model, and update the response level grouping results.

[0015] In the above technical solution, the technical effects and advantages provided by the present invention are: The present invention constructs a user response feature vector and divides the response hierarchy, enabling the system to dynamically allocate incentive resources according to user sensitivity and form a differentiated and distributed scheduling strategy in combination with the grid operation status. At the same time, the introduction of real-time behavior feedback and grid load data to construct a response feedback matrix makes the strategy execution effect visible and quantifiable, and on this basis, realizes the adaptive iterative optimization of incentive parameters and behavior models. Ultimately, a closed-loop control mechanism with "identification-guidance-feedback-correction" as the core is formed, which takes into account the individual charging utility of electric vehicle users and the overall operational safety of the power system, effectively improves the response efficiency of incentive delivery, the dynamic adaptability of the scheduling strategy and the load balancing of the grid operation, and significantly enhances the intelligence level and robustness of the large-scale electric vehicle charging scheduling system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0017] Figure 1 This is a flow chart of the dual-objective optimization method for electric vehicle charging scheduling based on incentive game of the present invention. DETAILED DESCRIPTION

[0018] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0019] The present invention provides Figure 1 The dual-objective optimization method for electric vehicle charging scheduling based on incentive game shown in the figure includes the following steps: S100: Build a user behavior dynamic response recognition model, obtain historical charging behavior sequences and incentive triggering time nodes, extract the response changes of each electric vehicle user under different incentive conditions, and generate corresponding response feature vectors; To effectively identify the charging behavior responses of electric vehicle users under different incentive conditions, a dynamic user behavior response identification method is constructed. This method extracts the sensitivity characteristics of individual users to external incentive information through in-depth analysis of multi-source data and behavioral modeling, providing a basis for behavioral grouping and parameter setting for subsequent differentiated charging regulation. The specific steps include: Obtain historical charging behavior data from electric vehicle users and associated incentive trigger information. In practice, by compiling statistics on user charging behavior over a specific time period, key data is collected, including but not limited to charging initiation time, duration, actual connected power, charging location, charging completion rate, and historical pricing information. All incentive triggering events are annotated with their time points and policy content to establish a temporal alignment between charging behavior and incentive events. To ensure data continuity and relevance, the behavioral data is time-normalized to remove data deviations introduced by abnormal behavior (such as power outages and fault interruptions). This constructs a sequence of behavior-incentive interaction data pairs based on a unified time base.

[0020] Based on the alignment of historical charging behavior sequences with incentive information, we extract responses to each user's behavioral changes. By constructing time period comparison windows, we statistically analyze the differences in user behavior characteristics within adjacent time windows before and after the incentive period. This allows us to extract behavioral characteristics such as whether a user initiated charging, adjusted power, or changed locations during the incentive period. To better characterize behavioral trends, we further introduce advanced behavioral statistics such as behavioral fluctuation indicators, relative difference ratios, and incentive window delays. These multi-dimensional indicators can describe the direction, intensity, and lag of users' responses under different types of incentives, thereby uncovering patterns in the dynamic changes of individual behaviors.

[0021] In the process of identifying the dynamic response of electric vehicle users' charging behavior, in order to more accurately describe the changing trends of users' behavior patterns before and after the implementation of incentive strategies, three representative statistical quantities are introduced: behavior fluctuation index, relative difference ratio, and incentive action window delay, which are used to construct a multi-dimensional quantitative expression of behavioral changes.

[0022] The behavioral volatility index refers to the magnitude of the fluctuation in a user's charging behavior within a specific time window. It is typically expressed as a measure of charging initiation time, duration, or power level, such as standard deviation, coefficient of variation, or sliding differential. Its purpose is to measure the stability of a user's charging behavior before and after the incentive intervention, thereby assessing the predictability and response jitter characteristics of their behavior patterns. For example, if a user's charging time fluctuates significantly over multiple incentive cycles, their volatility index will be large, indicating that the incentive has a strong but unstable impact on their behavior.

[0023] The relative difference ratio (RDR) refers to the difference between the average value of a key behavioral variable (such as access time or power level) and the baseline value in the adjacent time periods before and after the incentive event. It is typically expressed as "(post-incentive mean - pre-incentive mean) / pre-incentive mean." It measures the intensity, direction, and significance of a user's response and is a key indicator for determining whether users significantly adjust their behavior in response to the incentive. A higher RDR indicates a greater likelihood that users will adjust their strategies based on the principle of maximizing their own benefits.

[0024] The incentive window latency refers to the time interval between when an incentive is triggered and when a user's behavior actually changes, typically measured in minutes or hours. This metric reflects the time lag in user behavior and is particularly useful for evaluating users who don't respond immediately to incentives. It helps distinguish between "agile users" and "lagging users," further enabling the design of lead times and duration for incentive push notifications.

[0025] In practice, the acquisition of these three metrics is based on the time alignment of historical user behavior data with incentive policy records. Statistical analysis is performed by constructing sliding time windows: first, data is segmented into windows of equal time before and after the incentive event; second, the mean, variance, and time difference of each user's behavioral variable within the window are calculated; finally, the statistical results are normalized and encoded to construct a response feature vector. These metrics complement each other, collectively depicting the magnitude, direction, and temporal characteristics of user behavioral responses, providing a refined behavioral data foundation for subsequent user clustering and behavioral modeling.

[0026] After extracting the behavioral response indicators, a feature expression structure is constructed to generate a response feature vector for each electric vehicle user. Specifically, the behavioral response indicators extracted above are used as feature dimensions, and a uniform normalized encoding method is used to construct a high-dimensional vector representation structure. In this response feature vector, each dimension represents the typical behavioral response performance of a user under a certain type of incentive conditions (such as peak and valley electricity prices, instant subsidies, point rebates, etc.), and the vector as a whole reflects the user's incentive sensitivity distribution. To improve the separability of this feature vector in the subsequent grouping process, a covariance weight adjustment mechanism is introduced to assign differentiated weights to different response dimensions. This makes it easier to cluster users with similar behavioral patterns in the feature space, thereby enhancing the robustness and stability of the recognition method.

