Parking lot charging pile management method and system

Abstract

The embodiment of the invention relates to the field of parking lot management, and provides a parking lot charging pile management method and system, and the method comprises the steps: carrying out the management of charging piles based on the vehicle information and charging demand information submitted by a plurality of parking lot users, the real-time state information and parameter information of the charging piles in a plurality of parking lots, and the charging payment rules of corresponding regions; screening the candidate charging piles; predicting the estimated time when the multi-ground vehicle arrives at each target parking lot, and performing priority ranking on the target charging piles in combination with the idle time periods of the target charging piles and the passing efficiency of the parking lots; a reservation request submitted by a user for the sorted target charging piles is received, a reservation instruction is sent to a terminal corresponding to the target charging piles, and the use time period of the target charging piles is locked; and when the multi-ground vehicle arrives at the target parking lot and charging is started, fee settlement is carried out based on the charging payment rule and the charging time period information of the target area. Cross-site charging pile resource matching, reservation locking and dynamic scheduling are realized, and the charging pile utilization rate and the parking lot collaborative management efficiency are improved.

Description

Technical Field

[0001] This application relates to the field of parking lot management, and more specifically to a parking lot charging pile management method and system. Background Technology

[0002] With the widespread adoption of new energy vehicles, the number of multi-regional vehicles (such as vehicles with license plates from one or more regions within the Guangdong-Hong Kong-Macao Greater Bay Area) is increasing year by year. These vehicles need to frequently park and use charging stations in multiple parking lots across the Greater Bay Area, and the charging fees and payment methods for multi-regional vehicles vary in different regions. As a core infrastructure for replenishing energy for new energy vehicles, the management efficiency of parking lot charging stations directly affects the user experience of multi-regional vehicles and the development of the new energy vehicle industry.

[0003] Existing technologies often limit the management of charging piles to the independent control scope of a single parking lot or a single operating entity. They only carry out reservation, billing, status monitoring and operation and maintenance management for charging piles within the site. They do not take into account the actual scenario of vehicles from multiple regions (vehicles with license plates from multiple regions in the Guangdong-Hong Kong-Macao Greater Bay Area) frequently using charging piles across regions and sites, and the different charging and payment methods in different regions. Furthermore, they do not consider the data interoperability and resource coordination needs of charging pile management systems in different parking lots, nor do they take into account the differentiated adaptation of charging and payment standards in different regions.

[0004] However, due to the dispersed driving trajectories of vehicles from multiple locations, the cross-regional usage scenarios, the independent management systems of charging piles in different parking lots, the inability to share data, the lack of unified user authentication standards, and the inability to coordinate and schedule charging resources across different sites, coupled with the inconsistent charging and payment methods for vehicles from multiple locations in different regions, existing technologies have failed to achieve precise matching between the charging needs of vehicles from multiple locations and the charging pile resources in multiple locations. They have also failed to achieve compatibility and adaptation between charging and payment methods in different regions. As a result, existing technologies suffer from poor user convenience, low overall utilization rate of charging pile resources, cumbersome reservation and billing settlement for cross-regional charging (especially affected by different charging and payment methods, making the settlement process complex), and are prone to resource waste such as vehicles from multiple locations not being able to find suitable charging piles and idle charging piles not being shared. Furthermore, there are problems such as difficulty in unified management of charging records for vehicles from multiple locations, untimely operation and maintenance response, and incompatibility of payment settlement in different regions. These issues fail to meet the efficient and convenient charging needs of vehicles from multiple locations and also restrict the efficient utilization of parking lot charging pile resources across the entire area. Summary of the Invention

[0005] This application provides a parking lot charging pile management method and system, which can realize intelligent matching, reservation locking and dynamic scheduling of charging pile resources across sites, reduce user charging waiting time and overall usage costs, improve the utilization rate of charging piles and the efficiency of parking lot collaborative management, and reduce resource waste caused by charging resource vacancy and reservation conflicts.

