Charging scheduling method and device, cloud server and storage medium

Abstract

This application provides a charging scheduling method, apparatus, cloud server, and storage medium. The method, applied to vehicles, includes: acquiring the status of at least one charging station and the status of at least one vehicle waiting to be charged within a target area; determining an observation vector for each vehicle waiting to be charged based on the status of the at least one charging station and the status of the at least one vehicle waiting to be charged; and generating a charging scheduling result based on preset time cost and load rate constraints, using the observation vector of each vehicle waiting to be charged. Therefore, this application solves the problems of overload and traffic congestion at popular charging stations caused by users blindly selecting charging stations due to a lack of dynamic information, as well as scheduling response delays under high concurrency requests. It can significantly improve the balance and response efficiency of charging resource allocation, reduce the average waiting time for users, and improve overall charging service satisfaction.

Description

Technical Field

[0001] This application relates to the field of vehicles, and more specifically, to charging scheduling methods, apparatus, cloud servers, and storage media in the field of vehicles. Background Technology

[0002] With the rapid growth of electric vehicle ownership, urban charging infrastructure is facing unprecedented scheduling and service pressures. The current charging ecosystem suffers from core problems such as users blindly choosing charging stations, unbalanced station loads, slow system response, and low efficiency in processing multiple requests. These problems not only reduce the user's charging experience but also pose potential threats to power grid security and urban traffic. When a vehicle needs charging, the relevant technologies typically rely on the integration of map navigation platforms and charging service platforms to assist users in making decisions. The system first obtains the user's real-time location through GPS (Global Positioning System) or onboard positioning modules, and then connects to the data interface of third-party charging station operators to aggregate static information from various charging stations, including geographical location, charging station type, and rated power. Based on this static information, users can independently filter and select the nearest charging station or one that matches their vehicle model's needs on the map interface.

[0003] However, when the demand for electric vehicle charging in a region surges, a large number of users simultaneously select charging stations based on the nearest or matching power displayed on the map. This easily leads to a concentrated influx of users to popular stations in a short period, while other stations that are slightly farther away but actually available are ignored, resulting in a severe uneven distribution of charging resources. Secondly, because the platform only provides static data and lacks feedback on key dynamic indicators such as real-time queue length, estimated waiting time, and current grid load, users cannot predict the actual availability after arrival, leading to a significant increase in blind decision-making and a degraded user experience. Summary of the Invention

[0004] This application provides a charging scheduling method, device, cloud server, and storage medium. This method solves the problems of overload and traffic congestion at popular charging stations caused by users blindly selecting charging stations due to a lack of dynamic information, as well as scheduling response delays under high concurrency requests. It can significantly improve the balance and response efficiency of charging resource allocation, reduce the average waiting time for users, and improve overall charging service satisfaction.

[0005] Firstly, a charging scheduling method is provided, the method comprising: Acquire the status of at least one charging station and at least one vehicle to be charged within the target area; The observation vector for each vehicle to be charged is determined based on the state of the at least one charging station and the state of the at least one vehicle to be charged. Based on preset time cost and load rate constraints, charging scheduling results are generated according to the observation vector of each vehicle to be charged.

[0006] By using the above-mentioned technical means, the real-time status of charging stations and vehicles waiting to be charged is acquired synchronously, and an observation vector for each vehicle waiting to be charged is constructed. Then, under the preset time cost and load rate constraints, a charging scheduling result is generated, so that the scheduling decision reflects the actual arrival and waiting time of users, and avoids charging station overload or resource idleness. This improves the overall operating efficiency and stability of the charging network while ensuring user experience.

[0007] In conjunction with the first aspect, in some possible implementations, determining the observation vector for each vehicle to be charged based on the state of the at least one charging station and the state of the at least one vehicle to be charged includes: The number of queues, the number of currently available charging piles, and the number of available charging piles in the predicted time period are determined based on the status of the at least one charging station. Based on the status of the at least one vehicle to be charged, determine the current location, current remaining battery power, current driving speed, charging power demand type, and driving time of each vehicle to reach each charging station. The observation vector for each vehicle to be charged is obtained based on the number of queues at each charging station, the number of currently available charging piles, the number of available charging piles within the predicted time period, the current location, current remaining battery power, current driving speed, charging power demand type, and the driving time of each vehicle to be charged to each charging station.

[0008] By using the above-mentioned technical means, an observation vector is constructed based on the number of queues at charging stations, the current and predicted number of available charging piles, as well as the current location, remaining battery power, driving speed, charging power demand type, and travel time to each charging station of the vehicles waiting to be charged. This allows the observation vector of each vehicle to accurately reflect the matching relationship between the charging demand of each vehicle waiting to be charged and the service capacity of each charging station, thereby providing a quantitative basis for subsequent scheduling.

[0009] Combining the first aspect and the above implementation methods, in some possible implementation methods, the step of generating charging scheduling results based on the observation vector of each vehicle to be charged, according to preset time cost and load rate constraints, includes: Based on preset time cost and load rate constraints, an objective function is constructed, and based on the objective function, a charging scheduling result is generated according to the observation vector of each vehicle to be charged.

[0010] By using the above-mentioned technical means, an objective function is constructed based on preset time cost and load rate constraints, and charging scheduling results are generated by combining the observation vector of each vehicle to be charged. This can optimize the total time cost while actively avoiding charging station overload or resource idleness, thereby improving the overall service reliability and resource utilization efficiency.

[0011] In conjunction with the first aspect and the above implementation methods, in some possible implementation methods, before obtaining the status of at least one charging station and at least one vehicle to be charged within the target area, the method further includes: Get the number of charging requests from all vehicles waiting to be charged within the target area in the same time window, and determine whether the number of charging requests is less than a preset threshold. If the number of charging requests is less than the preset threshold, then the current location of each vehicle to be charged and the current location of each charging station are obtained. Based on the current location of each vehicle to be charged and the current location of each charging station, at least one candidate charging station is determined for each vehicle to be charged. The scores of at least one candidate charging station corresponding to each vehicle to be charged are determined to obtain a score set of all candidate charging stations corresponding to each vehicle to be charged. The highest score corresponding to each vehicle to be charged is selected from the score set of all candidate charging stations corresponding to each vehicle to be charged. The charging scheduling result of each vehicle to be charged is determined according to the candidate parking station corresponding to the highest score of each vehicle to be charged.

[0012] By employing the aforementioned technical means, before scheduling, it is determined whether the number of charging requests within the target area in the same time window is less than a preset threshold. When the request volume is low, a scoring strategy is used to avoid system delays or waste of computing resources. At the same time, by combining the real-time location information of vehicles and charging stations, candidate charging stations are screened for each vehicle, and the optimal option is selected based on the score. This not only improves the matching efficiency and utilization rate of charging resources, but also reduces the vehicle's empty driving distance and user waiting time, thereby optimizing the response speed of the charging scheduling system and improving the user experience.

[0013] In combination with the first aspect and the above implementation methods, in some possible implementation methods, determining the score of at least one candidate charging station corresponding to each vehicle to be charged includes: Determine the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging stations and the queue number for each vehicle's at least one candidate charging station; The score of each vehicle is calculated by weighting the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging piles at each candidate charging station, and the number of queues.