[0027] The "covariance weight adjustment mechanism" refers to the dynamic assignment of weights to each dimension (i.e., behavioral response characteristics under different incentive conditions) after constructing the user response feature vector. This weight reflects the importance of each dimension in distinguishing user behavior patterns based on the covariance structure between the different dimensions. In this invention, this mechanism enhances the separability and clustering effectiveness of user features in high-dimensional space. This allows users with similar behavioral patterns to be more concentrated in subsequent gradient response groupings, while users with significantly different behaviors are more easily distinguished, thereby improving the robustness and stability of the recognition results.

[0028] Specifically, in the response feature dataset of a large number of user samples, the covariance matrix between each dimension is calculated to evaluate the degree of linear dependence between different feature dimensions. Dimensions with smaller covariance (indicating greater independence and significant differences in user behavior) are given higher weights to emphasize their important contribution to the division of behavioral patterns. Dimensions with larger covariance, strong redundancy, or high distribution concentration among multiple users are given lower weights to reduce their interference with cluster boundaries. Weight adjustment can be achieved through principal component analysis (PCA) or feature covariance normalization transformation, remapping the feature vectors into the weight-compressed feature space. This process not only avoids the influence of invalid or weak discriminant features on the grouping results, but also helps to build a user classification system with a clearer structure and obvious response trends, thereby providing a more stable grouping basis for the differentiated configuration of the subsequent distributed incentive scheduling table.

[0029] To ensure that the response feature vector can dynamically adapt to changing trends in user behavior, a response vector update mechanism is established. By continuously collecting a new round of user charging behavior data during subsequent charging scheduling operations, performing time series comparative analysis with the original behavior characteristics, and dynamically adjusting the corresponding response vector dimensions, the continuous learning and evolution of the user behavior model can be achieved. This update mechanism can not only identify long-term drift in user behavior patterns, but also quickly respond to the triggering of short-term abnormal behavior, thereby maintaining the timeliness and accuracy of the recognition results. Ultimately, based on the constructed response feature vector, a personalized behavior response label can be calibrated for each user, providing a data foundation and behavioral basis for the subsequent response gradient grouping and differential incentive control strategy formulation.

[0030] The core role of the step "building a dynamic user behavior response identification model, acquiring historical charging behavior sequences and incentive triggering time nodes, extracting the response changes of each EV user under different incentive conditions, and generating corresponding response feature vectors" is to provide a precise, dynamic, and personalized behavioral profiling foundation for the subsequent differentiated regulation of EV user behavior. In large-scale charging scheduling scenarios, EV users exhibit significant differences in their response to incentive policies, with some users sensitively and quickly adjusting their behavior, while others are slow or even unresponsive. Ignoring these individual differences and directly applying a unified incentive signal can easily lead to user strategy convergence, triggering risks such as centralized access and secondary load peaks. Therefore, this step integrates historical user charging behavior data with incentive triggering time series data to quantify the response characteristics of each user under different incentive conditions. A high-dimensional response feature vector is constructed, encompassing dimensions such as response direction, response intensity, and response lag, comprehensively characterizing the sensitivity, stability, and consistency of user behavior. This feature vector not only serves as input for subsequent user grouping and incentive priority setting, but also participates in dynamic scheduling strategy optimization and behavior evolution modeling. It possesses long-term adaptability and updability, and is the core behavior driver within the overall dual-objective optimization control framework. This step enables us to break away from the traditional scheduling mindset of "uniform processing by user" and move towards a new paradigm of intelligent scheduling of "precise control based on behavioral differences."

[0031] S200, grouping the behavior responses of electric vehicle users based on the response characteristic vectors to form a plurality of response level groups with different incentive sensitivity gradients, and setting a corresponding incentive control priority for each response level group; To accurately classify differences in electric vehicle user behavior and prioritize control response strategies, a grouping method based on response feature vectors was constructed. This method divides a large number of electric vehicle users into multiple response hierarchical groups with different incentive sensitivity gradients. A corresponding incentive control priority was set for each group to support the subsequent configuration of differentiated strategies in the distributed incentive scheduling table. This method has good individual recognition capabilities and group scheduling adaptability, and specifically includes the following steps: Collect the generated set of user response feature vectors as the input data source for this step. Each response feature vector consists of multiple dimensions, which correspond to the response behavior characteristics of the user under different incentive conditions, such as response intensity, response direction, response volatility, response delay, etc. In order to improve the discrimination of the groups, all response feature vectors are normalized so that they can be analyzed for spatial distribution under a unified numerical scale. At the same time, a covariance weight adjustment mechanism is introduced according to the actual distribution characteristics of each dimension to weight the response feature dimensions, thereby enhancing the influence of highly discriminative features in the clustering process and ensuring that users with similar behavior patterns have higher spatial aggregation.

[0032] Based on the weighted set of response feature vectors, a multi-level behavioral response clustering analysis is performed. Unsupervised learning methods, such as Gaussian mixture models, fuzzy C-means algorithms, or spectral clustering, are used to automatically segment user samples within the feature space. This clustering process does not pre-set the number of users. Instead, it adaptively determines the optimal number of clusters based on the natural clustering trends within the response feature space, forming several behavioral response pattern categories. To prevent misclassification of edge cases, a response stability indicator is designed as an auxiliary evaluation basis, allowing for secondary correction and classification of marginal users to improve overall clustering quality.