[0006] In a first aspect, embodiments of this application provide a parking lot charging pile management method, the method comprising: It obtains vehicle information and charging demand information submitted by car users in multiple locations, and obtains real-time status information, parameter information, and corresponding regional charging and payment rules for charging piles in multiple parking lots. Based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and the charging and payment rules of the corresponding area, a cross-site charging pile resource matching model is constructed with the optimization objectives of charging pile resource utilization rate, user charging cost and charging waiting time. Candidate charging piles are screened to obtain at least one target charging pile. Based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and the reservation lock duration of charging stations, the estimated time for vehicles in multiple locations to arrive at each target parking lot is predicted. Then, the estimated time, the idle time of the target charging station, and the traffic efficiency of the parking lot are combined to prioritize at least one target charging station. The system receives reservation requests from users for sorted target charging stations and sends reservation instructions to the terminal corresponding to the target charging station to lock in the usage time of the target charging station. After vehicles from multiple locations arrive at the target parking lot and start charging, the charging process data is acquired, and the cost is calculated based on the charging payment rules and charging time information of the target area. After charging is completed, the system will process the settlement based on the user's linked payment method and the fee calculation result, generate the corresponding charging record, and update the status information of the target charging station.

[0007] Secondly, embodiments of this application provide a parking lot charging pile management system, which has functions corresponding to the parking lot charging pile management method provided in the first aspect above. These functions can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions, and these modules can be software and / or hardware.

[0008] In one embodiment, the parking lot charging pile management system includes: The input / output module is configured to obtain vehicle information and charging demand information submitted by users in multiple locations, and to obtain real-time status information, parameter information, and corresponding payment rules of charging piles in multiple parking lots. The processing module is configured to: construct a cross-site charging pile resource matching model based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and the corresponding area's payment rules; optimize charging pile resource utilization, user charging costs, and charging waiting time; filter candidate charging piles to obtain at least one target charging pile; construct a vehicle arrival time prediction model based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and charging pile reservation lock duration; predict the estimated arrival time of vehicles at each target parking lot; and prioritize at least one target charging pile by combining the idle time of the target charging pile and the parking lot's traffic efficiency; receive reservation requests submitted by users for the ranked target charging piles; send reservation instructions to the corresponding terminals of the target charging piles to lock the usage time of the target charging piles; after vehicles in multiple locations arrive at the target parking lot and start charging, acquire charging process data during the charging process and calculate the cost based on the payment rules of the target area and the charging time information; after charging is completed, perform settlement processing based on the user's bound payment method and the cost calculation result, generate the corresponding charging record, and update the status information of the target charging pile.

[0009] Thirdly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the parking lot charging pile management method as described in the first aspect.

[0010] Fourthly, embodiments of this application provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the parking lot charging pile management method described in the first aspect.

[0011] Fifthly, embodiments of this application provide a computer program product containing instructions, which, when run on a computer, causes the computer to execute the parking lot charging pile management method provided in the first aspect.

[0012] Compared to existing technologies, this application embodiment integrates the real-time status of charging piles across different locations, regional payment rules, and user charging needs into a unified model. It then combines vehicle arrival time prediction with target charging pile idle time matching to form an automated management process encompassing vehicle-charging pile resource matching, vehicle arrival prediction, charging pile reservation locking, and intelligent charging billing settlement. Because this application embodiment is based on a cross-location multi-objective optimization resource matching model and introduces arrival time prediction-driven reservation scheduling, rather than the charging pile selection and reservation based on single-location static information or manual rules in existing technologies, it can achieve more accurate target charging pile recommendation and reservation locking in scenarios involving frequent parking across multiple regions and parking lots, reducing vehicle charging waiting time and the probability of reservation conflicts. Therefore, this application embodiment achieves consistency and executability of cross-regional billing settlement by incorporating regional payment rules and charging time information into the cost calculation process. Since the status of the target charging pile is updated in a timely manner after settlement and a charging record is archived, the charging pile resource allocation and reservation scheduling results obtained in this application embodiment can effectively help improve resource utilization and reduce overall user costs, thereby improving the convenience of multi-location vehicle charging and the overall service experience. Attached Figure Description

[0013] Figure 1 This is a system schematic diagram of the parking lot charging pile management method in the embodiments of this application; Figure 2 This is a flowchart illustrating the parking lot charging pile management method according to an embodiment of this application; Figure 3 This is a schematic diagram of the parking lot charging pile management system according to an embodiment of this application.

