A charging control method and system for an electric vehicle

Abstract

The application discloses a charging control method and system of an electric vehicle, applied to the technical field of charging control, and comprises the following steps: obtaining historical charging data of each electric vehicle to be charged; extracting charging behavior features from each historical charging data, and determining the loss degree of each electric vehicle to be charged based on the charging behavior features; obtaining a corresponding power consumption influence coefficient according to the loss degree of each electric vehicle to be charged; generating an initial charging scheme based on the geographical distribution data of charging facilities and the real-time position information, trip information and power consumption influence coefficient of each electric vehicle to be charged; optimizing the initial charging scheme to obtain a charging control scheme of all electric vehicles to be charged; and sending a control instruction to each electric vehicle to be charged, wherein the control instruction is obtained from the charging control scheme. The charging control method and system of the electric vehicle provided by the application effectively improve the reliability of charging control.

Description

Technical Field

[0001] This invention relates to the field of charging control technology, and in particular to a charging control method and system for electric vehicles. Background Technology

[0002] Compared to traditional gasoline-powered vehicles, electric vehicles use electricity as their core driving force, significantly reducing reliance on fossil fuels and thus lowering greenhouse gas and pollutant emissions at the source. Efficient and intelligent charging methods are crucial for ensuring the performance of electric vehicles, directly impacting charging efficiency, battery life, and overall vehicle reliability.

[0003] Current electric vehicle charging control methods involve generating a low-charge alarm signal based on the battery's State of Charge (SOC), prompting the driver to navigate to a nearby charging station using navigation software. This method often fails to effectively and promptly charge electric vehicles, leading to breakdowns or forced low-speed driving due to insufficient battery power, increasing driving safety risks, ultimately causing traffic congestion and reducing the overall efficiency of the road network. Summary of the Invention

[0004] This invention provides a charging control method and system for electric vehicles to solve the technical problem that existing charging control methods simply charge based on the battery's state of charge (SOC), leading to unreliability of the charging control method, thereby improving the reliability of charging control.

[0005] To address the aforementioned technical problems, this invention provides a charging control method and system for electric vehicles, the method comprising: Obtain historical charging data for each electric vehicle waiting to be charged; Extract charging behavior features from each of the historical charging data, and determine the degree of damage of each of the electric vehicles to be charged based on the charging behavior features; Based on the degree of loss of each electric vehicle to be charged, a corresponding power consumption impact coefficient is obtained; Based on the geographical distribution data of charging facilities, as well as the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged, an initial charging plan is generated. With the goal of minimizing the charging waiting time in the initial charging scheme, the initial charging scheme is input into the constructed collaborative optimization model for optimization processing to obtain the charging control schemes for all the electric vehicles to be charged. Control commands are sent to each of the electric vehicles to be charged, wherein the control commands are obtained from the charging control scheme.

[0006] Preferably, the step of extracting charging behavior features from each of the historical charging data and determining the degree of loss of each of the electric vehicles to be charged based on the charging behavior features includes: By analyzing the charging times and corresponding power data in the historical charging data, charging anxiety characteristics are obtained. Statistical processing is performed on the single-charge replenishment amount in the historical charging data to obtain charging depth characteristics; Based on the charging anxiety characteristics and the charging depth characteristics, a battery loss assessment index is constructed. Based on the battery loss assessment index, the degree of loss of each of the electric vehicles to be charged is obtained.

[0007] Preferably, the step of analyzing the charging time and corresponding power data in the historical charging data to obtain charging anxiety characteristics includes: The charging time and corresponding power data are extracted from the historical charging data. Based on the power data, the charging time is classified to obtain classified charging behaviors; The charging anxiety characteristics are obtained by analyzing and processing the categorized charging behaviors.

[0008] Preferably, the initial charging plan is generated based on the geographical distribution data of charging facilities, the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged, including: The real-time location information and the travel information are processed to obtain the constraints; Based on the constraints, the geographical distribution data of the charging facilities are processed to obtain a set of candidate charging stations; The power consumption impact coefficient is used to filter the candidate charging station set to generate an initial charging scheme.

[0009] Preferably, the collaborative optimization model includes: The collaborative optimization model is iteratively updated using multi-agent reinforcement learning technology.

[0010] Another aspect of the present invention provides a charging control system for an electric vehicle, comprising: The acquisition module is used to acquire historical charging data for each electric vehicle to be charged. An extraction module is used to extract charging behavior features from each of the historical charging data, and to determine the degree of loss of each of the electric vehicles to be charged based on the charging behavior features. The loss module is used to obtain the corresponding power consumption impact coefficient based on the degree of loss of each electric vehicle to be charged. An initial module is used to generate an initial charging plan based on the geographical distribution data of charging facilities, the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged. The optimization module is used to minimize the charging waiting time in the initial charging scheme by inputting the initial charging scheme into the constructed collaborative optimization model for optimization processing, so as to obtain the charging control scheme of all the electric vehicles to be charged. The control module is used to send control commands to each of the electric vehicles to be charged, wherein the control commands are obtained from the charging control scheme.

[0011] Preferably, the extraction module includes: The analysis unit is used to analyze the charging time and corresponding power data in the historical charging data to obtain charging anxiety characteristics; The statistical unit is used to perform statistical processing on the single charging replenishment amount in the historical charging data to obtain charging depth characteristics. The indicator unit is used to construct a battery loss assessment indicator based on the charging anxiety feature and the charging depth feature; A loss unit is used to obtain the degree of loss of each of the electric vehicles to be charged based on the battery loss assessment index.