[0033] "Multi-level behavioral response cluster analysis" uses unsupervised learning methods to classify differences in user behavioral responses based on the response feature vectors of electric vehicle users, without relying on manually assigned labels. This allows users with similar behavioral patterns to be clustered into similar categories, forming multiple response hierarchies. The core function of this analysis is to automatically identify different types of user groups from the data on their charging behavior responses to incentive policies. This provides a basis for refined grouping for the subsequent implementation of differentiated incentive regulation strategies, avoiding the policy mismatch and resource waste associated with traditional coarse-grained segmentation. This process typically constructs a high-dimensional feature space and analyzes it using unsupervised clustering algorithms such as Gaussian mixture models, fuzzy C-means, or spectral clustering. Gaussian mixture models are suitable for situations where behavioral characteristics exhibit continuous probability distributions. Fuzzy C-means allows users to belong to multiple behavioral categories simultaneously, making it suitable for scenarios with fuzzy behavioral boundaries. Spectral clustering, using graph partitioning as a framework, identifies non-convex group distributions and adapts to complex behavioral similarity relationships. During implementation, the weighted response feature vectors are first fed into a clustering algorithm. Internal evaluation metrics (such as the silhouette coefficient and Davies-Bouldin index) are then used to adaptively determine the optimal number of clusters. This algorithm then outputs the behavioral category label for each user. The resulting multiple behavioral response categories form a multi-level response gradient grouping, providing actionable and dynamically updateable data support for subsequent prioritization of incentives and control and generation of distributed scheduling strategies.

[0034] After clustering, the overall sensitivity index to external incentive signals is calculated based on the average response eigenvector of each cluster group. This sensitivity index comprehensively considers factors such as the average response amplitude, response direction consistency, and response lag time of users within the group. Higher values ​​indicate that users are more susceptible to incentive intervention and behavior adjustment. Based on the size of this sensitivity index, all cluster groups are sorted and divided into multiple response gradient hierarchical groups, defined as "high sensitivity group," "medium sensitivity group," and "low sensitivity group," to clarify the incentive response ability level of different user groups.

[0035] Based on the incentive sensitivity of each response level group, the corresponding incentive control priority is set. Specifically, response level groups with high priority will be allocated more incentive resources, earlier incentive delivery time, or stronger price elasticity to fully stimulate their control potential; low-priority groups will have their incentive intensity appropriately weakened, or a hysteresis strategy will be adopted to avoid incentive waste and policy congestion. This control priority not only serves as the basis for setting subsequent scheduling strategies, but can also be adjusted dynamically. When the behavioral characteristics of a response group change significantly during operation, its priority can be automatically updated to achieve the coordinated evolution of grouping and control strategies.

[0036] The core purpose of the step of "grouping EV users' behavioral responses based on response eigenvectors, forming multiple response hierarchical groups with varying incentive sensitivity gradients, and assigning corresponding incentive control priorities to each response hierarchical group" is to break the inherent "homogeneous treatment" of users in traditional charging scheduling and achieve hierarchical classification and differentiated precise implementation of user control strategies. In practice, EV users exhibit significant individual differences in their behavioral responses to various incentives, such as price, subsidies, and point rebates. Some users are highly sensitive and immediately adjust their charging behavior upon the issuance of incentives, while others are less responsive, with minimal or even no response. Without identifying and grouping these response differences, directly applying a unified incentive strategy would not only fail to effectively control grid load but could also lead to negative consequences such as resource waste, secondary peaks, and even traffic congestion. Therefore, by analyzing response eigenvectors, constructing a high-dimensional behavioral representation space, and using a clustering algorithm within this space to automatically group users, users can be divided into multiple behavioral hierarchical groups, such as high-sensitivity, medium-sensitivity, and low-sensitivity, thereby understanding the actual responsiveness of different user groups to control signals. Based on this division, incentive control priorities are set for different hierarchical groups to achieve optimal allocation of incentive resources: incentives are prioritized for groups with high response potential to achieve maximum benefit guidance; for groups with low response, push notifications are delayed and reduced in frequency to avoid ineffective stimulation. This step not only improves overall scheduling efficiency and response flexibility, but also lays the foundation for a behavior-driven distributed control strategy, reflecting a shift from "average incentives" to "precise incentives." It is a key step in building an intelligent, multi-level, and adaptive charging control system.

[0037] S300: Based on the response level groups and incentive control priorities, the distributed incentive scheduling table for control is generated by combining the behavioral response characteristics of each response level group with the current grid operation status. The incentive scheduling table includes incentive amplitude parameters, charging access period parameters, and control intervention strategy parameters. To achieve refined and differentiated guidance for different types of electric vehicle users, a distributed incentive schedule for charging regulation is constructed based on the established response level groups and their incentive regulation priorities, combined with the behavioral response characteristics of each level group and the real-time grid operation status. The schedule is dynamically generated and scalable, and includes core scheduling information such as incentive amplitude parameters, charging access time parameters, and regulation intervention strategy parameters. This schedule not only enables customized configuration of incentive delivery content and paths, but also supports coordinated adjustments over time, grid status, and user behavior feedback. Specifically, it includes the following steps: Based on the obtained user response level grouping results, the response behavior distribution characteristics of each response level group in different historical incentive scenarios are statistically analyzed. Indicators such as response rate, behavior adjustment amplitude, lag time, and unexpected behavior frequency are extracted. Combined with current grid operating state parameters (such as regional load forecast value, time period power margin, electricity price fluctuation range, voltage safety margin, etc.), a scheduling scenario identification dataset is constructed. To ensure the coordination and consistency between strategy design and grid demand, it is necessary to integrate the user group behavior characteristics and the matching relationship between power supply capacity in this dataset to identify which user levels have the greatest regulation potential in the current or future time period and can be incentivized without triggering load risks.