[0014] In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts. Detailed Implementation

[0015] This application provides a parking lot charging pile management method and system, applicable to charging pile management systems that support cross-regional charging management of vehicles from multiple locations and collaborative charging across multiple parking lots. Exemplarily, the charging pile management system may include a resource matching device and a scheduling and settlement device, which can be deployed integratedly or separately. The resource matching device is used at least to filter candidate charging piles based on vehicle information, charging demand information, real-time charging pile status information, parameter information, and regional charging and payment rules to obtain target charging piles. The scheduling and settlement device is used to prioritize and reserve charging piles based on the vehicle's estimated arrival time, the target charging pile's idle time, and the parking lot's traffic efficiency, and to complete billing, settlement, and status updates after charging is completed.

[0016] The resource matching device can be an application that executes a cross-site charging pile resource matching model, or a server or terminal device with the application installed. The scheduling and settlement device can be a management program that executes arrival time prediction, reservation scheduling, cost calculation, and settlement processing. The management program may include, for example, a time-series prediction model and payment rule adaptation module. The scheduling and settlement device can also be a terminal device or cloud server that has deployed the model and module.

[0017] The solutions provided in this application address the technical problems existing in the current charging settlement and parking fee calculation processes for multi-location vehicles, as illustrated in the following embodiments: A multi-region vehicle refers to a vehicle with license plates from multiple regions or with the right to travel between multiple regions. For example, a multi-region vehicle can be a vehicle with license plates from one or more regions within the Guangdong-Hong Kong-Macao Greater Bay Area, and can park in multiple parking lots across regions within the Greater Bay Area and use charging stations.

[0018] In existing technologies, parking lot charging pile management typically adopts an independent management approach by a single parking lot or a single operating entity. This means that charging pile reservations, status monitoring, fee calculations, and payment settlements are only completed within their respective site areas. The reason for adopting this approach is that the system construction entities, data standards, authentication rules, and settlement interfaces of different parking lots are usually independent of each other. The cost of achieving cross-system interconnection and collaborative transformation is high. Therefore, existing technologies are mostly based on localized and decentralized management methods.

[0019] However, this approach is not designed for scenarios where vehicles from multiple locations frequently park and use charging stations across different areas and venues. It fails to achieve unified management and coordinated scheduling of charging station status data, user information, and payment rules across different parking lots. This results in difficulties in accurately matching the charging needs of vehicles from multiple locations with the charging station resources at multiple venues. Furthermore, it leads to incompatibility and compatibility issues between different regional charging methods and payment methods. Consequently, problems such as cumbersome cross-regional charging reservations, complex fee settlements, low efficiency in finding charging stations for users, a coexistence of idle and scarce charging station resources, difficulty in unified management of charging records, and untimely operation and maintenance responses arise.

[0020] Compared to existing technologies, this application embodiment acquires vehicle information and charging demand information submitted by users in multiple locations, and combines this with real-time status information, parameter information, and corresponding regional payment rules of charging piles in multiple parking lots to construct a cross-site charging pile resource matching model. This model filters candidate charging piles to obtain target charging piles. Simultaneously, based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and charging pile reservation lock duration, the estimated arrival time of vehicles in each target parking lot is predicted. Priority is then assigned and reservations are made based on the idle time slots of the target charging piles and the parking lot's traffic efficiency. After charging is completed, the cost is calculated, payment is settled, and status is updated according to the rules of the target area.

[0021] Through the above approach, the embodiments of this application realize the overall scheduling of charging resources in multiple locations and the compatibility and adaptation of charging and payment methods in different regions. It effectively solves the problems in the prior art such as the inability to share charging pile resources across locations, difficulty in finding charging piles for vehicles in multiple locations, numerous reservation conflicts, complex settlement links, and difficulty in unified management of charging records. This improves the utilization rate of charging pile resources, the efficiency of cross-regional charging management, and the charging convenience for users of vehicles in multiple locations.