[0012] Preferably, the analysis unit includes: The charging time and corresponding power data are extracted from the historical charging data. Based on the power data, the charging time is classified to obtain classified charging behaviors; The charging anxiety characteristics are obtained by analyzing and processing the categorized charging behaviors.

[0013] Preferably, the initial module includes: The constraint unit is used to process the real-time location information and the travel information to obtain constraint conditions; The candidate unit is used to process the geographical distribution data of the charging facilities based on the constraints to obtain a set of candidate charging stations. The filtering unit is used to filter the candidate charging station set using the power consumption impact coefficient and generate an initial charging scheme.

[0014] Preferably, the collaborative optimization model includes: The collaborative optimization model is iteratively updated using multi-agent reinforcement learning technology.

[0015] Compared with the prior art, the beneficial effects of the present invention are at least one of the following: This technical solution acquires historical charging data from each electric vehicle waiting to be charged, extracts charging behavior characteristics to determine the degree of wear and tear on each vehicle, and calculates the corresponding power consumption impact coefficient accordingly. Then, combining the geographical distribution data of charging facilities, real-time vehicle location information, travel information, and the power consumption impact coefficient, an initial charging plan is generated. Based on this, minimizing the charging waiting time in the initial charging plan is the objective, and the plan is input into a pre-constructed collaborative optimization model for optimization. Finally, a charging control plan for all electric vehicles waiting to be charged is obtained, and control commands generated by this plan are sent to each vehicle. This solution effectively overcomes the problems of delayed response, blind routes, and resource mismatch caused by relying solely on battery SOC-triggered alarms and drivers passively searching for charging stations in existing technologies. It avoids safety hazards such as breakdowns and low-speed driving due to insufficient battery power, thereby improving charging timeliness and accuracy, reducing driving safety risks, alleviating traffic congestion, and improving the overall road network operating efficiency. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of a charging control method for an electric vehicle according to one embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a charging control system for an electric vehicle in one embodiment of the present invention; Figure label: Among them, 11. Acquisition module; 12. Extraction module; 13. Loss module; 14. Initialization module; 15. Optimization module; 16. Control module. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] In the description of this invention, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0019] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0020] In the description of this invention, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0021] Compared to traditional gasoline-powered vehicles, electric vehicles are powered by electricity, significantly reducing reliance on fossil fuels and emissions. Therefore, efficient and intelligent charging methods are crucial for their performance, battery life, and operational reliability. Current charging control methods typically rely on the driver manually navigating to a charging station after the vehicle's State of Charge (SOC) is triggered. This response is delayed, easily leading to battery depletion, breakdowns en route, or low-speed driving, increasing safety risks and potentially causing traffic congestion and reducing road network efficiency.

[0022] One embodiment of the present invention provides a charging control method for electric vehicles. For details, please refer to [link / reference needed]. Figure 1 , Figure 1 The diagram shows a flowchart of a charging control method for an electric vehicle according to one embodiment of the present invention, including: S1. Obtain historical charging data for each electric vehicle waiting to be charged; S2. Extract charging behavior features from each historical charging data point, and determine the degree of damage to each electric vehicle to be charged based on the charging behavior features. S3. Based on the degree of loss of each electric vehicle to be charged, obtain the corresponding power consumption impact coefficient; S4. Based on the geographical distribution data of charging facilities and the real-time location information, travel information and power consumption impact coefficient of each electric vehicle to be charged, generate an initial charging plan. S5. With the goal of minimizing the charging waiting time in the initial charging scheme, the initial charging scheme is input into the constructed collaborative optimization model for optimization processing to obtain the charging control scheme of all electric vehicles to be charged. S6. Send control commands to each electric vehicle waiting to be charged, wherein the control commands are obtained from the charging control scheme.

[0023] First, obtain the historical charging data for each electric vehicle waiting to be charged. Electric vehicles waiting to be charged refer to those currently requiring charging and in a waiting-to-charge state. Historical charging data refers to all detailed records related to charging for these electric vehicles since they were put into use, specifically including charging time and duration, remaining battery power at the start of charging and full battery power at the end of charging, real-time and average charging power during charging, charging voltage and current, number of charging attempts and intervals between each charge, charging location and charging equipment number, as well as energy consumption data and records of abnormal charging events during the charging process.

[0024] This data is acquired to extract charging behavior characteristics and determine the degree of vehicle wear and tear. Various parameters during vehicle charging can be collected in real time by the onboard terminal's data acquisition module and stored in the onboard database. Simultaneously, relevant data about the vehicle charging at the charging facility is collected by the charging facility's data acquisition unit and uploaded to a cloud server. This data is then transmitted via V2X (Vehicle-to-Everything) technology. Everything (Vehicle-to-Everything) technology enables data interaction and synchronization between the vehicle-mounted database and the cloud server. Staff can use the data interface to call a structured query language to filter and extract data from the cloud server and the vehicle-mounted database. For different brands and models of electric vehicles, a unified data communication protocol, such as a message queue telemetry transmission protocol, can be used to achieve compatible data collection, ensuring the comprehensiveness and accuracy of data acquisition. This provides reliable basic data for subsequent charging behavior feature extraction and vehicle wear assessment, ensuring the scientific and rational nature of subsequent charging control schemes. The vehicle-mounted terminal acquisition module is a hardware device installed on the electric vehicle to collect vehicle operation and charging-related data. The charging facility's data acquisition unit is a data acquisition component integrated into charging piles and other charging equipment to record vehicle charging parameters. The cloud server is a remote server used to centrally store and manage historical charging data of all electric vehicles awaiting charging. The data interface is an interface component used to enable data transmission and interaction between different devices and databases.