[0038] Based on the above identification results, the incentive amplitude parameter of each response level group is determined. This parameter is used to quantify the incentive intensity that should be given to the user group at this level under the current grid operation state, including but not limited to numerical quantities such as price discount ratio, point return ratio, and subsidy amount. The parameter design process adopts a response elasticity matching mechanism, that is, based on the user's historical response function to different incentive amplitudes, combined with the current load peak shaving target or load valley filling demand, the most cost-effective incentive amplitude is reversely deduced to ensure the maximum user behavior guidance efficiency within the incentive cost control range. For users in the highly sensitive group, a smaller incentive amplitude can be used to induce expected behavior, while for the low-sensitivity group, the incentive amplitude can be appropriately increased and the delivery time can be delayed to share the response burden and improve the regulation coverage.

[0039] The "response elasticity matching mechanism" refers to a strategic design method that dynamically determines the optimal incentive amplitude based on the relationship between historical user behavior data and behavioral changes under incentive conditions. In this invention, the core function of this mechanism is to measure the "degree of behavioral change caused by a unit of incentive intensity" (i.e., response elasticity) based on the user's historical response function to different incentive amplitudes. This is then matched with the current grid-side scheduling objectives (such as peak shaving or valley filling), and the most cost-effective incentive intensity value that can both induce the expected user behavior and avoid wasting incentive resources under current conditions is reversed. The implementation of this mechanism involves three key steps: First, based on historical data, a behavioral response function model for each response level group under different incentive amplitudes is constructed. The model can be expressed as a nonlinear function of charging behavior variables (such as access probability and power adjustment ratio) with respect to the incentive amplitude. Second, based on the load regulation requirements of the current scheduling period, the required behavioral change is calculated, and the target response value is set accordingly. Finally, by reversely calculating this response function, the minimum incentive amplitude required to achieve the target response value is solved, thereby determining the incentive intensity adapted to the response group in the current period. This method enables the incentive strategy to take into account both user behavior differences and grid target requirements, effectively improving the cost-effectiveness of incentive guidance. It is an important strategy for achieving precise regulation and optimal resource allocation.

[0040] A "historical response function" is a mathematical function that reflects the relationship between incentive intensity and user behavior response, based on past charging behavior data of electric vehicle users under different incentive conditions and time periods. Its core function is to quantify the sensitivity and response patterns of user behavior to incentive changes, predicting whether and to what extent users will adjust their charging behavior under a given incentive magnitude. Specifically, the independent variable of the historical response function is typically an incentive parameter (such as price discount or subsidy amount), and the dependent variable is a behavioral variable (such as the probability of initiating charging, access time delay, or power selection change). The response function model is trained using curve fitting, nonlinear regression, or machine learning methods (such as gradient boosting trees and support vector regression) by collecting historical user behavior data under multiple incentive scenarios. This function not only reveals the behavioral resilience of users to incentive changes but also serves as a foundational tool for subsequent calculations of "incentive magnitude," "control strategy design," and "hierarchical response classification." The historical response function enables more accurate predictions of future user behavior, a key foundation for refined behavior guidance and efficient resource allocation.

[0041] To further refine user behavior regulation, corresponding charging access period parameters are configured for each response level group. Specifically, the load fluctuation curve and adjustable capacity distribution of the target area power grid are first analyzed, and combined with the user behavior prediction curve, multiple target access time windows with low load or sufficient spare capacity are identified. These time windows are then matched with different response level groups, prioritizing highly sensitive groups with high stability and strong behavioral consistency to the critical peak-shaving period, while users with delayed responses are guided into the buffer time band. These access period parameters are set in the form of intervals, and combined with the incentive information push and request guidance mechanism, they provide input basis for subsequent path recommendations.

[0042] After completing the setting of the incentive amplitude parameters and access period parameters, set the corresponding control intervention strategy parameters for each response level group. This parameter is used to define the real-time intervention measures and restrictions that should be taken during the execution of charging scheduling, including load constraint boundaries, response failure readjustment strategies, user behavior deviation warning mechanisms, etc. The core of the control intervention strategy parameters is to ensure the security and stability of scheduling execution. When some users fail to follow the recommended path behavior or the grid load forecast deviates, the strategy can be immediately triggered to respond and adjust, such as restricting subsequent access, temporarily adjusting the incentive plan, or reallocating the access time window, so as to maintain the dynamic balance of the overall scheduling structure.

[0043] The core function of this step, "Based on the response level groups and incentive control priorities, and combining the behavioral response characteristics of each response level group with the current grid operating status, a distributed incentive scheduling table for control is generated. The incentive scheduling table includes incentive amplitude parameters, charging access period parameters, and control intervention strategy parameters.", is to build an execution strategy system that can perform refined charging guidance and load control based on user behavior differences and the dynamic state of the grid, so as to achieve the unity of personalized response control and load balancing goals in the charging scheduling process. This step is a key link that connects the previous and the next, inheriting the behavioral cognitive foundation of the previous user response grouping and priority identification, and directly outputting an operational incentive strategy deployment table as the basis for scheduling execution. By introducing the incentive amplitude parameter, the minimum effective incentive intensity can be dynamically set based on the elasticity of user behavior, thereby achieving a balance between resource conservation and guidance efficiency. By setting the charging access period parameter, users with different response capabilities can be reasonably allocated to time windows with different load carrying capacities, avoiding policy clustering, response congestion, and the risk of secondary peaks. The control intervention strategy parameter provides security during the execution process. Once user behavior deviates from expectations or the grid load becomes abnormal, the parameter can trigger corresponding restrictions, adjustments, or emergency strategy compensation mechanisms. The generation of a distributed incentive schedule not only gives incentive control a visual, configurable, and dynamically updateable operational structure, but also realizes the transition from "global unified scheduling" to "clustered and precise guidance." It is the core technical support for building an adaptive, flexible, and highly responsive intelligent charging scheduling system.