[0022] In some implementations, the management platform and parking lot terminals are deployed separately, referring to... Figure 1 The parking lot charging pile management method provided in this application embodiment can be based on Figure 1 The diagram illustrates a parking lot charging station management system. This system may include a server 01 and terminal devices 02.

[0023] The server 01 can be a parking lot charging pile management platform, which can deploy cross-site resource matching programs, vehicle arrival time prediction programs, reservation and scheduling programs, and billing and settlement programs.

[0024] The terminal device 02 can be a parking lot charging station terminal and / or a user terminal, in which terminal interaction programs, data acquisition programs and payment processing programs can be deployed.

[0025] The terminal device 02 can forward received vehicle information, charging demand information, charging pile status information, and reservation requests to the server 01. The server 01 can filter target charging piles based on the vehicle information, charging demand information, charging pile status information, and regional payment rules. It then combines the estimated arrival times of vehicles from multiple locations, the idle time of the target charging piles, and the parking lot's traffic efficiency to generate a priority ranking result and a reservation lock instruction, which is then sent to the terminal device 02. The terminal device 02 can also collect charging process data during charging and report it to the server 01. The server 01 can perform cost calculation and settlement processing based on the charging process data and target area rules, update the target charging pile status, and generate corresponding charging records.

[0026] It should be noted that the computing devices involved in the embodiments of this application may be servers and / or terminal devices.

[0027] The server involved in the embodiments of this application can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0028] The terminal devices involved in the embodiments of this application can be devices that provide voice and / or data connectivity to users, handheld devices with wireless connectivity, or other processing devices connected to a wireless modem. Examples include mobile phones (or "cellular" phones) and computers with mobile terminals, such as portable, pocket-sized, handheld, computer-embedded, or vehicle-mounted mobile devices that exchange voice and / or data with a wireless access network. Examples include Personal Communication Service (PCS) phones, cordless phones, Session Initiation Protocol (SIP) phones, Wireless Local Loop (WLL) stations, Personal Digital Assistants (PDAs), and other devices.

[0029] Reference Figure 2 , Figure 2 This is a flowchart illustrating a parking lot charging pile management method provided in an embodiment of this application. The method can be executed by a parking lot charging pile management system. The method includes the following steps: Step 101: Obtain vehicle information and charging demand information submitted by users in multiple locations, and obtain real-time status information, parameter information, and corresponding payment rules for charging piles in multiple parking lots.

[0030] In step 101, the vehicle information submitted by users of vehicles from multiple locations may include information on the region corresponding to the vehicle's license plate, vehicle type, vehicle battery capacity, remaining battery power, vehicle charging interface type, and user identity association information.

[0031] In one embodiment, the vehicle license plate area information can be used to identify the area to which the vehicle belongs and match the corresponding area's charging rules; the vehicle type information can be used to determine the compatibility between the vehicle model and the charging pile; the vehicle battery capacity information and the vehicle's remaining power information can be used to estimate the current energy replenishment needs and expected charging time; the vehicle charging interface type information can be used to filter available candidate charging piles; and the user identity association information can be used for subsequent identity verification, reservation locking, and settlement processing.

[0032] For example, when a user submits information about a vehicle's regional license plate, battery capacity of 75 kWh, current remaining charge of 30%, and interface type of DC fast charging, the management platform can initially exclude charging piles with incompatible interfaces or mismatched power, and provide input conditions for subsequent selection of target charging piles.

[0033] In one specific embodiment, the information acquisition in step 101 is completed in three stages: user terminal data collection, vehicle-side data synchronization, and platform-side verification. Upon first use, the user completes account registration and vehicle binding on their terminal. The user terminal receives the vehicle license plate information, vehicle type, common charging interface type, and payment account information entered by the user, and submits the results to the management platform. The management platform calls the license plate recognition and license plate database comparison service to identify and verify the area corresponding to the vehicle license plate, obtaining the vehicle license plate area information and user identity association information. For data that passes verification, the management platform generates a unique vehicle identifier and writes it into the user information sub-database for subsequent reservation and settlement processes.