[0014] By using the aforementioned technical means, and comprehensively considering the distance between the vehicle and the candidate charging station, the number of currently available charging piles, and the number of people in the queue, and by applying a weighted score, the service quality of each candidate charging station for the corresponding vehicle can be more comprehensively and accurately reflected, thereby improving user satisfaction.

[0015] In conjunction with the first aspect and the above implementation methods, in some possible implementation methods, determining at least one candidate charging station corresponding to each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station includes: Based on the current location of each vehicle to be charged and the current location of each charging station, determine whether there is at least one charging station within a preset range of each vehicle to be charged; If at least one charging station exists within a preset range for each vehicle to be charged, then at least one candidate charging station corresponding to each vehicle to be charged is determined based on the charging stations within the preset range for each vehicle to be charged.

[0016] By using the above-mentioned technical means, based on the current location of each vehicle to be charged, existing charging stations within its preset range are selected as candidate objects. This can effectively limit the search space, avoid redundant calculations for distant or unrelated charging stations, and thus significantly reduce the computational complexity and response latency of the scheduling algorithm.

[0017] In combination with the first aspect and the above implementation methods, in some possible implementation methods, after determining whether there is at least one charging station within a preset range of each vehicle to be charged, the following steps are included: If there is no charging station within the preset range of any vehicle, the preset range of any vehicle is updated by a preset step size, and it is re-determined whether there is at least one charging station within the preset range of any vehicle, until at least one charging station exists within the preset range of any vehicle.

[0018] By employing the aforementioned technical means, when no charging station is found within the vehicle's preset range, the search range is dynamically expanded with a preset step size and the judgment is iteratively made until at least one feasible charging station is obtained. This effectively solves the problem of missing candidate stations caused by an initial range that is too small or the sparse distribution of charging facilities in the area. At the same time, the gradual expansion method avoids the waste of computing resources.

[0019] Secondly, a charging scheduling device is provided, the device comprising: The acquisition module is used to acquire the status of at least one charging station and the status of at least one vehicle to be charged within the target area; An observation module is used to determine the observation vector for each vehicle to be charged based on the status of the at least one charging station and the status of the at least one vehicle to be charged. The scheduling module is used to generate charging scheduling results based on the observation vector of each vehicle to be charged, according to preset time cost and load rate constraints.

[0020] In conjunction with the first aspect, in some possible implementations, the observation module includes: The first determining unit is used to determine the number of queues, the number of currently available charging piles, and the number of available charging piles in the predicted time period for each charging station based on the status of the at least one charging station. The second determining unit is used to determine the current location, current remaining battery power, current driving speed, charging power demand type, and driving time of each vehicle to be charged to each charging station based on the status of the at least one vehicle to be charged. The first generation unit is used to obtain the observation vector of each vehicle to be charged based on the number of queues at each charging station, the number of currently available charging piles and the number of available charging piles in the predicted time period, as well as the current location, current remaining battery power, current driving speed, charging power demand type and driving time of each vehicle to be charged to each charging station.

[0021] In combination with the first aspect and the above implementation methods, in some possible implementations, the scheduling module includes: The second generation unit is used to construct an objective function based on the preset time cost and load rate constraints, and to generate charging scheduling results based on the objective function and the observation vector of each vehicle to be charged.

[0022] In conjunction with the first aspect and the above implementation methods, in some possible implementation methods, before acquiring the status of at least one charging station and the status of at least one vehicle to be charged within the target area, the acquisition module further includes: The first acquisition unit is used to acquire the number of charging requests of all vehicles waiting to be charged in the same time window within the target area, and to determine whether the number of charging requests is less than a preset threshold. The second acquisition unit is used to acquire the current location of each vehicle to be charged and the current location of each charging station when the number of charging requests is less than the preset threshold. The third determining unit is used to determine at least one candidate charging station corresponding to each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station. The third generation unit is used to determine the score of at least one candidate charging station corresponding to each vehicle to be charged to obtain a score set of all candidate charging stations corresponding to each vehicle to be charged, select the highest score corresponding to each vehicle to be charged from the score set of all candidate charging stations corresponding to each vehicle to be charged, and determine the charging scheduling result of each vehicle to be charged according to the candidate parking station corresponding to the highest score of each vehicle to be charged.

[0023] In combination with the first aspect and the above implementation methods, in some possible implementations, the third generation unit includes: The determination sub-unit is used to determine the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging piles and the queue number of each vehicle's at least one candidate charging station; The calculation subunit is used to calculate the score of at least one candidate charging station corresponding to each vehicle based on the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging piles and the queue number of at least one candidate charging station for each vehicle.

[0024] In combination with the first aspect and the above implementation methods, in some possible implementation methods, the determining sub-unit includes: The determination sub-component is used to determine, based on the current location of each vehicle to be charged and the current location of each charging station, whether there is at least one charging station within a preset range of each vehicle to be charged. A determining sub-component is used to determine at least one candidate charging station corresponding to each vehicle to be charged, based on the charging stations within the preset range of each vehicle to be charged, when at least one charging station exists within the preset range of each vehicle to be charged.

[0025] In combination with the first aspect and the above implementation methods, in some possible implementation methods, after determining whether at least one charging station exists within a preset range of each vehicle to be charged, the determining sub-component includes: The update sub-component is used to update the preset range of any vehicle with a preset step size when there is no charging station within the preset range of any vehicle, and to re-determine whether there is at least one charging station within the preset range of the vehicle until at least one charging station exists within the preset range of the vehicle.

[0026] Thirdly, a cloud server is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to perform the method of the first aspect or any possible implementation thereof.

[0027] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof.

[0028] Fifthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof. Attached Figure Description

[0029] Figure 1 This is a flowchart of a charging scheduling method provided according to an embodiment of this application; Figure 2 This is a block diagram of a charging scheduling device provided according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a cloud server provided according to an embodiment of this application. Detailed Implementation

[0030] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0031] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0032] Before introducing the charging scheduling method of the embodiments of this application, let's first introduce the technical problems of electric vehicle charging scheduling schemes in related technologies.

[0033] (1) Users blindly select charging stations, resulting in low charging satisfaction. The specific reasons are as follows: Electric vehicle charging takes a long time, and when users initiate a charging request, they cannot obtain core information such as the number of available charging stations and queuing status in real time; at the same time, users' driving routes are easily affected by factors such as traffic congestion and weather changes, which leads to users having to rely on experience or randomly select charging stations, and they cannot make reasonable choices that suit their own journeys, ultimately resulting in high blindness in charging station selection and low charging satisfaction.

[0034] (2) Load imbalance at charging stations leads to power grid safety risks and traffic congestion. The specific reasons are as follows: Due to the lack of effective global scheduling guidance, a large number of electric vehicles are charging in a disorderly manner. Charging stations in some geographically advantageous or densely populated areas are prone to a surge in charging requests, while charging stations in remote areas may be idle for a long time. The uneven distribution of charging demand will lead to a sudden increase in local power grid load, exceeding the power grid carrying capacity threshold and causing safety risks. At the same time, the concentrated charging vehicle flow will also cause traffic congestion around the charging stations.

[0035] (3) The scheduling system has high latency and cannot adapt to large-scale real-time charging scenarios. The specific reasons are as follows: The centralized intelligent scheduling system in the relevant technology needs to collect real-time status data of all charging stations and electric vehicles in the whole area. It has the problems of high status / action dimension and long data transmission link, resulting in significant delay in global data acquisition. When the number of charging requests in the area surges, the system cannot respond quickly to a large number of concurrent requests, and thus cannot adapt to the city-level large-scale real-time charging recommendation scenario.