[0025] Preferably, charging behavior features are extracted from each historical charging data point, and the degree of damage to each electric vehicle to be charged is determined based on these features. The charging time and corresponding power data in the historical charging data are analyzed to obtain charging anxiety features; the single-charge replenishment power in the historical charging data is statistically processed to obtain charging depth features; a battery damage assessment index is constructed based on the charging anxiety features and charging depth features; and the degree of damage to each electric vehicle to be charged is obtained based on the battery damage assessment index. The charging time and corresponding power data are extracted from the historical charging data; the charging time is classified based on the power data to obtain categorized charging behaviors; and the categorized charging behaviors are analyzed to obtain charging anxiety features.

[0026] The core of this project involves extracting charging behavior features from the historical charging data of each electric vehicle and determining the degree of vehicle wear based on these features. This is achieved by mining key information from historical charging data to accurately assess the wear and tear on the vehicle's battery, providing a basis for obtaining the power consumption impact coefficient and generating charging solutions. The charging behavior features mainly include charging anxiety features and charging depth features. Charging anxiety features reflect the urgency of electric vehicle users to charge, while charging depth features reflect the amount of power replenished during a single charge. Battery wear assessment indicators are specific parameters used to quantify the wear and tear on electric vehicle batteries. The degree of wear is a value calculated from the battery wear assessment indicators that reflects the extent of battery aging and performance degradation.

[0027] First, charging times and corresponding battery levels are extracted from historical charging data. Charging time refers to the specific time each charging is initiated, and battery level refers to the remaining battery charge at that time. Next, charging times are categorized based on the battery level data using K-means clustering. The categorization is based on the battery level: charging times with low remaining battery charge are grouped into one category, those with medium remaining battery charge into another, and those with high remaining battery charge into a third. This results in categorized charging behavior, which is the set of charging time categories for different remaining battery levels. Then, the frequency and interval of each category are statistically analyzed, combined with time-series analysis to uncover the distribution patterns of charging times, thus revealing charging anxiety characteristics. If the frequency of charging times corresponding to the low remaining battery category is high and the interval is short, it indicates a high level of charging anxiety; conversely, if the frequency is low and the interval is short, it indicates a low level of charging anxiety. The anxiety level is low. Simultaneously, statistical processing is performed on the single-charge replenishment amount in historical charging data. The single-charge replenishment amount is the difference between the amount of electricity at the end of a single charge and the amount at the beginning of the charge. The arithmetic mean method is used to calculate the average of all single-charge replenishment amounts over a period of time, and the standard deviation is combined to calculate the fluctuation range of the replenishment amount, thus obtaining the charging depth characteristic. A high average replenishment amount with small fluctuations indicates a stable and deeper charging depth, while a low average replenishment amount indicates an unstable and shallow charging depth. Subsequently, a battery loss assessment index is constructed based on the charging anxiety characteristic and the charging depth characteristic. The anxiety coefficient corresponding to the charging anxiety characteristic and the depth coefficient corresponding to the charging depth characteristic are weighted and summed. The weights are determined using the analytic hierarchy process (AHP) to obtain the specific value of the battery loss assessment index. Finally, the battery loss assessment index value of each electric vehicle to be charged is compared with a preset loss standard to quantify the degree of loss of each electric vehicle to be charged. A higher value indicates more severe loss. The advantage of doing this is that it can accurately quantify the vehicle battery wear and tear, providing a reliable basis for the subsequent calculation of the power consumption impact coefficient. Time series analysis is a data analysis method used to analyze the distribution patterns and trends of time series data. Arithmetic mean is a statistical method used to calculate the average value of a set of data. Standard deviation is a statistical indicator used to measure the dispersion of a set of data. Analytic hierarchy process (AHP) is a system analysis method used to determine the weights of multiple indicators.

[0028] Furthermore, based on the degree of loss of each electric vehicle to be charged, the corresponding power consumption impact coefficient is obtained.

[0029] A corresponding power consumption impact coefficient is obtained based on the degree of wear and tear of each electric vehicle to be charged. This coefficient quantifies the impact of battery wear on actual power efficiency, typically ranging from 0 to 1. Higher wear levels result in a larger coefficient, indicating a more significant negative impact on power efficiency, while lower wear levels result in a smaller coefficient, signifying a less pronounced impact. As previously explained, the degree of wear reflects the aging and performance degradation of the electric vehicle battery, and this explanation will not be repeated here. This approach incorporates battery wear factors into the charging solution development process, ensuring that the resulting charging plan better reflects the actual power consumption of the vehicle.