[0044] S400: When an electric vehicle user initiates a charging connection request, differentiated incentive information is pushed based on the response level group to which the user belongs and the control parameters in the distributed incentive scheduling table to guide the user to generate an actual charging access path sequence; To achieve differentiated, real-time guidance for electric vehicle users and avoid secondary load peaks caused by concentrated grid access, a differentiated incentive push and path guidance method combining user response hierarchical groups with a distributed incentive schedule is proposed. This method triggers a behavior recognition and policy matching mechanism at the critical node where a user initiates a charging connection request, ultimately guiding the user to select the access path that best meets grid regulation objectives. This process includes the following steps: When an electric vehicle user initiates a charging connection request, the corresponding response feature vector is automatically identified and the response hierarchical group to which it belongs is retrieved. This hierarchical group is derived through multidimensional clustering based on the user's historical incentive response behavior, representing the user's level of responsiveness when faced with incentive information, such as high sensitivity, medium sensitivity, or low sensitivity. After identification, the distributed incentive scheduling table corresponding to the user in the current scheduling cycle is simultaneously extracted. This table contains the incentive amplitude parameters, access period parameters, and control strategy constraint parameters that match the response group, forming the basis for personalized configuration of the user's control strategy.

[0045] Based on the extracted dispatch parameters, the optimal incentive strategy is calculated under the current grid load status and user demand conditions. The incentive strategy generation process not only considers the incentive control priority of the user's hierarchical group, but also combines information such as the load forecast value, voltage margin, and power adjustment margin of the grid nodes in the current time period to evaluate whether the target user can be directed to a low-load period or an area with higher available capacity. On this basis, the incentive cost-effectiveness of the user under different time windows and access node combinations is calculated. Based on the incentive response function of the user's historical behavior, the strategy combination that meets the minimum incentive cost and can achieve the maximum behavioral adjustment range is selected as the differentiated incentive push content.

[0046] After the incentive strategy is generated, differentiated incentive information is pushed to the user terminal in a parameterized structure. This incentive information not only includes charging price discounts, points rebates, or reward subsidy values, but also clearly specifies the recommended access time window, recommended access location range, and corresponding validity period. To increase the probability of user execution, the push information also combines successful response case prompts from users with similar behavior with a dynamic risk warning mechanism, for example, informing users that "accessing during this recommended time period will earn XX points and reduce the risk of queuing." At the same time, the policy execution conditions are bound in the background. Once the user confirms the recommended path and accesses it, the behavior recording and feedback monitoring process will automatically begin.

[0047] After the user receives the differentiated incentive information and independently selects the charging period and access path, the access path information they confirm is registered as the "path sequence" within that round of scheduling. This path sequence includes elements such as user identification, access time point, access location, and power request amount, and serves as the input basis for subsequent grid-side load adjustment, capacity allocation, and dynamic forecasting modeling. At the same time, the path sequence is also included in the user behavior database as a behavior execution record, providing actual execution feedback data for the next response feature identification and incentive strategy iteration. Through the above mechanism, a closed-loop behavior guidance chain is formed from user group identification → differentiated incentive push → path decision support → actual path implementation, achieving two-way collaborative optimization between scheduling goals and user autonomous choices.

[0048] The core function of this step, "When an electric vehicle user initiates a charging connection request, differentiated incentive information is delivered based on their response level group and the control parameters in the distributed incentive schedule, guiding the user to generate a specific charging access path sequence," is to achieve a precise closed-loop transition from user classification and identification to individual control execution. This allows the previously identified user behavior characteristics and policy optimization results to be applied to actual scheduling operations, thereby achieving a coordinated interaction between electric vehicle users and the power grid. This process breaks away from the traditional "unified rules, passive access" EV charging model and instead adopts a proactive control mechanism of "response identification - hierarchical push - behavior guidance." This mechanism accurately delivers incentive content based on the user's behavioral response capabilities, guiding the user to charge at the appropriate time, location, and power level. This step not only makes the allocation of incentive resources more precise and targeted, avoiding excessive incentives or resource waste, but also effectively distributes charging load, reducing local congestion and grid stress caused by the concentration of homogeneous users. At the system level, this step, through the dynamic generation of user access path sequences, enables proactive control and capacity management of grid load, resulting in more even load distribution and controllable scheduling. At the user level, this differentiated guidance strategy enhances individual engagement and response rates while maintaining user decision-making freedom, strengthening acceptance of incentive execution and enhancing behavioral stability. Therefore, this step is a key bridge to achieving the goal of tailored intelligent charging scheduling to individual needs and the times. It plays a central role in connecting scheduling strategies with actual behavior and serves as the execution vehicle for strategy implementation and behavioral transformation within the entire dual-objective optimization system.

[0049] S500, when the electric vehicle user performs the charging task according to the access path sequence, the charging behavior data and the grid load status data in the corresponding time period are collected, and a response feedback matrix between the user path behavior and the grid response status is constructed; To dynamically analyze the correlation between EV user behavior responses and the actual grid operating status, a response feedback evaluation mechanism was established. A response feedback matrix was constructed, with user access paths as the primary focus and grid load fluctuations as the primary control. This matrix not only evaluates the effectiveness of differential incentive strategies but also provides real-time data support for subsequent incentive parameter correction and identification model optimization. This process includes the following steps: As electric vehicle users perform charging tasks along the recommended access paths, their charging behavior data is continuously recorded. Collected data includes, but is not limited to, key indicators such as charging initiation time, access location, charging duration, average power level, maximum power value, power demand completion rate, and task interruption frequency. To ensure the continuity and accuracy of data collection, a high-frequency sampling mechanism is used to periodically scan the behavior process, ensuring that the behavioral changes throughout the charging process are fully captured. Furthermore, to account for user differences, all data records are linked to user IDs and route numbers to establish clear behavioral attribution.