[0036] (4) Long waiting time and low charging efficiency in multi-request scenarios. The specific reasons are: the scheduling schemes in related technologies mostly use the charging station as the intelligent agent. When multiple charging requests occur in a short period of time, the decision is made based on the order of the charging requests. It is impossible to execute the charging station allocation action in parallel for multiple charging requests. The sequential decision will cause the electric vehicle that initiates the request later to wait for the previous request to complete the allocation before it can enter the decision process, which directly prolongs the overall charging waiting time and reduces the charging efficiency.

[0037] To address the aforementioned technical problems, this application provides a charging scheduling method. This method acquires the status of at least one charging station and at least one vehicle waiting to be charged within a target area. Based on the status of these two locations, an observation vector for each vehicle is determined. Then, based on preset time cost and load rate constraints, a charging scheduling result is generated using the observation vector of each vehicle. This solution addresses the problems of overload and traffic congestion at popular charging stations due to users blindly selecting charging stations without dynamic information, as well as scheduling response delays under high concurrency requests. It significantly improves the balance and response efficiency of charging resource allocation, reduces average user waiting time, and enhances overall charging service satisfaction.

[0038] Specifically, such as Figure 1 As shown, the charging scheduling method proposed in this application includes the following steps: In step S101, the status of at least one charging station and the status of at least one vehicle to be charged within the target area are obtained.

[0039] The target area is a pre-defined geographical area used to define the operational boundaries of the charging dispatch service. The status of the charging station reflects comprehensive information about the service capacity of the charging station in the current and future period, which may include: station location, total number of charging piles, number of currently available charging piles, queue number, and real-time occupancy status of each charging pile. The vehicle waiting to be charged is the vehicle that initiated the charging request. The status of the vehicle waiting to be charged is the operation and demand information of the vehicle that initiated the charging request at the time of the request, which may include the vehicle's current location coordinates, remaining battery percentage, user charging preference type, current driving speed, and destination constraints.

[0040] Understandably, when there are a large number of vehicles waiting to be charged and charging station resources are limited, allowing users to choose charging stations independently can easily lead to problems such as overload at popular stations and vehicle congestion. Therefore, when there are a large number of vehicles waiting to be charged in the target area at the same time window, the status of charging stations and vehicles waiting to be charged in the target area can be obtained, and charging scheduling can be carried out accordingly.

[0041] Specifically, within a preset time window (e.g., within 1 minute), the system aggregates all requests from vehicles waiting to be charged, forms a scheduling batch, and obtains the status of all vehicles waiting to be charged within the scheduling batch and the status of charging stations in the target area, so as to perform charging scheduling based on the status of all vehicles waiting to be charged within the scheduling batch and the status of charging stations in the target area. It should be noted that...

[0042] Optionally, in some embodiments, before obtaining the status of at least one charging station and at least one vehicle to be charged within the target area, the method further includes: obtaining the number of charging requests from all vehicles to be charged within the target area in the same time window, and determining whether the number of charging requests is less than a preset threshold; if the number of charging requests is less than the preset threshold, obtaining the current location of each vehicle to be charged and the current location of each charging station; determining at least one candidate charging station corresponding to each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station; determining the score of at least one candidate charging station corresponding to each vehicle to be charged to obtain a score set of all candidate charging stations corresponding to each vehicle to be charged; selecting the highest score corresponding to each vehicle to be charged from the score set of all candidate charging stations corresponding to each vehicle to be charged; and determining the charging scheduling result of each vehicle to be charged based on the candidate parking station corresponding to the highest score corresponding to each vehicle to be charged.

[0043] Furthermore, in some embodiments, determining the score of at least one candidate charging station corresponding to each vehicle to be charged includes: determining the distance between each vehicle and the corresponding at least one candidate charging station, the number of currently available charging piles and the queue number of the at least one candidate charging station for each vehicle; and performing a weighted calculation based on the distance between each vehicle and the corresponding at least one candidate charging station, the number of currently available charging piles and the queue number of the at least one candidate charging station for each vehicle to be charged, to obtain the score of at least one candidate charging station corresponding to each vehicle to be charged.

[0044] The time window is a pre-set scheduling cycle unit, preferably 1 minute, used to aggregate continuously arriving charging requests to form a scheduling batch; the preset threshold is a pre-set critical value for the number of requests, preferably 3 to 10 vehicles per time window, used to distinguish between low-density and high-density scheduling scenarios; the candidate charging station is an available charging station that meets the charging needs of any vehicle to be charged within a preset range (e.g., within a radius of 5 kilometers); the score is used to quantify the recommendation degree of the corresponding candidate charging station.

[0045] Understandably, when the number of charging requests is low, vehicles are less likely to cause congestion. In this case, vehicles can adopt a lightweight scheduling method to select the optimal charging station.

[0046] Specifically, this embodiment of the application can obtain the number of charging requests from all vehicles waiting to be charged within the same time window in the target area, and determine whether the number of charging requests is less than a preset threshold. If the number of charging requests is less than the preset threshold, it can be determined as a low-density scenario, and enter a lightweight recommendation mode. The system obtains the current location of each vehicle waiting to be charged and the geographical location of each charging station, and filters at least one candidate charging station for each vehicle based on geographical proximity and service compatibility (such as fast charging / slow charging matching). Subsequently, for each vehicle, the distance between it and the corresponding at least one candidate charging station is calculated, and a weighted calculation is performed by combining the number of currently available charging piles and the number of queues at each candidate charging station. For example, the calculation formula can be: Score = ,in, , , The system uses weighted coefficients for distance, the number of currently available charging stations, and the number of people in the queue, respectively, to generate a score set for each vehicle to be charged and its corresponding candidate stations. Finally, the candidate charging station with the highest score is selected as the charging scheduling result for that vehicle, and the recommendation information is sent to the user terminal, which significantly reduces system response latency and computational overhead.

[0047] Therefore, this application embodiment uses a weighted scoring system that incorporates distance, the number of currently available charging piles, and the number of queues, taking into account both user travel efficiency and the real-time service capabilities of the stations. On the other hand, this simplified process is only enabled when the number of requests is below a preset threshold, effectively saving computing resources and improving response speed. In addition, this solution can complement charging scheduling strategies in high-density scenarios, improving user experience while reducing system latency.

[0048] Optionally, in some embodiments, determining at least one candidate charging station corresponding to each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station includes: determining whether there is at least one charging station within a preset range of each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station; if there is at least one charging station within the preset range of each vehicle to be charged, then determining at least one candidate charging station corresponding to each vehicle to be charged based on the charging stations within the preset range of each vehicle to be charged.

[0049] The preset range is a pre-defined geographical area radius centered on the current location of the vehicle to be charged, used to limit the search boundary of candidate charging stations.

[0050] Specifically, in this embodiment of the application, the current location of all vehicles to be charged and the geographical location information of all charging stations in the area are obtained. For each vehicle to be charged, a search is performed with its current location as the center and a preset range as the radius to determine whether there is a charging station in the area. If there is, the charging stations located within the preset range are used as the candidate charging station set for the vehicle.

[0051] Therefore, the embodiments of this application effectively reduce the search space for charging station selection, avoid indiscriminate traversal of all charging stations, and significantly improve computational efficiency and response speed; at the same time, by limiting the preset range, it ensures that the recommended candidate charging stations are close to the vehicle, reducing the user's driving mileage and waiting time, and improving the user experience.