[0034] In one specific embodiment, vehicle battery capacity information and remaining battery power information are obtained in two ways. The first method involves establishing a data authorization connection between the user terminal and the vehicle's infotainment system, reading basic vehicle parameters and real-time battery power after user authorization. The second method involves the user manually entering the battery capacity and using charging history records to retrospectively calculate and correct the remaining battery power range. The management platform can perform consistency comparisons between the two data sources. When the deviation between the real-time battery power read by the infotainment system and the historical charging curve exceeds a preset threshold, the platform triggers a secondary confirmation to improve the reliability of the input data. After confirmation, the platform uses the battery capacity and remaining battery power as the core inputs for calculating charging demand.

[0035] In one specific embodiment, the vehicle charging interface type information is determined jointly through vehicle model database matching and charging pile trial handshake. The management platform first retrieves the default interface type from the vehicle model parameter database based on the vehicle brand, model, and year. Then, after the user arrives at a candidate parking lot, the charging pile terminal initiates an interface handshake test to confirm whether the physical interface matches the communication protocol. If the vehicle model database result matches the handshake result, the interface type is confirmed as a valid interface type. If they do not match, the handshake test result is used, and the vehicle model parameter database record is updated, thereby improving the accuracy of subsequent screening.

[0036] For example, a user binds a vehicle with a regional license plate to their user terminal, enters the vehicle type as a pure electric sedan, and authorizes the vehicle's data access system. The platform identifies that the vehicle's battery capacity is 75 kWh, the real-time remaining charge is 30%, and the interface type is a DC fast charging interface. Subsequently, the platform synchronizes candidate charging pile parameters from multiple parking lots and finds that some piles only support AC slow charging or have a rated power lower than the user's needs. Based on the interface type and power conditions, the platform first eliminates incompatible devices, and then submits the selected charging piles to the subsequent resource matching model for sorting. Through the above process, step 101 can form a structured, verifiable set of vehicle information and demand information that can be directly used for decision-making calculations before reservation.

[0037] In step 101, the charging demand information submitted by users of vehicles in multiple locations may include target charging area information, target parking lot range information, planned charging time information, expected charging power information, target battery capacity information, acceptable waiting time information, and charging budget information.

[0038] In one embodiment, the target charging area information and target parking lot range information are used to limit the candidate site set, the planned charging time information is used to match the charging pile's idle time period, the expected charging power information is used to match the charging pile's output capacity, the target power information is used to determine the scale of this charging task, the acceptable waiting time information is used for user experience constraints in the ranking stage, and the charging budget information is used to constrain the final recommendation result.

[0039] For example, when a user's need is to replenish energy from 30% to 80% in a designated area within a certain time period, and the acceptable waiting time does not exceed 20 minutes and the charging budget does not exceed the preset amount, the management platform can prioritize recommending target charging piles that can meet the interface adaptation, power requirements and cost constraints within that time period.

[0040] In one specific embodiment, the charging demand information in step 101 is collected by the user terminal, and the management platform completes structured parsing and consistency verification. After the user enters the reservation page, the user terminal first obtains the user's current location, and the user selects the target charging area and the target parking lot range. The target parking lot range can be determined by the user manually selecting parking lots from the list, or it can be automatically generated by the user selecting a preset radius range centered on the current location. The user terminal uploads the selected area code and parking lot set identifier to the management platform, and the management platform limits the candidate site set accordingly.

[0041] In one specific embodiment, planned charging time information is obtained through a time period selection control, allowing users to input their planned arrival and expected departure times. The management platform calculates the overlap between this time window and the idle periods of charging piles in each parking lot, eliminating candidate charging piles that do not meet the time constraints. For cross-day reservation scenarios, the platform maps the planned charging time to the corresponding regional time-of-use electricity price range, providing time period input for subsequent cost prediction.

[0042] In one specific embodiment, the desired charging power information and target battery level information are obtained through a combination of explicit user input and automatic system derivation. Users can directly select fast or slow charging levels on the terminal, or input a target battery percentage. The management platform calculates the charging capacity based on the vehicle's battery capacity and current remaining battery level, and matches the desired charging power with the rated output capacity of the candidate charging stations. If the user-input target battery level is lower than the current battery level or exceeds the charging limit, the platform triggers an error correction prompt and guides the user to adjust the parameters.