[0030] First, the wear levels of all electric vehicles awaiting charging are standardized using the min-max normalization method. This maps the wear level values ​​of each vehicle to a range of 0 to 1, eliminating the impact of differences in wear level values ​​between different vehicles. The standardized calculation formula is: the standardized wear level equals the current wear level minus the minimum wear level of all vehicles, divided by the difference between the maximum and minimum wear levels of all vehicles. Then, a calculation model for the power consumption impact coefficient is constructed based on the standardized wear level. A linear regression method is used to establish the correspondence between the standardized wear level and the power consumption impact coefficient. By collecting sample data of standardized wear levels and actual power consumption impact coefficients from historical data, the linear regression model is trained to determine the regression coefficients and intercept term. After training, the standardized wear levels of each electric vehicle awaiting charging are calculated. By substituting the values ​​into the model, the corresponding power consumption impact coefficient can be calculated. To ensure the accuracy of the calculation results, the calculated power consumption impact coefficient needs to be verified and outliers removed. If the power consumption impact coefficient of a vehicle exceeds the reasonable range of 0 to 1, interpolation is used to correct it so that the coefficient returns to a reasonable range. The min-max standardization method is a standardization processing method that maps data to a specified range through linear transformation. The linear regression method is a statistical analysis method used to establish a linear relationship between two or more variables. The interpolation method is a numerical calculation method used to supplement and correct outlier or missing data. The advantage of doing this is that it can accurately quantify the impact of battery loss on vehicle power consumption, providing data support for the subsequent generation of a scientific and reasonable charging plan, ensuring that the charging plan meets the vehicle's charging needs while also taking into account power consumption efficiency.

[0031] Furthermore, based on the geographical distribution data of charging facilities and the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged, an initial charging plan is generated. The real-time location information and travel information are processed to obtain constraints; based on the constraints, the geographical distribution data of charging facilities are processed to obtain a set of candidate charging stations; the power consumption impact coefficient is used to filter the set of candidate charging stations to generate the initial charging plan.

[0032] An initial charging plan is generated based on the geographical distribution data of charging facilities, as well as the real-time location, travel information, and power consumption impact coefficient of each electric vehicle to be charged. The geographical distribution data refers to detailed data such as the specific geographical coordinates of all charging facilities in the area, the number of available charging piles, charging power levels, and operating hours. Real-time location information refers to the precise geographical coordinates of the electric vehicle to be charged, which can be collected in real time through the Global Positioning System (GPS). Travel information refers to the expected mileage and arrival time of the electric vehicle's subsequent route. Constraints refer to the conditions determined based on the real-time location and travel information that restrict the vehicle's choice of charging stations. The candidate charging station set refers to the set of charging stations selected from all charging facilities that meet the constraints. The initial charging plan refers to the preliminary charging time and charging amount plan for each electric vehicle to be charged, determined based on the data. The power consumption impact coefficient, which quantifies the impact of battery wear on power efficiency, has already been explained and will not be repeated here. This is done to initially match vehicles with charging facilities, laying the foundation for subsequent optimization of the charging plan.

[0033] First, real-time location and trip information are processed. Using a GIS (Geographic Information System), the vehicle's real-time location coordinates are matched with the geographic coordinates of the trip route. The distance from the vehicle's current location to each potential charging point along the route and the estimated travel time are calculated. Simultaneously, combined with the vehicle's estimated mileage and power consumption impact coefficient, the maximum drivable range of the vehicle is determined, thus deriving constraints. These constraints mainly include the maximum distance between the charging station and the vehicle's current location, the availability of charging stations within the vehicle's trip route coverage area, and the matching of the vehicle's estimated charging time. Then, based on these constraints, the geographic distribution data of charging facilities is processed. A spatial query algorithm is used to filter charging stations that meet all constraints from all charging facilities, integrating them into a candidate charging station set. During the filtering process, GIS technology is used to verify in real-time whether the geographical location of the charging station is within the vehicle's drivable range, while simultaneously retrieving the real-time operation data of each charging facility. The data is used to confirm whether the number of available charging piles and operating hours meet the constraints. Then, the candidate charging station set is screened using a power consumption impact coefficient. A weighted scoring method is used to score the candidate charging stations, with scoring indicators including the distance between the charging station and the vehicle, the charging power of the charging station, and the power consumption impact coefficient. The weight of the power consumption impact coefficient is set according to the degree of vehicle wear and tear; the higher the degree of wear and tear, the greater the weight of the power consumption impact coefficient. Charging stations with high charging power and suitable distance are prioritized. Finally, based on the scoring results, the candidate charging station with the highest comprehensive score is selected for each electric vehicle to be charged. Simultaneously, combining the vehicle's current battery level, estimated driving range, and power consumption impact coefficient, the required charging amount and estimated charging time are calculated, thus generating an initial charging plan. Spatial query algorithms are used to query spatial objects that meet specific conditions in geospatial data. The weighted scoring method is a screening method that assigns weights to different indicators and calculates a comprehensive score. The advantage of this approach is that it can quickly screen out charging facilities that meet the vehicle's needs, initially achieving a reasonable match between the vehicle and charging resources, and improving the efficiency and accuracy of subsequent charging plan optimization.

[0034] Next, with the goal of minimizing the charging wait time in the initial charging scheme, the initial charging scheme is input into the constructed collaborative optimization model for optimization, resulting in charging control schemes for all electric vehicles to be charged. The collaborative optimization model is iteratively updated using multi-agent reinforcement learning technology.