[0052] Optionally, in some embodiments, after determining whether there is at least one charging station within a preset range of each vehicle to be charged, the process includes: if there is no charging station within the preset range of any vehicle, then updating the preset range of any vehicle by a preset step size, and re-determining whether there is at least one charging station within the preset range of any vehicle, until there is at least one charging station within the preset range of any vehicle.

[0053] The preset step size is a predefined range expansion increment, for example, 1 kilometer.

[0054] Specifically, after screening charging stations based on an initial preset range, if the embodiments of this application identify vehicles that have not been matched with any charging stations within the initial range, the system increments its search radius by a preset step size (e.g., 1 kilometer) to form a new extended range, and performs the search again to determine whether there is a charging station within the new range; if there is still no charging station, it continues to iterate and expand at the same step size until at least one charging station is successfully matched within the extended range.

[0055] Therefore, the embodiments of this application effectively avoid the situation where there are no candidate charging stations due to uneven distribution of charging infrastructure or vehicles located in remote areas, ensuring that every vehicle waiting to be charged can obtain feasible charging options; by gradually expanding rather than conducting a large-scale search at once, both computational efficiency and coverage integrity are taken into account, avoiding resource waste.

[0056] In step S102, the observation vector of each vehicle to be charged is determined based on the state of at least one charging station and the state of at least one vehicle to be charged.

[0057] Further, in some embodiments, determining the observation vector for each vehicle to be charged based on the status of at least one charging station and the status of at least one vehicle to be charged includes: determining the queue size, the number of currently available charging piles, and the number of available charging piles within a predicted time period for each charging station based on the status of at least one charging station; determining the current location, current remaining battery level, current driving speed, charging power demand type, and travel time for each vehicle to arrive at each charging station based on the status of at least one vehicle to be charged; and obtaining the observation vector for each vehicle to be charged based on the queue size, the number of currently available charging piles, the number of available charging piles within a predicted time period, the current location, current remaining battery level, current driving speed, charging power demand type, and travel time for each vehicle to arrive at each charging station.

[0058] Among them, the queue number is the number of vehicles waiting to charge in the charging station; the number of currently available charging piles is the number of charging piles in the charging station that are currently not occupied and can provide charging services for vehicles; the number of available charging piles in the predicted time period is the number of available charging piles in the predicted time period; the current remaining power is the current state of charge of the vehicle's battery; the current driving speed is the current moving speed of the vehicle; the charging power demand type is the type of charging power required by the vehicle, such as slow charging, fast charging, or supercharging; the observation vector is a multi-dimensional data set constructed for each vehicle to be charged, used to describe the comprehensive perception state of each vehicle to be charged in the current charging environment.

[0059] It is understood that the embodiments of this application can construct the observation vector of each vehicle to be charged from the following dimensions to provide key decision information, thereby avoiding blind decision-making due to missing information and preventing computational difficulties caused by dimensional redundancy.

[0060] 1. Vehicle status (4-dimensional): Current location, current remaining battery power, current driving speed, and charging power demand type of each vehicle; 2. Trip-related status (1-dimensional): Travel time for each vehicle to reach each charging station; 3. Charging station status (3-dimensional): Queue size, number of currently available charging piles, and number of available charging piles within a predicted time period (e.g., the next 30 minutes) for each charging station.

[0061] Specifically, this embodiment of the application can obtain the queue size, the number of currently available charging piles, and the number of available charging piles within a predicted time period obtained through a preset prediction model for each charging station. Simultaneously, it obtains the status of each vehicle waiting to be charged, including its current location, current remaining battery power, current driving speed, and charging power demand type. Based on map software, it determines the travel time for each vehicle to reach each candidate charging station. Subsequently, the above two types of information are structurally integrated: the queue size, the number of currently available charging piles, and the number of available charging piles within the predicted time period for each charging station are combined with the current location, current remaining battery power, current driving speed, charging power demand type, and travel time for each vehicle to reach each charging station to obtain an observation vector containing multi-dimensional numerical features, providing reliable data for charging scheduling decisions.

[0062] It should be noted that before constructing the observation vector for each vehicle to be charged, the number of available charging piles within the prediction time period can be generated through a preset prediction module. In this embodiment, the number of available charging piles within the prediction time period can be predicted by combining historical behavior, real-time occupancy, and external factors (weather, holidays). Specifically, the input data of the prediction model needs to include three core features to ensure the accuracy and real-time performance of the prediction: 1. Historical charging behavior data: historical charging records of each charging station (start time, end time, charging duration, user type for each charge), used to mine the periodic patterns of charging behavior; 2. Real-time operation data: current occupancy status of each charging pile (idle / charging / faulty), estimated end time of vehicles currently charging, and number of vehicles queuing at the charging station; 3. External correlation data: real-time traffic conditions (affecting the time it takes for vehicles to arrive at the charging station), weather data (rainy days / high temperatures increase charging demand), time period / holiday tags (such as a surge in demand for charging stations in commercial areas during holidays).

[0063] Furthermore, the prediction module in this embodiment can adopt a layered architecture, employing adaptive prediction strategies for different scenarios to balance computational efficiency and accuracy: 1. The base layer is used for statistical time-series prediction. It periodically fits the available number of charging piles at a single charging station, for example, using exponential smoothing to predict charging demand fluctuations over the next 30 minutes based on historical data, outputting a baseline value for the number of available charging piles; 2. The enhancement layer is used for multi-feature machine learning prediction. It integrates real-time operational data with externally correlated data, training a multi-feature prediction model using tree models such as LightGBM, focusing on correcting biases in the base layer (such as demand changes caused by sudden weather or resource fluctuations caused by temporary charging pile failures); 3. The adaptation layer is used for reinforcement learning environment model linkage. It combines the environment model of a multi-agent reinforcement learning framework, taking the allocation result of the current charging request (i.e., the agent's action selection) as input, iteratively updating the predicted value of the future available number of charging piles. For example, when the model recommends charging station A for an agent, it dynamically reduces the predicted value of the available number of charging piles for that station over the next 30 minutes, creating a closed loop between prediction and actual scheduling actions.

[0064] Finally, to address the uncertainties of real-time scenarios, this application embodiment employs a dynamic correction mechanism to adjust the prediction results: 1. Rolling window update: The prediction model is updated every 5 minutes with the latest data such as charging end events and changes in charging pile status to ensure the timeliness of input features. 2. Abnormal event-triggered correction: When abnormal situations such as charging pile malfunctions or users prematurely ending charging occur, local prediction value correction is immediately triggered to avoid recommendation bias caused by static prediction.

[0065] Understandably, this prediction will be incorporated into the observation vector as the future state of the charging station, allowing each electric vehicle agent to not only see the current number of available charging stations but also predict resource changes over the next 30 minutes when making decisions. For example, if a charging station currently has no available charging stations but predicts that two stations will become available in the next 10 minutes, the agent can choose to go to that station, thus avoiding missing out on better options due to misjudgments of the current state. If a charging station is currently available but predicts that it will be fully occupied in the next 20 minutes, the agent will proactively avoid it, thereby reducing queuing caused by blindly selecting charging stations at the source.

[0066] In step S103, based on preset time cost and load rate constraints, charging scheduling results are generated according to the observation vector of each vehicle to be charged.