[0050] Synchronously collect grid load status data for the time period corresponding to the user's charging task. This data includes operational status information such as node load values, load fluctuations, voltage offsets, current changes, power factors, and grid frequency within the scheduling cycle. In particular, the real-time operational status of the user's connected location node and its surrounding area should be collected, and time windows should be segmented and strictly aligned with the user behavior data in terms of timing to achieve a "one-to-one" status correspondence. To improve data consistency and comparability, the grid status data is smoothed and anomalies are eliminated to eliminate errors caused by external disturbances.

[0051] Based on the above two data sources, a two-dimensional response feedback matrix between user path behavior and grid response status is constructed. The row vectors of the matrix represent the user's behavior sequence under different access paths, and the column vectors represent various indicators of the grid status during the time period of the path. Each matrix unit reflects the actual feedback of the grid operation status caused by a certain behavior selection. For example, the "high power-short period access" behavior may correspond to indicator responses such as "current surge-voltage drop". In order to enhance the expressive power of the matrix, normalization processing, behavior label mapping and load disturbance weight correction mechanism are introduced, so that the matrix not only has a visual behavior-load mapping relationship, but also can serve as the input basis for calculating comprehensive indicators such as guidance deviation, strategy effectiveness and grid stability contribution rate.

[0052] The constructed response feedback matrix is ​​incorporated into the data feedback engine for subsequent incentive strategy evaluation and adaptive correction of the scheduling mechanism. The response feedback matrix can be used as an overlay record of multiple rounds of charging behavior. By comparing and analyzing user behavior across cycles, the stability and consistency of user behavior, as well as the long-term impact trend of the incentive scheme on the quality of grid operation, can be analyzed. At the same time, by clustering and tracking high-deviation units in the matrix, phenomena such as "behavior deviation from the recommended path" and "abnormal load response" are identified, and feedback is fed back to the incentive parameter control mechanism to trigger iterative optimization of the incentive amplitude, access period, and intervention strategy. Ultimately, a closed-loop evaluation mechanism with real behavior as the core and system response as the basis is constructed to achieve quantitative evaluation of the incentive guidance effect, risk warning of grid stability, and continuous evolution of the behavior recognition model, significantly improving the intelligence, adaptability, and robustness of the entire charging scheduling system.

[0053] The "data feedback engine" refers to a logical processing mechanism within the electric vehicle charging dispatch system specifically designed to receive, integrate, and analyze user charging behavior data and grid operating status data. It serves as the "feedback hub" of the entire dispatch system, responsible for converting actual operational results from the execution phase into a basis for policy evaluation and model updates. In this invention, the data feedback engine functions on three levels: first, it dynamically captures and aligns user behavior after executing the recommended access path with the grid load response, forming a response feedback matrix; second, it analyzes differences and identifies trends within this matrix to identify whether the incentive guidance is achieving its intended goals, such as whether users are deviating from the path or whether the grid is experiencing load anomalies due to concentrated behavior; and third, it feeds these analysis results back to the front end of the policy control chain, triggering parameter adjustments to the incentive amplitude, access period, or behavior recognition model. This mechanism achieves a closed-loop control logic from "push-execute-perceive-correct," ensuring that the incentive dispatch strategy is not only optimized based on theoretical models but also dynamically adapts to actual operational results, enhancing its self-learning capabilities and environmental adaptability, thereby continuously improving the accuracy, stability, and intelligence of charging dispatch.

[0054] The core function of this step, "As electric vehicle users execute charging tasks along a sequence of access paths, their charging behavior data and grid load status data for the corresponding time period are collected, and a response feedback matrix is ​​constructed between user path behavior and grid response status." This step aims to establish a closed-loop data structure and a causal mapping relationship between user behavior and actual grid operation, enabling quantitative evaluation and adaptive optimization of scheduling strategy effectiveness. In the previous step, customized charging access paths were delivered to different users through response identification and incentive guidance. However, the effectiveness of these strategies, their stable peak-shaving and valley-filling effects, and the resulting grid stress can only be accurately determined through the simultaneous observation and coordinated analysis of behavior and load. Therefore, this step simultaneously collects user behavior data (such as access time, power level, and duration) as they execute access paths, along with key grid status data (such as load variation, voltage, current, and frequency) during the same time period. This constructs a high-resolution response feedback matrix, where each cell represents the specific response of a particular user behavior to the grid state under a specific scheduling context. This mechanism can not only identify which behavioral combinations are beneficial to the grid and which pose risks, but also reveal the degree of deviation in policy guidance, that is, whether users strictly follow the recommended path to access, whether their behavior deviates from the expected goals, and whether it triggers abnormal grid operation. By continuously constructing and iteratively updating this feedback matrix, it is possible to dynamically judge the effectiveness of incentive regulation, promptly identify problems such as strategy failure, user resistance, and mass congestion, and provide detailed and reliable data for the next step of incentive parameter correction and behavior model update. Therefore, this step is a key link in achieving "verifiable" behavior guidance, "traceable" grid impact, and "evolvable" scheduling strategies. It is the core support for building a globally optimal, distributed intelligent, closed-loop adaptive electric vehicle charging scheduling system.