[0043] In one specific embodiment, the acceptable waiting time and charging budget information are filled in by the user in the preference settings interface. The acceptable waiting time is used to constrain the waiting cost item in the ranking model, and the charging budget is used to constrain the total cost limit. When generating candidate results, the management platform first calculates the expected queuing time and expected total cost of each candidate solution, then filters out solutions that exceed the waiting threshold or budget threshold, retaining the executable candidate set.

[0044] For example, a user submits the following requirements: The target charging area is Xiangzhou District, Zhuhai; the target parking lot is within 5 kilometers of the current location; the planned charging time is from 6 PM to 8 PM; the desired charging power is no less than 60 kilowatts; the target battery level is to increase from 30% to 80%; the acceptable waiting time is no more than 15 minutes; and the charging budget is no more than 80 yuan. After receiving this requirement, the management platform first filters candidate parking lots by region and range, then matches available charging piles by time window, subsequently verifies accessibility based on power capacity and target energy replenishment, and finally outputs a recommended solution based on waiting time and budget constraints. Through this process, the user's subjective needs are transformed into calculable, filterable, and sortable structured constraints.

[0045] In step 101, the real-time status information of charging piles in multiple parking lots may include the idle status, occupied status, fault status, remaining charging time, current queue length, and reservation lock status of each charging pile.

[0046] In one embodiment, the idle state and occupied state are used to determine whether the charging pile is available for immediate use, the fault state is used to remove unavailable equipment, the remaining charging time and the current queue length are used to estimate the waiting cost, and the reservation lock state is used to avoid duplicate allocation of locked resources.

[0047] For example, when a charging station in a parking lot is idle and not reserved, it can be directly added to the available charging set. When another charging station is occupied and has a long remaining charging time, the management platform can lower its priority.

[0048] In step 101, the parameter information of the charging piles in multiple parking lots may include charging power level information, charging interface type information, rated voltage range information, rated current range information, compatible vehicle model information, and equipment health status information.

[0049] In one embodiment, charging power level information is used to match the user's expected charging efficiency, charging interface type information is used to complete physical access adaptation, rated voltage range information and rated current range information are used to ensure charging safety boundaries, compatible vehicle model information is used to improve screening accuracy, and device health status information is used to reduce the risk of failure.

[0050] For example, when a user's vehicle only supports a certain type of fast charging interface and expects higher power charging, the management platform can prioritize charging piles with compatible interfaces and power that meet the threshold requirements, and remove devices with abnormal health status from the candidate set.

[0051] In one specific embodiment, the acquisition of parameter information for multiple parking lot charging piles in step 101 can be completed collaboratively by three parts: static parameter archiving, real-time operation reporting, and session handshake verification. The management platform first establishes a communication connection with the charging pile terminals in each parking lot, and during the device access phase, reads the device's factory parameters and maintenance log parameters to obtain charging power level information, charging interface type information, rated voltage range information, rated current range information, and compatible vehicle model information. This information is written as basic parameters into the charging pile parameter database and bound to the unique identifier of the charging pile for subsequent rapid retrieval and filtering.

[0052] In one specific embodiment, the charging pile terminal reports operating parameters and health parameters to the management platform according to a preset cycle. Operating parameters include the current maximum output power, real-time voltage range, real-time current range, and interface availability status. Health parameters include module temperature rise status, insulation detection status, communication link stability status, fault code records, and recent maintenance records. The management platform generates equipment health status information based on the reported data and classifies the equipment health status into available, limited available, and unavailable levels. When a charging pile has an over-temperature alarm, insulation abnormality, or critical fault code, the system marks it as unavailable and excludes it from the candidate set.

[0053] In one specific embodiment, to improve parameter accuracy, the management platform performs a session handshake verification before and after a user initiates a reservation or arrives at the charging station. The charging station terminal and the vehicle side perform an interface handshake and capability negotiation to confirm the interface physical type, protocol version, acceptable voltage and current range, and maximum charging power. If the handshake result matches the parameter database record, the original parameter state is maintained. If an inconsistency occurs, the management platform updates the corresponding parameters based on the handshake result and records the parameter change time and source to avoid incorrect recommendations due to outdated parameters.