[0035] With the goal of minimizing the charging wait time in the initial charging scheme, the initial charging scheme is input into a pre-constructed collaborative optimization model for optimization, resulting in charging control schemes for all electric vehicles waiting to be charged. The collaborative optimization model is a mathematical model used to coordinate all electric vehicles and charging facility resources, with the core objective of minimizing charging wait time, to achieve collaborative allocation of charging resources. Charging wait time refers to the time an electric vehicle must wait for the preceding vehicle to finish charging after arriving at a candidate charging station before it can begin charging. The initial charging scheme is a preliminary determination of the vehicle-charging station matching and charging-related parameters. The charging control scheme is a directly executable scheme, determined after optimization, specifying the final charging time, charging sequence, and charging amount for each electric vehicle waiting to be charged. Multi-agent reinforcement learning technology is a machine learning technique where multiple agents collaborate, continuously learning and optimizing strategies through interaction with the environment to achieve the optimal overall goal. This approach aims to optimize the inefficiencies in the initial charging scheme, reduce charging wait time, and improve charging efficiency.

[0036] First, a collaborative optimization model is constructed. The model's input includes candidate charging station information, real-time location information, and power consumption impact coefficients for each vehicle in the initial charging plan, as well as data on the geographical distribution of charging facilities, the number of available charging piles, and real-time charging progress. The objective function of the model is set to minimize the total charging wait time of all electric vehicles waiting to be charged. Constraints are also included, such as the charging time for each vehicle not exceeding the charging station's operating hours, each charging pile only charging one vehicle at a time, and the vehicle's charging amount meeting the power consumption needs of subsequent trips. The constraints must be set in conjunction with vehicle trip information and real-time charging facility data to ensure the feasibility of the model's optimization results. After the model is constructed, iterative updates are performed using multi-agent reinforcement learning technology. Each electric vehicle waiting to be charged is treated as an agent, and each charging facility is treated as a node in the environment. The agent's actions include selecting charging stations and adjusting charging times. Environmental feedback includes indicators such as charging wait time and charging facility utilization. The model continuously interacts with the environment, adjusting its strategy based on reward values ​​received from environmental feedback. Reward values ​​are negatively correlated with charging wait time; shorter wait times result in higher rewards. Simultaneously, multiple agents share charging resource information through a collaborative communication mechanism, avoiding resource conflicts and redundant selections. In each iteration, the model adjusts the charging scheme based on the agents' strategies, calculating the charging wait time for the new scheme. If the wait time does not reach its minimum, the iteration continues until the total charging wait time is minimized or the preset number of iterations is reached. After iteration, the initial charging scheme is input into the optimized collaborative optimization model. The model then reallocates charging resources according to the optimal strategy, adjusting the charging sequence and charging time at charging stations for each vehicle, resulting in a charging control scheme for all electric vehicles awaiting charging. The collaborative communication mechanism enables information sharing and collaborative decision-making among multiple agents. The objective function is a mathematical expression used to measure the model's optimization objective. The preset number of iterations is a fixed number set to control the model's iteration efficiency. This approach minimizes charging wait time, achieves optimal allocation of charging resources, improves the charging experience for electric vehicles, and enhances the utilization rate of charging infrastructure.

[0037] Finally, control commands are sent to each electric vehicle waiting to be charged, and these control commands are derived from the charging control scheme.

[0038] Control commands are sent to each electric vehicle waiting to be charged. These commands are generated by the charging control scheme and can be recognized and executed by the vehicle's onboard control system. Specific details include the geographical coordinates and navigation route of the target charging station, the estimated arrival time, charging start and end times, charging power and charge amount parameters, and error handling instructions during the charging process. It has already been clarified that the electric vehicles waiting to be charged are those currently in a charging-ready state. The charging control scheme is an optimized and final executable charging-related plan, which will not be explained again here. This is done to ensure the optimized charging control scheme is implemented effectively, guaranteeing that each electric vehicle can complete charging according to the optimal plan.

[0039] First, the charging control scheme is analyzed and processed. The data parsing module extracts the charging parameters for each electric vehicle to be charged. These parameters are then converted into a command format recognizable by the vehicle control system. During the conversion, a unified vehicle communication protocol, such as the Controller Area Network (CAN) bus protocol, must be followed to ensure the compatibility and recognizability of the control commands. After parsing and conversion, a bidirectional communication link is established between the cloud control platform and the on-board terminals of each electric vehicle using V2X technology. Once the communication link is established, the cloud control platform verifies the identity of each vehicle's on-board terminal. After successful verification, the corresponding control commands are encrypted using advanced encryption standards to prevent tampering or leakage. After encryption, the cloud control platform sends the control commands to the corresponding electric vehicles according to their vehicle identifiers. The vehicle's onboard terminal receives control commands, decrypts them, and then transmits them to the vehicle control system. The control system verifies the command content, confirming that the parameters meet the vehicle's battery performance and driving requirements. It then executes the corresponding control operation, guiding the vehicle to the target charging station and completing charging according to set parameters within the specified time. Simultaneously, it provides real-time feedback on charging status information to the cloud control platform. The data parsing module extracts and converts charging control scheme parameters. The onboard terminal is a device installed on the electric vehicle to receive and transmit control commands and interact with the vehicle control system. The cloud control platform is a remote platform for managing all vehicle charging control command transmissions and status monitoring. Identity verification confirms the legitimacy of the onboard terminal, and encryption protects the control commands. This approach ensures the effective implementation of the charging control scheme, enabling precise control of the charging process for each electric vehicle, improving charging efficiency and safety.