[0067] Optionally, in some embodiments, based on preset time cost and load rate constraints, a charging scheduling result is generated according to the observation vector of each vehicle to be charged, including: constructing an objective function based on preset time cost and load rate constraints, and generating a charging scheduling result based on the objective function and the observation vector of each vehicle to be charged.

[0068] The preset time cost and load rate constraints can be: ; ; The objective function is: ; in, For vehicle index, i.e., the first vehicle; For the charging station index, i.e., the first One charging station; For the first The car arrived at the The total time cost required to complete queuing and start charging at each charging station; For decision variables; This is the sum of the total time costs for all vehicles. To be allocated to charging stations The total number of vehicles, For charging stations The number of charging stations For charging stations load rate, This represents the total number of charging stations within the target area. The balancing coefficient is a parameter used to adjust the weight between time cost and load balancing. .

[0069] Understandably, this application constructs pre-defined time cost and load rate constraints with the objectives of minimizing the total time cost of all vehicles waiting to be charged (the time cost of each vehicle is the time from its current location to the target charging station plus queuing time) and achieving load balancing across all charging stations. Based on these constraints, an objective function is constructed, and the time cost of each vehicle traveling to any charging station is calculated based on its observation vector. Subsequently, decision variables were introduced. Represent the allocation relationship and impose two types of core constraints: 1. Unique assignment constraint: This ensures that each vehicle is assigned to only one charging station; 2. Capacity feasibility constraints Ensure that the number of vehicles allocated to any charging station does not exceed the total number of charging piles to avoid overload.

[0070] Based on this, an objective function is constructed, which consists of two parts: Part One Minimize the total time cost of all vehicles waiting to be charged, thereby improving user efficiency.

[0071] The second part incentivizes each charging station's load rate through a secondary penalty. Falling within the ideal range If the load rate exceeds 0.8 (overload) or falls below 0.2 (resource idle), a positive penalty will be introduced, and the greater the deviation, the stronger the penalty.

[0072] Finally, under the constraints, the scheduling combination that minimizes the objective function is solved, thus generating a globally optimized charging scheduling result.

[0073] Therefore, the embodiments of this application achieve synergistic optimization of time efficiency and resource balance. On the one hand, minimizing the total time cost directly improves user satisfaction and reduces unnecessary driving and waiting; on the other hand, through the load rate penalty mechanism, it effectively avoids the phenomenon of some charging stations being overcrowded while others are idle for a long time, extending the service life of equipment and improving the overall utilization rate of infrastructure. In addition, the flexible balance coefficient in the objective function... This allows the system to dynamically adjust and optimize its focus based on actual operational needs (such as emphasizing balance during peak hours and speed during off-peak hours).

[0074] As another embodiment of this application, charging scheduling can also be performed based on a multi-agent cooperative approach. Specifically, after obtaining the observation vector in step S102, the observation vector can be input into a preset agent model to obtain the charging scheduling result.

[0075] The pre-set agent model in this application embodiment is trained using the CTDE (Centralized Training with Decentralized Execution) strategy and mean field theory.

[0076] For example, a multi-agent reinforcement learning model is constructed with the goal of minimizing the time cost of all agents and balancing the load of charging stations. The CTDE strategy is adopted: during the training phase, global information (states and actions of all agents) is used to optimize the model to ensure that the strategy is globally optimal; during the execution phase, each agent makes independent decisions based only on its own observation vectors, without relying on global data transmission, which greatly reduces scheduling latency.

[0077] During the training phase, a hierarchical training logic is adopted, from single-agent pre-training to multi-agent collaborative training. This ensures the basic decision-making ability of individual agents while achieving the optimal global objective through collaborative training, relying entirely on the CTDE strategy for implementation.

[0078] (a) Single agent pre-training stage.

[0079] The training objective in the single-agent pre-training phase is for each vehicle to select the optimal charging station based on its own state and basic charging station information, thereby laying the foundation for multi-agent collaboration. The specific steps are as follows: 1. Training environment initialization: Build a single-agent simulation training environment, import the basic data of all charging stations in the target area, such as location, total number of charging piles, historical charging duration distribution, and driving time mapping table under different time periods and road conditions; set the training period to simulate charging scenarios for 1000 natural days, and randomly generate charging requests for different time periods every day.

[0080] 2. Definition of agent state and action: A single agent only inputs its own state (current location, remaining battery power, estimated time to reach each charging station) and the real-time status of the charging stations (number of available charging stations at each station, current queue number). The action space is the set of all available charging stations.

[0081] 3. Independent exploration and reward feedback: An ε-greedy strategy (ε initial value 0.8, linearly decaying to 0.1 with training rounds) is adopted to allow the agent to explore the action space. Under ε probability, the agent randomly selects a charging station, and under 1-ε probability, the agent selects the charging station with the highest estimated reward based on the current state. After each completion of a full charging process (T0-T3), the reward is calculated based on the actual driving time plus queuing time (<60 minutes gives a positive reward, value = 60 - actual time; ≥60 minutes or charging failure gives 0 reward).

[0082] 4. Parameter Iterative Optimization: The single agent uses DQN (Deep Q-Network) as the basic model. The network input is the state feature vector, and the output is the Q value of each charging station (corresponding to the estimated reward for selecting that station). After every 10 charging scenarios (episodes) are completed, the loss function (MSE loss, i.e., the deviation between the predicted Q value and the actual reward) is calculated, and the network parameters are updated by gradient descent until the average time cost of the agent selecting charging stations stabilizes at the target threshold (e.g., within 40 minutes), at which point the pre-training ends.

[0083] (ii) Multi-agent collaborative training phase.

[0084] The training objectives in the multi-agent collaborative training phase are to minimize the overall time cost of all agents and achieve load balancing at the charging station. This involves integrating the CTDE strategy with mean-field theory, and the specific steps are as follows: 1. Collaborative training environment setup: Based on the single-agent environment, it is extended to a multi-agent simulation environment, which supports the simultaneous generation of multiple charging requests within a preset time window Δt (1 minute) (the number of agents m fluctuates randomly, ranging from 5 to 50 per minute, simulating the charging request density changes in real-world scenarios); it also synchronously imports global data interfaces for centralized collection of the status, actions, and environmental feedback data of all agents.

[0085] 2. Intensive training phase: Data aggregation: The centralized trainer collects the local state of all agents, the selected charging station actions, the actual reward value, and the real-time load data of each charging station (changes in the number of available charging piles and the number of people in the queue) in real time to form a global training sample set.

[0086] 3. Mean Field Theory Fusion: For the problem of variable number of agents m, the trainer calculates the average of the actions of all other agents for each agent i (i.e., the average number of times each charging station is selected). This average is used as the "virtual environment state" to supplement the state features of agent i, which approximates the complex interaction between multiple agents and reduces the dimensionality of the state space.

[0087] 4. Global loss calculation: The total time cost of all agents is used as the core loss term, and a load balancing penalty term is added (if the load rate of a charging station is >80% or <20%, an additional loss value is added) to construct a comprehensive loss function, ensuring that the training pursues both time optimization and resource balance.