[0054] For example, a user's vehicle only supports DC fast charging and expects a charging power of at least 60 kilowatts. After the management platform synchronizes the parameters of 20 candidate charging piles from three parking lots, it first eliminates AC interface devices based on interface type, then eliminates devices below the threshold based on rated power and real-time output power, and finally performs safety boundary matching based on the vehicle's negotiation capabilities according to the rated voltage range and rated current range. For two devices with insulation abnormality alarms, the system marks their health level as unavailable and removes them from the candidate set. The devices that are ultimately retained simultaneously meet the conditions of interface compatibility, power compliance, safety boundary attainment, and health status availability, and are used as input objects for subsequent resource matching and priority ranking.

[0055] In another embodiment, compatible vehicle information can also be incrementally updated through historical charging sessions. The management platform statistically analyzes the historical handshake success rate, abnormal interruption rate, and average charging efficiency of the same vehicle model on different charging piles to form a vehicle compatibility profile. When a new round of screening begins, the system prioritizes devices with higher historical compatibility, while meeting hardware parameter constraints, thereby further improving recommendation accuracy and charging success rate.

[0056] In step 101, the payment rules for the corresponding region may include time-of-use electricity pricing rules, service fee rules, parking fee rules, cross-regional billing conversion rules, available payment channel rules, and payment authentication rules.

[0057] In one embodiment, the time-of-use pricing rule is used to determine the unit electricity price during peak, normal, and off-peak hours; the service fee rule is used to calculate additional fees; the parking billing rule is used to generate parking fees; the cross-regional billing conversion rule is used to handle differences in billing standards across different regions; the available payment channel rule is used to limit the available payment methods; and the payment authentication rule is used to verify the compliance of payment accounts.

[0058] For example, when the target area uses higher electricity prices during peak hours and only supports some payment channels, the management platform can consider the cost impact and payment accessibility simultaneously during the screening and sorting stages to avoid recommending target charging stations to users that can charge but cannot be settled.

[0059] Taking vehicles with license plates from different regions as an example, in one embodiment, the vehicle license plate region is used as one of the dimensions for matching the billing rules. When vehicles with license plates from different regions are charged in the same parking lot, in addition to the basic electricity price, there may also be differences in service fees, parking fee reductions, cross-regional clearing fees, and payment channel rates.

[0060] For example, in Zone A, the charging price for local vehicles during weekdays is 1.10 yuan per kilowatt-hour, with a service fee of 0.30 yuan per kilowatt-hour, and parking fees are waived for the first hour. In Zone B, the charging price during the same period is 1.10 yuan per kilowatt-hour, with a service fee of 0.40 yuan per kilowatt-hour, and parking fees are not waived. If both types of vehicles charge 30 kilowatt-hours and park for 90 minutes, the total cost for Zone A vehicles can be calculated as 33 yuan for electricity, 9 yuan for service, and 10 yuan for parking, totaling 52 yuan. The total cost for Zone B vehicles can be calculated as 33 yuan for electricity, 12 yuan for service, and 20 yuan for parking, totaling 65 yuan. This demonstrates that, under the same location and charging conditions, differences in service fees and parking policies alone can lead to significant cost differences.

[0061] In another embodiment, the system introduces a cross-regional settlement adjustment item for cross-regional vehicles.

[0062] For example, vehicles with local license plates registered in Region C are not charged a cross-regional clearing fee when charging at stations within that region. Vehicles with Region D license plates are charged a 2 yuan clearing fee per order when charging across regions, plus a 0.5% channel fee on some third-party payment channels based on the order amount. If the base order fee is 80 yuan, a Region C vehicle will ultimately pay 80 yuan. A Region D vehicle, if using the corresponding payment channel, will ultimately pay 80 yuan plus 2 yuan plus 0.4 yuan, totaling 82.4 yuan. This difference primarily stems from the differences in the cross-regional clearing mechanism and payment channel rules.

[0063] In another embodiment, the reservation breach rules are linked to the license plate area policy.