[0040] One embodiment of the present invention provides a charging probability model based on the charging behavior characteristics of electric vehicle users. This model not only focuses on the inherent physical parameters of the vehicle (such as battery capacity and power consumption per unit distance), but also deeply analyzes the user's psychological and behavioral patterns, including their travel habits, charging preferences, and psychological expectations regarding remaining battery power. In this way, the model can more accurately predict the user's charging needs in specific situations and determine the charging power demand of a single electric vehicle accordingly.

[0041] In this model, two key concepts are introduced: minimum acceptable SOC and expected SOC. The minimum acceptable SOC is a crucial threshold for users to decide whether to charge upon arrival at their destination—if the current SOC is below this value, the user will immediately choose to charge; otherwise, charging is unnecessary. A user's range anxiety level is reflected in their expected battery level. Specifically, when the remaining battery level is below the user's expected minimum, the likelihood of grid-connected charging increases significantly; conversely, when the battery level exceeds the expected maximum during charging, the likelihood of interrupting charging also increases. It is noteworthy that users with lower psychological tolerance are more inclined to charge immediately when the battery level reaches the expected minimum, while users with higher psychological tolerance are more likely to terminate charging prematurely when the battery level reaches the expected maximum. Therefore, the probability of a user connecting to the grid for charging is inversely proportional to their psychological tolerance, while the probability of interrupting charging is directly proportional to it.

[0042] To quantify the impact of users' psychological tolerance on charging behavior, the model further establishes a mechanism for determining expected battery capacity. The core idea is that users' expected battery capacity is not fixed but dynamically adjusted based on their psychological characteristics. Specifically, users with lower psychological tolerance have higher requirements for the safety margin of remaining battery capacity—they want to retain more power after arriving at their destination to cope with unexpected trips or avoid the anxiety of "running out of power"; conversely, users with higher psychological tolerance can tolerate lower remaining battery capacity and are more accepting of uncertainties regarding battery life.

[0043] Based on this principle, the model defines two key expected battery levels: the first is the minimum expected battery level upon arrival at a location, which the user believes they must charge immediately below this level, otherwise they will feel uneasy; the second is the minimum expected battery level before leaving a location, which the user hopes to maintain at least before departure. If the battery level falls below this value, the user tends to continue charging rather than set off. Both of these expected levels are driven by the core variable of the user's psychological tolerance: the lower the psychological tolerance (e.g., a value close to 0), the higher the expected battery level; the higher the psychological tolerance (e.g., a value close to 1), the lower the expected battery level.

[0044] To reflect this nonlinear relationship, the model introduces several shape parameters to adjust the sensitivity of expected charge levels to changes in psychological tolerance. By adjusting these parameters, different "anxiety curves" can be generated—for example, for users with low psychological tolerance (e.g., set to 0.4), their expected charge levels rise rapidly near the psychological threshold, exhibiting high sensitivity; while for users with high psychological tolerance (e.g., set to 0.8), their expected charge levels change more gradually, demonstrating stronger adaptability. The resulting grid-connected charging probability and interrupted charging probability curves can realistically reflect the differentiated decision-making tendencies of different user groups when faced with the same charge levels, thus providing refined behavioral basis for subsequent charging demand prediction and strategy optimization.

[0045] Considering the stochastic nature of electric vehicle charging behavior, this model employs a non-stationary Markov chain to describe the dynamic changes in battery charge. Within this framework, each charge level is considered a discrete state, and increases or decreases in charge (caused by charging or driving) constitute state transitions. A key feature of this Markov chain is that the next state depends only on the current state, simplifying the modeling of complex stochastic processes. To ensure the model objectively reflects the reversibility of the charging and power consumption processes, detailed equilibrium conditions are introduced during sampling. Based on the electric vehicle's operating state, the entire state space is divided into three basic states: charging, parking, and driving, further combined into two modes—grid-connected mode (connected to the grid and absorbing power) and off-grid mode (driving or parking, not drawing power from the grid).

[0046] The model further defines the specific rules for state transitions. For example, starting charging from state i, the probability of an increase in battery power Δk is determined by factors such as charger efficiency, rated power, and charging / discharging time; while the decrease in battery power due to driving depends on mileage, power consumption per unit distance, and average vehicle speed. Furthermore, the model assumes that the initial battery power of the electric vehicle follows a uniform distribution and provides a method for calculating the battery power after reaching the destination. Based on this, the charging power demand of a single electric vehicle can be determined, and the total charging power demand at any given time is obtained by summing the individual demands of all vehicles currently charging. This prediction process consists of three modules: input, calculation, and output. The input module integrates multi-source information such as vehicle parameters, driving habits, power system, and traffic system; the calculation module extrapolates the various trips and charging parameters of a single vehicle in chronological order; and the output module aggregates individual demands in the spatiotemporal dimension to form the overall electricity demand distribution.

[0047] After obtaining the total power demand, the system enters the charging decision-making phase. The city is divided into several regions, and the set R represents all regions. A new decision moment occurs whenever a user completes a trip, ends charging, or is unable to proceed with the next trip due to insufficient battery power. All such moments constitute the set U. At each decision moment u, the user must choose an action aᵤ based on the current vehicle state sᵤ. Here, the binary variable xᵤ indicates whether the user waits for the next trip (xᵤ=0) or begins charging (xᵤ=1); the continuous variable yᵤ indicates the planned charging level. The vehicle state sᵤ is a composite vector containing the spatiotemporal location (current region nᵤ and time tᵤ) and the battery state (remaining battery power kᵤ, battery power change gᵤ, and battery health qᵤ).