[0088] Specifically, step (1): Data acquisition, the centralized trainer obtains the actual driving time of all agents (m) in the current collaborative group. Expected queuing time And the current number of occupied charging piles and the total number of charging piles in all charging stations (n) within the target area; Step (2): Calculate the total time cost loss. —Summarize the time costs of all agents as the core loss term; (3) Step 3: Calculate the load rate of each charging station —Calculate the real-time load rate based on the number of occupied charging piles at each charging station and the total number of charging piles; Step (4): Calculate the load balancing penalty. —For charging stations with load rates exceeding 20%-80%, calculate the penalty value; the more unbalanced the load, the greater the penalty value. Step (5): Calculate the comprehensive loss value. — Combining time cost loss and load balancing penalty, the final loss value is obtained after weighting, which serves as the basis for updating model parameters.

[0089] Then, construct the following data: (1) Comprehensive loss function (core formula for training optimization): ; (2) Total time cost loss (Core loss item): ; (3) Charging station load rate (Basis for calculating penalty items): ; (4) Load balancing penalty (Constrained load imbalance): .

[0090] in, The overall loss value is the core optimization objective of model training. The smaller the value, the lower the time cost and the more balanced the load on the charging station. : The number of agents (i.e., the number of charging requests) within the current time window Δt, with a value range of 5-50 (adapting to fluctuations in real-world scenarios). : The estimated travel time (in minutes) for the i-th agent to reach the recommended charging station, obtained from the agent's local observation; : The estimated queuing time (in minutes) after the i-th agent arrives at the charging station, calculated based on the predicted availability of charging piles in the next 30 minutes; Penalty coefficient, empirically set at 0.3~0.5, can be adjusted according to the scenario (if the focus is on load balancing, it can be appropriately increased to 0.4~0.5; if the focus is on time cost, it can be adjusted to 0.3~0.4). The total number of charging stations in the target area is determined based on the actual application scenario; The number of currently occupied charging stations at the j-th charging station. : Total number of charging piles at the j-th charging station (all data collected in real time); Penalty logic: when (Load rate < 20%, pile idle) or When the load rate is greater than 80% and the pile is overloaded, a penalty value is generated. The greater the deviation, the higher the penalty value.

[0091] 5. Parameter synchronization update: The centralized trainer is based on the comprehensive loss function and updates the model parameters of all agents through gradient backpropagation. That is to say, each agent shares the same basic model structure, and only the feature extraction layer parameters of local observation are kept independent. After each day of simulation, the parameters are synchronized to all agents.

[0092] 6. Decentralized Validation Phase: After parameter synchronization, each agent is separated from the centralized trainer and makes independent decisions based solely on its own local observation data, simulating decentralized execution in real-world scenarios. The trainer only collects decision results and environmental feedback without interfering with the decision-making process. If the global total time cost decrease rate is less than 1% for 10 consecutive natural days and the average load rate of the charging station remains stable at 40%-60%, then the collaborative training ends.

[0093] During the execution phase of the pre-defined agent model, each agent operates completely independently, without relying on a centralized trainer. Decisions are made solely based on local data and the pre-trained model. The core objective is to reduce real-time scheduling latency, as defined below: (a) Intelligent agent input.

[0094] The input is an 8-dimensional feature vector, all acquired through local sensors or lightweight data interfaces, requiring no data transfer across agents. Specifically, this includes: 1. Vehicle status (4 dimensions): Current location, current remaining battery power, current driving speed, and charging power demand type for each vehicle to be charged; 2. Trip-related status (1D): Travel time for each vehicle to reach each charging station. 3. Charging station status (3D): Queue size for each charging station, number of currently available charging piles, and number of available charging piles in the predicted time period (next 30 minutes).

[0095] (ii) Intelligent agent output.

[0096] The output is a discrete decision result: from all available charging stations in the target area (excluding faulty and unavailable charging stations), select the charging station with the lowest estimated total time cost (driving + queuing), and output the unique identifier (ID) of the charging station and a suggested route (generated based on the current location).

[0097] After the agent outputs a decision, it only synchronizes the selection result to the charging station management platform to update the station's real-time load, without interacting with other agents to exchange decision information. If the road conditions or charging station status change abruptly during the journey, such as sudden congestion or charging pile failure, the agent will re-execute the above input-output process based on the latest local observation data to dynamically adjust the recommendation result.

[0098] Meanwhile, mean field theory is introduced to adapt to scenarios with a variable number of agents: by averaging the actions of other agents, the expected reward of a single agent is approximately obtained, avoiding model failure due to fluctuations in the number of charging requests within Δt, and improving adaptability to large-scale scenarios.

[0099] In practical use, each electric vehicle that initiates a charging request within a preset time window Δt (default value is 1 minute) is set as an independent intelligent agent. If there are m charging requests within Δt, a collaborative group containing m intelligent agents is formed. This setting enables parallel processing of all intelligent agents in multi-request scenarios, laying the foundation for subsequent synchronous decision-making. Then, multi-dimensional charging environment status is collected to ensure that the intelligent agent's decision-making has sufficient data support: (1) Real-time status of charging stations: collect the current number of charging requests and the number of available charging piles at each charging station to clarify the current resource supply and demand situation; (2) Intelligent agent travel status: collect the estimated travel time of each intelligent agent to each charging station to match the user's travel needs; (3) Future status of charging stations: predict and collect the number of available charging piles at each charging station within 30 minutes after the current time to avoid charging stations with tight resources in the future.

[0100] For agent collaboration groups within Δt, synchronous decision-making is used instead of sequential decision-making: the model simultaneously performs charging station allocation actions for all agents in the collaboration group, rather than allocating them one by one according to the request order. This avoids invalid waiting for agents with later requests, and at the same time, through global objective constraints, it ensures that the charging stations selected by each agent form a resource-balanced distribution, reducing local overload.

[0101] In addition, the preset intelligent agent model in this application embodiment can also be set with a feedback iteration module to iteratively update during actual use. The reward value and model parameters are updated based on the actual charging results. As the usage scenarios accumulate, the recommendation accuracy can be continuously improved. The feedback iteration module can integrate charging data to form a scheduling sample dataset, providing data support for subsequent model fine-tuning and scenario expansion. The status acquisition module supports multiple data interaction methods such as wireless and wired, and can be adapted to most existing charging station management platforms and electric vehicle terminals.

[0102] The charging scheduling method of this application embodiment is verified below with reference to specific embodiments.

[0103] Example scenario setting: A preset time window Δt = 1 minute, within which 3 charging requests are initiated sequentially (corresponding to 3 intelligent agents, denoted as A1, A2, and A3). There are 3 available charging stations in the target area (denoted as S1, S2, and S3). The initial states of each charging station are as follows: S1 (2 available charging piles, no queue), S2 (1 available charging pile, no queue), S3 (1 available charging pile, no queue). The estimated travel time for each intelligent agent to reach each charging station is as follows: A1 to S1 takes 8 minutes, S2 takes 12 minutes, and S3 takes 15 minutes; A2 to S1 takes 10 minutes, S2 takes 6 minutes, and S3 takes 14 minutes; A3 to S1 takes 13 minutes, S2 takes 9 minutes, and S3 takes 7 minutes.