[0055] S600, based on the analysis results of the response feedback matrix, identifies the degree of deviation between the user behavior response and the recommended access path, as well as the difference between the grid load status and the scheduling expectation during the corresponding period. Based on the deviation degree and load difference information, the incentive amplitude parameters, access period parameters, and control intervention strategy parameters are dynamically modified, the distributed incentive scheduling table is updated, and the user behavior identification model is simultaneously optimized to achieve closed-loop adaptive optimization control between electric vehicle user behavior regulation and grid system operation; To achieve dynamic matching between EV user behavior control strategies and grid operating conditions, a closed-loop adaptive optimization control system based on a response feedback matrix is ​​constructed. Based on long-term accumulated user behavior data and grid operating status data, behavioral deviations and load anomalies during incentive strategy execution are identified. The strategy parameters and behavior model are then modified accordingly to achieve dynamic optimization of the control strategy. This process includes the following steps: Perform periodic analysis on the constructed response feedback matrix. The response feedback matrix consists of the user's behavior data after executing the recommended access path and the grid load status data during the time period. Each row in the matrix represents a charging path behavior sample, and each column represents the corresponding grid operation status indicator. During the analysis process, two aspects of data deviation are identified: one is the degree of deviation between the user's actual behavior and the recommended path, and the other is the difference between the actual load status of the grid and the load change expected by the strategy. Among them, the degree of behavior deviation can be calculated through quantitative indicators such as path similarity, execution rate, and time period deviation value; and the grid load deviation can be expressed by the residual mean and peak offset between the actual load curve and the expected curve.

[0056] Based on the above-mentioned deviation identification results, the incentive strategy correction condition judgment rules are established. When the user deviation degree exceeds the set threshold, or the strategy fails for a long time on certain access paths, even if the incentive is pushed, it fails to guide the user to access the target period; or when the actual load of the power grid deviates from the expected trend many times, and risk signals such as secondary load concentration and equipment overload appear, the strategy correction mechanism is automatically triggered. At this time, it is necessary to comprehensively consider the elastic model of user behavior response and the load response characteristics to re-evaluate the effectiveness and cost-effectiveness of the current incentive strategy. In order to avoid excessive adjustments due to abnormal cases, the behavior confidence interval and load disturbance filtering mechanism are introduced in the judgment process to improve the stability and robustness of the strategy correction.

[0057] After the correction mechanism is triggered, the corresponding incentive amplitude parameters, access period parameters and control intervention strategy parameters are dynamically adjusted according to the degree of deviation and the magnitude of the load difference. Specifically, for response level groups with severe deviation behavior, the incentive amplitude can be increased to enhance their willingness to participate, or their access period can be adjusted to a window with more behavioral acceptance. At the same time, for areas where load anomalies frequently occur, the control intervention strategy can be strengthened, such as restricting low-response users from accessing during sensitive periods, or releasing incentives to high-response users in advance to share the regulation pressure. All adjustment parameters are synchronously written into the distributed incentive scheduling table in real time, replacing the original policy items, to ensure that the control strategy always reflects the current operating environment and actual behavior.

[0058] While updating the incentive schedule, the user behavior recognition model is also optimized. Specifically, this involves integrating newly added actual user behavior data with existing training data, updating the user response feature vectors and behavior label weights, and retraining the recognition model to adapt to changing trends in user behavior over time. The optimized recognition model will be used to update the behavior hierarchical groupings during the next round of user charging requests, driving the continuous co-evolution of scheduling strategies and actual user behavior.

[0059] Based on the analysis of the response feedback matrix, the system identifies the degree of deviation between user behavior responses and the recommended access path, as well as the difference between the grid load status during the corresponding time period and the dispatch expectations. Based on this deviation and load difference information, the system dynamically adjusts the incentive amplitude parameters, access time parameters, and control intervention strategy parameters, updates the distributed incentive schedule, and simultaneously optimizes the user behavior identification model, achieving closed-loop adaptive optimization control between EV user behavior regulation and grid operation. This step aims to establish a closed-loop optimization mechanism that dynamically updates the dispatch strategy and behavior model based on actual user behavior feedback and grid operation status, enabling self-correction, continuous learning, and real-time control capabilities, thereby achieving coordinated stability between behavior and power. In actual operation, although personalized incentive strategies and access paths have been formulated for users, and guided through response grouping and incentive push notifications, user behavior is influenced by factors such as personal preferences, temporary travel needs, and price sensitivity fluctuations, resulting in some deviations between actual charging behavior and the recommended path. Furthermore, grid operation may deviate from the original dispatch expectations due to factors such as the external environment, grid load forecast errors, or concentrated user responses. Therefore, this step analyzes the response feedback matrix formed between user path behavior and grid status data to identify which paths and user groups exhibit persistent deviations and which scheduling periods frequently experience load anomalies, thereby identifying any issues or failures in the existing strategy. Based on this identification, it is possible to specifically adjust the corresponding incentive amplitude (e.g., appropriately increasing subsidies for low-response groups), adjust access time configuration (e.g., directing some users to time windows with more stable loads), and adjust the control intervention strategy (e.g., shortening the response lag window and increasing the policy response tolerance). Importantly, this step also simultaneously updates the behavior recognition model, incorporating new behavior data into feature training to adapt the model to the long-term evolution of user behavior. This closed-loop process shifts from a "prediction-driven" approach to a "feedback-driven" approach, achieving dynamic coordination between behavior control, grid security, and strategic resource utilization efficiency. This represents the key link in the overall incentive game scheduling approach that is most intelligent, evolutionary, and possesses the highest control depth.