[0048] When a user chooses to charge, they must travel to the nearest charging station within their area. Due to the limited number of charging facilities in each area, users must also consider queuing time, a random variable affected by the region and arrival time. The total charging time (i.e., "offline time") consists of the travel time to the charging station, queuing time, and actual charging time, while the charging cost is directly related to the planned charging amount. For the next trip rᵤ (defined by the origin region, destination region, and trip time), the system determines whether the remaining battery power is sufficient to support the trip. If the battery power is insufficient, the user cannot proceed with the trip and must make a new decision.

[0049] The entire charging and travel decision-making process is modeled as a sequential decision problem. Its state space S encompasses all possible scenarios from daily use to battery degradation; the action space A consists of the aforementioned (xᵤ, yᵤ) pairs. The state transition rules specify in detail how the battery level, the change in accumulated charge, and the spatiotemporal location are updated after taking action aᵤ. For example, if charging is chosen, the remaining charge increases, and gᵤ is recorded as a positive value; if traveling is chosen, the charge decreases, and gᵤ is recorded as a negative value; if the charge is insufficient to support the journey, the state remains unchanged.

[0050] The core objective of this invention is to find an optimal strategy π* to maximize the "optimal use" value of electric vehicles. To this end, each electric vehicle is embedded with an intelligent agent. This agent assists the user in making real-time decisions based on local state information and records experience data such as state, actions, and immediate rewards during the decision-making process. All agents periodically upload data to a central learning agent, which integrates global information, updates the global strategy using reinforcement learning algorithms, and feeds the optimized strategy back to each terminal agent. This centralized-decentralized architecture not only promotes experience sharing but also effectively balances exploration and utilization through multi-agent strategies, enabling individual agents, even those lacking experience in a particular area, to make better decisions with the help of collective wisdom.

[0051] Through the aforementioned mechanism, the system can continuously iterate and optimize charging and travel strategies to adapt to dynamic and uncertain external environments. Ultimately, this solution not only effectively alleviates users' range anxiety by ensuring the vehicle has sufficient charge when needed through advance planning, but also significantly improves the psychological comfort and overall user experience of driving electric vehicles.

[0052] One embodiment of the present invention provides a charging control system for electric vehicles. For details, please refer to [link / reference]. Figure 2 , Figure 2 The diagram shown is a flowchart of a charging control system for an electric vehicle according to one embodiment of the present invention, including: Module 11 is used to acquire historical charging data for each electric vehicle to be charged; Extraction module 12 is used to extract charging behavior features from each historical charging data and determine the degree of loss of each electric vehicle to be charged based on the charging behavior features. The loss module 13 is used to obtain the corresponding power consumption impact coefficient based on the degree of loss of each electric vehicle to be charged. The initial module 14 is used to generate an initial charging plan based on the geographical distribution data of charging facilities and the real-time location information, travel information and power consumption impact coefficient of each electric vehicle to be charged. The optimization module 15 is used to minimize the charging waiting time in the initial charging scheme by inputting the initial charging scheme into the constructed collaborative optimization model for optimization processing, so as to obtain the charging control scheme of all electric vehicles to be charged. The control module 16 is used to send control commands to each electric vehicle to be charged, wherein the control commands are obtained from the charging control scheme.

[0053] Preferably, the extraction module 12 includes: The analysis unit is used to analyze the charging time and corresponding power data in historical charging data to obtain charging anxiety characteristics; The statistical unit is used to statistically process the single-charge replenishment amount in historical charging data to obtain charging depth characteristics. The indicator unit is used to construct battery loss assessment indicators based on charging anxiety characteristics and charging depth characteristics. The loss unit is used to obtain the degree of loss of each electric vehicle to be charged based on the battery loss assessment index.

[0054] Preferably, the analysis unit includes: The charging time and corresponding power data are extracted from historical charging data. Based on the power data, the charging time is classified to obtain the categorized charging behavior; By analyzing and processing the categorized charging behaviors, charging anxiety characteristics were obtained.

[0055] Preferably, the initial module 14 includes: The constraint unit is used to process real-time location information and travel information to obtain constraint conditions; Candidate units are used to process the geographical distribution data of charging facilities based on constraints to obtain a set of candidate charging stations. The filtering unit is used to filter the candidate charging station set using the power consumption impact coefficient and generate an initial charging scheme.

[0056] Preferred collaborative optimization models include: The collaborative optimization model utilizes multi-agent reinforcement learning techniques to iteratively update the collaborative optimization model.

[0057] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0058] Accordingly, embodiments of the present invention provide a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform steps in the electric vehicle charging control method of the above embodiments, for example... Figure 1 Steps S1 to S6 as described above.