[0104] The charging scheduling scheme in related technologies allocates charging requests sequentially (A1, A2, A3) one by one, with the following decision results: ① A1 prioritizes S1 with the shortest travel time, occupying 1 available charging pile, leaving S1 with 1 available charging pile remaining; ② A2 prioritizes S2 with the shortest travel time, occupying 1 available charging pile, leaving S2 with no available charging pile remaining; ③ A3 prioritizes S3 with the shortest travel time, occupying 1 available charging pile, leaving S3 with no available charging pile remaining. The final load distribution is: S1 (occupying 1 / 2 of the total, load rate 50%), S2 (occupying 1 / 1 of the total, load rate 100%), S3 (occupying 1 / 1 of the total, load rate 100%). The total travel time for the three intelligent agents is 8 + 6 + 7 = 21 minutes. However, S2 and S3 have no available charging piles. If a new charging request arrives within the next minute, it must wait for the preceding vehicle to finish charging, resulting in a significant increase in waiting time.

[0105] The charging scheduling method adopted in this application embodiment: The model includes A1, A2, and A3 in the same cooperative group, and evaluates the total time cost and load balance of all agent-charging station combinations. The final allocation results are as follows: ① A1 is allocated to S1 (driving time 8 minutes), occupying 1 available charging pile; ② A2 is allocated to S1 (driving time 10 minutes), occupying the remaining 1 available charging pile in S1; ③ A3 is allocated to S3 (driving time 7 minutes), occupying 1 available charging pile. This allocation logic satisfies two objectives simultaneously: first, the total driving time is controllable (8+10+7=25 minutes, only 4 minutes longer than the sequential decision); second, the load balance is optimized (S1 load rate 100%, S2 load rate 0%, S3 load rate 100%, no single station overload and S2 is reserved for subsequent requests).

[0106] If a new agent A4 is added in the next minute Δt (which only takes 5 minutes to reach S2), the synchronous decision can directly assign A4 to S2 without waiting for the preceding vehicles in S2 and S3 to finish charging. In related technologies, A4 needs to wait for S2 and S3 to release available charging stations, adding at least 15-30 minutes of waiting time. It is evident that the synchronous decision avoids the ineffective waiting of subsequent requesting agents and achieves resource load balancing through global coordination, effectively solving efficiency and load issues in multi-request scenarios. Therefore, this application embodiment uses T0 (the time of initiating a charging request), T1 (the time of arriving at the charging station), T2 (the time of completing the queue), and T3 (the time of leaving after charging) as key time nodes. The model focuses on optimizing the travel time from T0 to T1 and the queuing time from T1 to T2 to control time costs and improve the overall charging experience.

[0107] In summary, to address the issues of blind user selection of charging stations, load imbalance, high scheduling latency, and long waiting times for multiple requests in existing electric vehicle charging scheduling, this paper proposes a reinforcement learning model that minimizes overall driving queuing time. This model is constructed by treating electric vehicles that initiate charging requests within a preset time window Δt as independent intelligent agents. It combines centralized training with distributed execution strategies, synchronous decision-making, and multi-dimensional charging environment status collection. This model achieves efficient and personalized charging station recommendations, reducing the total time cost of user driving and queuing, and significantly improving charging satisfaction. It also guides intelligent agents to select charging stations in a balanced manner, balancing the load of charging stations in different areas, reducing the risk of local grid overload, and alleviating traffic congestion in charging areas. Furthermore, it reduces global data dependence, enabling scheduling response speeds to adapt to large-scale real-time charging request scenarios. Synchronous decision-making by the intelligent agent collaborative group enables parallel allocation of multiple requests, significantly shortening electric vehicle charging waiting times and improving overall charging efficiency.

[0108] The charging scheduling method proposed in this application obtains the status of at least one charging station and at least one vehicle waiting to be charged within a target area. Based on the status of these two locations, an observation vector for each vehicle is determined. Then, based on preset time cost and load rate constraints, a charging scheduling result is generated using the observation vector of each vehicle. This method solves the problems of overload and traffic congestion at popular charging stations due to users blindly selecting charging stations without dynamic information, as well as scheduling response delays under high concurrency requests. It significantly improves the balance and response efficiency of charging resource allocation, reduces average user waiting time, and enhances overall charging service satisfaction.

[0109] Next, the charging scheduling device proposed according to the embodiments of this application is described with reference to the accompanying drawings.

[0110] Figure 2 This is a block diagram of a charging scheduling device according to an embodiment of this application.

[0111] like Figure 2 As shown, the charging scheduling device 10 includes: an acquisition module 100, an observation module 200, and a scheduling module 300.

[0112] The acquisition module 100 is used to acquire the status of at least one charging station and the status of at least one vehicle to be charged within the target area.

[0113] The observation module 200 is used to determine the observation vector for each vehicle to be charged based on the status of at least one charging station and the status of at least one vehicle to be charged.

[0114] The scheduling module 300 is used to generate charging scheduling results based on the observation vector of each vehicle to be charged, according to preset time cost and load rate constraints.

[0115] Optionally, in some embodiments, the observation module 200 includes: a first determining unit, a second determining unit, and a first generating unit.

[0116] The first determining unit is used to determine the number of queues, the number of currently available charging piles, and the number of available charging piles in the predicted time period for each charging station based on the status of at least one charging station.

[0117] The second determining unit is used to determine the current location, current remaining battery power, current driving speed, charging power demand type, and driving time of each vehicle to be charged to each charging station based on the status of at least one vehicle to be charged.

[0118] The first generation unit is used to obtain the observation vector of each vehicle to be charged based on the number of queues at each charging station, the number of currently available charging piles and the number of available charging piles in the predicted time period, as well as the current location, current remaining power, current driving speed, charging power demand type and driving time of each vehicle to be charged to each charging station.

[0119] Optionally, in some embodiments, the scheduling module 300 includes a second generation unit.

[0120] The second generation unit is used to construct an objective function based on preset time cost and load rate constraints, and generate charging scheduling results based on the objective function and the observation vector of each vehicle to be charged.

[0121] Optionally, in some embodiments, before acquiring the status of at least one charging station and at least one vehicle to be charged within the target area, the acquisition module 100 further includes: a first acquisition unit, a second acquisition unit, a third determination unit, and a third generation unit.

[0122] The first acquisition unit is used to acquire the number of charging requests from all vehicles waiting to be charged within the target area in the same time window, and to determine whether the number of charging requests is less than a preset threshold.

[0123] The second acquisition unit is used to acquire the current location of each vehicle to be charged and the current location of each charging station when the number of charging requests is less than a preset threshold.

[0124] The third determining unit is used to determine at least one candidate charging station corresponding to each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station.

[0125] The third generation unit is used to determine the score of at least one candidate charging station corresponding to each vehicle to be charged, obtain the score set of all candidate charging stations corresponding to each vehicle to be charged, select the highest score corresponding to each vehicle to be charged from the score set of all candidate charging stations corresponding to each vehicle to be charged, and determine the charging scheduling result of each vehicle to be charged based on the candidate parking station corresponding to the highest score of each vehicle to be charged.

[0126] Optionally, in some embodiments, the third generation unit includes: a determining subunit and a calculating subunit.

[0127] The determination subunit is used to determine the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging piles at at least one candidate charging station for each vehicle, and the number of queues.

[0128] The calculation subunit is used to calculate the score of at least one candidate charging station corresponding to each vehicle based on the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging piles at at least one candidate charging station for each vehicle, and the number of queues.

[0129] Optionally, in some embodiments, determining a subunit includes: determining a sub-component and determining a sub-component.

[0130] The judgment sub-component is used to determine whether there is at least one charging station within a preset range for each vehicle to be charged, based on the current location of each vehicle to be charged and the current location of each charging station.