[0060] This dual-objective optimization approach for electric vehicle charging scheduling, based on incentive game theory, enables in-depth identification and hierarchical guidance of individual user behavior, effectively addressing the convergence of user strategies and secondary load peaks caused by existing unified incentive strategies. By constructing user response feature vectors and classifying responses into hierarchical levels, this approach enables the system to dynamically allocate incentive resources based on user sensitivity and formulate differentiated, distributed scheduling strategies based on grid operating conditions. Furthermore, by incorporating real-time behavioral feedback and grid load data into a response feedback matrix, policy execution effects are visualized and quantified. This allows for adaptive iterative optimization of incentive parameters and behavioral models. The resulting closed-loop control mechanism, centered around "identification-guidance-feedback-correction," balances the individual charging utility of electric vehicle users with the overall operational safety of the power system. This effectively improves the responsiveness of incentive delivery, the dynamic adaptability of scheduling strategies, and the load balancing of grid operations, significantly enhancing the intelligence and robustness of large-scale electric vehicle charging scheduling systems.

[0061] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0062] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0063] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0064] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0065] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0066] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0067] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0068] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0069] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. A dual-objective optimization method for electric vehicle charging scheduling based on incentive game, characterized by: The following steps are involved: S100: Build a user behavior dynamic response recognition model, obtain historical charging behavior data and incentive triggering time nodes, extract user response changes under different incentive conditions, and generate a response feature vector; S200, grouping users based on response feature vectors to form multiple response level groups with different incentive sensitivity gradients, and setting corresponding incentive control priorities; S300, based on the response level groups and the incentive control priorities, combined with the response characteristics of each response level group and the current power grid operation status, generating a distributed incentive scheduling table; S400: When a user initiates a charging connection request, differentiated incentive information is pushed based on the user's response level group and the parameters in the distributed incentive scheduling table to guide the user to generate an access path sequence; S500, when a user performs a charging task according to an access path sequence, collecting the user's charging behavior data and grid load status data for the corresponding period, and constructing a feedback matrix between the user's path behavior and the grid response status; S600, based on the feedback matrix, identifies the degree of deviation between user behavior and the recommended path, as well as the difference between the grid status and scheduling expectations, dynamically corrects the incentive amplitude, access period and control strategy parameters, updates the scheduling table, and simultaneously optimizes the user behavior recognition model.

2. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 1 is characterized in that: Step S100 includes: Obtain historical charging behavior data and incentive triggering time nodes of electric vehicle users, and construct a behavior-incentive interaction data sequence; Extract indicator data on user behavior changes before and after incentives, including behavioral fluctuation indicators, relative difference ratios, and incentive window delays; Based on the extracted indicator data, the response feature vector is constructed, normalized coding is adopted and a covariance weight adjustment mechanism is introduced to enhance feature discrimination; During subsequent charging operations, the values ​​of each dimension of the response feature vector are continuously updated to achieve dynamic correction and adaptive learning of the user behavior model.

3. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 2 is characterized in that: The covariance weight adjustment mechanism includes: Calculate the covariance matrix between the dimensions in multiple user response feature vectors to determine the degree of linear correlation between the feature dimensions; High weights are assigned to feature dimensions whose covariance values ​​are lower than a set threshold to enhance their contribution to distinguishing user behavior patterns. Assign low weights to feature dimensions whose covariance values ​​are higher than a set threshold to suppress their interference in identifying user behavior features; The feature vector after weight adjustment is reconstructed to improve the clustering clarity and grouping effectiveness of user response features in the feature space.

4. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 1 is characterized in that: Step S200 includes: Perform unsupervised cluster analysis on the weighted user response feature vectors to form multiple behavioral response categories based on the feature space distribution; The average response eigenvector of each behavioral response category was calculated to determine its incentive sensitivity index; Divided into multiple response level groups based on the incentive sensitivity index; Set incentive control priorities for each response level group for subsequent incentive resource allocation and policy push timing configuration.

5. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 1 is characterized in that: Step S300 includes: Based on the historical response characteristics of the response level group and the current power grid operation status, a dispatch scenario recognition dataset is constructed; Based on the scheduling scenario identification dataset, combined with the user's historical response function and target scheduling requirements, the incentive amplitude parameters of each response level group are determined; Analyze the load fluctuation curve of the target area power grid and set the charging access period parameters for each response level group; Configure the corresponding control intervention strategy parameters to form a distributed incentive scheduling table including incentive amplitude, access period and intervention strategy.

6. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 1 is characterized in that: Step S400 includes: When an electric vehicle user initiates a charging connection request, it identifies the response level group to which it belongs and extracts the corresponding distributed incentive scheduling table parameters; Generate a differentiated incentive strategy with the lowest incentive cost and the largest behavioral adjustment range based on the distributed incentive schedule parameters and the current grid load status; Push incentive information including incentive amplitude, access time window, access location and validity period to user terminals; After the user selects, his access path sequence is registered, and the current path is used as a behavior execution record for subsequent behavior feedback and policy iteration analysis.

7. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 1 is characterized in that: Step S500 includes: Obtain the charging behavior data of electric vehicle users during the execution of the access path sequence and bind the user ID and path number; Collect the operating status data of the corresponding grid node and its adjacent area during the user's charging period, and perform time alignment and anomaly elimination processing; Construct a response feedback matrix with user path behavior as row vector and grid operation status as column vector; The response feedback matrix is ​​input into the data feedback engine for incentive strategy evaluation and control parameter update analysis.

8. The dual-objective optimization method for electric vehicle charging scheduling based on incentive game according to claim 1 is characterized in that: Step S600 includes: Periodically analyze the response feedback matrix to identify the degree of deviation between user behavior responses and recommended access paths, as well as the difference between grid load status and dispatch expectations; Determine the incentive strategy correction conditions based on the identified deviation degree and difference information; When the correction conditions are met, the incentive amplitude parameters, access period parameters and control intervention strategy parameters are dynamically adjusted, and the distributed incentive scheduling table is updated; Integrate the newly added user behavior data with the original training data, optimize the user behavior recognition model, and update the response level grouping results.

Images