[0059] This invention reconstructs the electric vehicle charging control logic from the perspective of system collaboration and data-driven approaches. It no longer relies on the driver's passive response after a State of Charge (SOC) alarm, but instead extracts charging behavior characteristics by analyzing historical charging data, quantifies the battery wear level of each electric vehicle waiting to be charged, and introduces a power consumption impact coefficient to more accurately reflect its actual energy consumption characteristics. Based on this, it integrates the geographical distribution of charging facilities, real-time vehicle location, and travel information to generate an initial charging plan that considers individual differences. Furthermore, it uses a collaborative optimization model to perform global optimization with the goal of minimizing charging waiting time, outputting a unified charging control plan and issuing control commands. This mechanism realizes a shift from passive emergency charging to proactive predictive guidance, not only improving the efficiency and accuracy of charging resource allocation but also effectively avoiding range anxiety and mid-journey stoppages caused by inaccurate power estimation or unreasonable route planning, thus ensuring driving safety.

[0060] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A charging control method for an electric vehicle, characterized in that, include: Obtain historical charging data for each electric vehicle waiting to be charged; Extract charging behavior features from each of the historical charging data, and determine the degree of damage of each of the electric vehicles to be charged based on the charging behavior features; Based on the degree of loss of each electric vehicle to be charged, a corresponding power consumption impact coefficient is obtained; Based on the geographical distribution data of charging facilities, as well as the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged, an initial charging plan is generated. With the goal of minimizing the charging waiting time in the initial charging scheme, the initial charging scheme is input into the constructed collaborative optimization model for optimization processing to obtain the charging control schemes for all the electric vehicles to be charged. Control commands are sent to each of the electric vehicles to be charged, wherein the control commands are obtained from the charging control scheme.

2. The charging control method for an electric vehicle as described in claim 1, characterized in that, The step of extracting charging behavior features from each of the historical charging data and determining the degree of damage of each of the electric vehicles to be charged based on the charging behavior features includes: By analyzing the charging times and corresponding power data in the historical charging data, charging anxiety characteristics are obtained. Statistical processing is performed on the single-charge replenishment amount in the historical charging data to obtain charging depth characteristics; Based on the charging anxiety characteristics and the charging depth characteristics, a battery loss assessment index is constructed. Based on the battery loss assessment index, the degree of loss of each of the electric vehicles to be charged is obtained.

3. The charging control method for an electric vehicle as described in claim 2, characterized in that, The analysis of charging times and corresponding power data in the historical charging data yields charging anxiety characteristics, including: The charging time and corresponding power data are extracted from the historical charging data. Based on the power data, the charging time is classified to obtain classified charging behaviors; The charging anxiety characteristics are obtained by analyzing and processing the categorized charging behaviors.

4. The charging control method for an electric vehicle as described in claim 1, characterized in that, The initial charging plan is generated based on the geographical distribution data of charging facilities, the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged, including: The real-time location information and the travel information are processed to obtain the constraints; Based on the constraints, the geographical distribution data of the charging facilities are processed to obtain a set of candidate charging stations; The power consumption impact coefficient is used to filter the candidate charging station set to generate an initial charging scheme.

5. The charging control method for an electric vehicle as described in claim 1, characterized in that, The collaborative optimization model includes: The collaborative optimization model is iteratively updated using multi-agent reinforcement learning technology.

6. A charging control system for an electric vehicle, characterized in that, include: The acquisition module is used to acquire historical charging data for each electric vehicle to be charged. An extraction module is used to extract charging behavior features from each of the historical charging data, and to determine the degree of loss of each of the electric vehicles to be charged based on the charging behavior features. The loss module is used to obtain the corresponding power consumption impact coefficient based on the degree of loss of each electric vehicle to be charged. An initial module is used to generate an initial charging plan based on the geographical distribution data of charging facilities, the real-time location information, travel information, and power consumption impact coefficient of each electric vehicle to be charged. The optimization module is used to minimize the charging waiting time in the initial charging scheme by inputting the initial charging scheme into the constructed collaborative optimization model for optimization processing, so as to obtain the charging control scheme of all the electric vehicles to be charged. The control module is used to send control commands to each of the electric vehicles to be charged, wherein the control commands are obtained from the charging control scheme.

7. The charging control system for an electric vehicle as described in claim 6, characterized in that, The extraction module includes: The analysis unit is used to analyze the charging time and corresponding power data in the historical charging data to obtain charging anxiety characteristics; The statistical unit is used to perform statistical processing on the single charging replenishment amount in the historical charging data to obtain charging depth characteristics. The indicator unit is used to construct a battery loss assessment indicator based on the charging anxiety feature and the charging depth feature; A loss unit is used to obtain the degree of loss of each of the electric vehicles to be charged based on the battery loss assessment index.

8. The charging control system for an electric vehicle as described in claim 7, characterized in that, The analysis unit includes: The charging time and corresponding power data are extracted from the historical charging data. Based on the power data, the charging time is classified to obtain classified charging behaviors; The charging anxiety characteristics are obtained by analyzing and processing the categorized charging behaviors.

9. The charging control system for an electric vehicle as described in claim 6, characterized in that, The initial module includes: The constraint unit is used to process the real-time location information and the travel information to obtain constraint conditions; The candidate unit is used to process the geographical distribution data of the charging facilities based on the constraints to obtain a set of candidate charging stations. The filtering unit is used to filter the candidate charging station set using the power consumption impact coefficient and generate an initial charging scheme.

10. The charging control system for an electric vehicle as described in claim 6, characterized in that, The collaborative optimization model includes: The collaborative optimization model is iteratively updated using multi-agent reinforcement learning technology.

Images