[0131] The determination sub-component is used to determine at least one candidate charging station for each vehicle to be charged, based on the charging stations within the preset range of each vehicle to be charged, when at least one charging station exists within the preset range of each vehicle to be charged.

[0132] Optionally, in some embodiments, after determining whether there is at least one charging station within a preset range for each vehicle to be charged, the determination sub-component includes: updating the sub-component.

[0133] The update sub-component is used to update the preset range of any vehicle with a preset step size when there is no charging station within the preset range of any vehicle, and to re-determine whether there is at least one charging station within the preset range of any vehicle, until there is at least one charging station within the preset range of any vehicle.

[0134] It should be noted that the foregoing explanation of the charging scheduling method embodiment also applies to the charging scheduling device of this embodiment, and will not be repeated here.

[0135] The charging scheduling device proposed in this application obtains the status of at least one charging station and at least one vehicle waiting to be charged within a target area. Based on the status of the charging station and the vehicle, it determines the observation vector for each vehicle. Then, based on preset time cost and load rate constraints, it generates a charging scheduling result for each vehicle. Therefore, this application solves the problems of overload and traffic congestion at popular charging stations caused by users blindly selecting charging stations due to a lack of dynamic information, as well as scheduling response delays under high concurrency requests. It can significantly improve the balance and response efficiency of charging resource allocation, reduce average user waiting time, and improve overall charging service satisfaction.

[0136] Figure 3 A schematic diagram of the structure of a cloud server provided in an embodiment of this application. The cloud server may include: The memory 301, the processor 302, and the computer program stored on the memory 301 and capable of running on the processor 302.

[0137] When the processor 302 executes the program, it implements the charging scheduling method provided in the above embodiments.

[0138] Furthermore, cloud servers also include: Communication interface 303 is used for communication between memory 301 and processor 302.

[0139] The memory 301 is used to store computer programs that can run on the processor 302.

[0140] The memory 301 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0141] If the memory 301, processor 302, and communication interface 303 are implemented independently, then the communication interface 303, memory 301, and processor 302 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0142] Optionally, in a specific implementation, if the memory 301, processor 302, and communication interface 303 are integrated on a single chip, then the memory 301, processor 302, and communication interface 303 can communicate with each other through an internal interface.

[0143] Processor 302 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0144] Furthermore, embodiments of this application also protect an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to perform the charging scheduling method provided in embodiments of this application.

[0145] This embodiment can divide the device into functional modules based on the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0146] When the functions of each module are divided according to their respective functions, the device may also include a draining module, an acquisition module, and a determination module. It should be noted that all relevant content regarding the steps involved in the above method embodiments can be referenced in the corresponding functional descriptions, and will not be repeated here.

[0147] It should be understood that the device provided in this embodiment is used to execute the above-described charging scheduling method, and therefore can achieve the same effect as the above-described implementation method.

[0148] When using integrated units, the device may include a processing module and a storage module. When applied to an automobile, the processing module can be used to control and manage the vehicle's movements. The storage module can be used to support the vehicle in executing program code, etc.

[0149] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits as disclosed in this application. The processor may also be a combination of computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module may be a memory.

[0150] In addition, the device provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute the charging scheduling method provided in the above embodiments.

[0151] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement a charging scheduling method provided in the above embodiment.

[0152] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to implement a charging scheduling method provided in the above embodiment.

[0153] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0154] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0155] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0156] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A charging scheduling method, characterized in that, Includes the following steps: Acquire the status of at least one charging station and at least one vehicle to be charged within the target area; The observation vector for each vehicle to be charged is determined based on the state of the at least one charging station and the state of the at least one vehicle to be charged. Based on preset time cost and load rate constraints, charging scheduling results are generated according to the observation vector of each vehicle to be charged.

2. The method according to claim 1, characterized in that, Determining the observation vector for each vehicle to be charged based on the status of the at least one charging station and the status of the at least one vehicle to be charged includes: The number of queues, the number of currently available charging piles, and the number of available charging piles in the predicted time period are determined based on the status of the at least one charging station. Based on the status of the at least one vehicle to be charged, determine the current location, current remaining battery power, current driving speed, charging power demand type, and driving time of each vehicle to reach each charging station. The observation vector for each vehicle to be charged is obtained based on the number of queues at each charging station, the number of currently available charging piles, the number of available charging piles within the predicted time period, the current location, current remaining battery power, current driving speed, charging power demand type, and the driving time of each vehicle to be charged to each charging station.

3. The method according to claim 1, characterized in that, The process of generating charging scheduling results based on the observation vector of each vehicle to be charged, according to preset time cost and load rate constraints, includes: Based on the preset time cost and load rate constraints, an objective function is constructed, and based on the objective function, a charging scheduling result is generated according to the observation vector of each vehicle to be charged.

4. The method according to claim 1, characterized in that, Before obtaining the status of at least one charging station and at least one vehicle to be charged within the target area, the process also includes: Get the number of charging requests from all vehicles waiting to be charged within the target area in the same time window, and determine whether the number of charging requests is less than a preset threshold. If the number of charging requests is less than the preset threshold, then the current location of each vehicle to be charged and the current location of each charging station are obtained. Based on the current location of each vehicle to be charged and the current location of each charging station, at least one candidate charging station is determined for each vehicle to be charged. The scores of at least one candidate charging station corresponding to each vehicle to be charged are determined to obtain a score set of all candidate charging stations corresponding to each vehicle to be charged. The highest score corresponding to each vehicle to be charged is selected from the score set of all candidate charging stations corresponding to each vehicle to be charged. The charging scheduling result of each vehicle to be charged is determined according to the candidate parking station corresponding to the highest score of each vehicle to be charged.

5. The method according to claim 4, characterized in that, The step of determining the score for at least one candidate charging station corresponding to each vehicle to be charged includes: Determine the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging stations and the queue number for each vehicle's at least one candidate charging station; The score of each vehicle is calculated by weighting the distance between each vehicle and at least one corresponding candidate charging station, the number of currently available charging piles at each candidate charging station, and the number of queues.

6. The method according to claim 4, characterized in that, The step of determining at least one candidate charging station corresponding to each vehicle to be charged based on the current location of each vehicle to be charged and the current location of each charging station includes: Based on the current location of each vehicle to be charged and the current location of each charging station, determine whether there is at least one charging station within a preset range of each vehicle to be charged; If at least one charging station exists within a preset range for each vehicle to be charged, then at least one candidate charging station corresponding to each vehicle to be charged is determined based on the charging stations within the preset range for each vehicle to be charged.

7. The method according to claim 6, characterized in that, After determining whether at least one charging station exists within a preset range for each vehicle to be charged, the process includes: If there is no charging station within the preset range of any vehicle, the preset range of any vehicle is updated by a preset step size, and it is re-determined whether there is at least one charging station within the preset range of any vehicle, until at least one charging station exists within the preset range of any vehicle.

8. A charging scheduling device, characterized in that, include: The acquisition module is used to acquire the status of at least one charging station and the status of at least one vehicle to be charged within the target area; A determination module is used to determine the observation vector of each vehicle to be charged based on the state of the at least one charging station and the state of the at least one vehicle to be charged. The scheduling module is used to generate charging scheduling results based on the observation vector of each vehicle to be charged, according to preset time cost and load rate constraints.

9. A cloud server, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the charging scheduling method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the charging scheduling method as described in any one of claims 1 to 7.

Images