[0064] For example, vehicles with license plates from Region E will incur a fixed late fee of 5 yuan if they fail to arrive after the scheduled time. Vehicles with license plates from Region F will be subject to a tiered lateness mechanism: 3 yuan for late arrivals within 15 minutes and 8 yuan for late arrivals exceeding 15 minutes. If both types of vehicles are late by 20 minutes, the lateness fee for Region E vehicles will be 5 yuan, and for Region F vehicles, it will be 8 yuan. The system can incorporate lateness risk costs into the overall cost prediction when recommending target charging stations to avoid cost discrepancies during the settlement phase.

[0065] In a comprehensive example, predicted bills can be output separately for vehicles with license plates from different regions.

[0066] For example, if a vehicle requires 25 kWh of refueling during peak hours and parks for 60 minutes, the predicted total cost for a vehicle with a license plate from Region G can be calculated based on a peak-hour electricity price of RMB 1.30 per kWh, a service fee of RMB 0.25 per kWh, and a 30-minute parking fee waiver. For a vehicle with a license plate from Region X, the predicted total cost can be calculated based on a peak-hour electricity price of RMB 1.30 per kWh, a service fee of RMB 0.35 per kWh, no parking fee waiver, and an additional cross-regional clearing fee. This allows for the provision of detailed breakdowns of costs and total costs for both candidate parking lots, enabling vehicles with license plates from different regions to know their feasible settlement costs before making a reservation.

[0067] As an optional embodiment, step 101 involves obtaining real-time status information, parameter information, and corresponding payment rules for charging piles in multiple parking lots, including: establishing communication connections with charging pile terminals and user terminals in each parking lot; collecting the idle status, occupancy status, fault status, and remaining charging time of each charging pile; collecting the location, access lane information, and parking space distribution information of each parking lot; collecting the charging power, charging interface type, and compatible vehicle information of each charging pile; and collecting the charging fee standards and payment method requirements for vehicles from multiple locations in each area.

[0068] For example, in a specific application scenario, the management platform establishes bidirectional communication connections with the charging pile terminals and user terminals in parking lots A, B, and C. The management platform receives charging pile status data uploaded by each parking lot according to a preset collection cycle. Parking lot A reports that some charging piles are idle, parking lot B reports that some charging piles are occupied with remaining charging time, and parking lot C reports that some charging piles are faulty. The management platform standardizes this status data to form a comprehensive status view, allowing for priority use of idle and healthy charging pile resources during subsequent selection.

[0069] Optionally, the global status view is a global resource status mapping formed by uniformly aggregating and standardizing the availability, health, timeliness, and waiting costs of charging piles in multiple parking lots. The global status view can be constructed in four stages: data access, data standardization, status fusion calculation, and view generation. First, the management platform accesses raw status data from the charging pile terminals in each parking lot. This raw status data includes at least the unique identifier of the charging pile, the identifier of the parking lot to which it belongs, the status type, the status timestamp, the remaining charging time, the fault code, and the time of the most recent heartbeat. Subsequently, the management platform maps the heterogeneous fields reported by different parking lots into a unified status field system to eliminate field differences caused by different equipment manufacturers and different site protocols.

[0070] Next, the management platform performs time alignment and validity verification on the standardized status data. Time alignment unifies data from different sources into the same sampling time window, while validity verification identifies missing data, duplicate reports, and abnormal data jumps. For data that has exceeded its validity period, the management platform marks it as invalid and triggers re-collection. For conflicting status data, the management platform arbitrates based on timestamp priority and device heartbeat reliability to obtain a unique valid status record for each charging pile within the current time window.

[0071] The management platform constructs a status scoring vector based on a unique and valid status record. This vector consists of an availability score, a health score, a timeliness score, and a waiting cost score. The availability score is determined based on idle status, occupied status, and reserved / locked status. The health score is determined based on the fault code level and the number of recent alarms. The timeliness score is determined based on status reporting latency and heartbeat stability. The waiting cost score is determined based on remaining charging time and queue length. The management platform merges these scoring vectors according to preset weights to obtain a comprehensive status value for each charging station.

[0072] Finally, the management platform aggregates the comprehensive status values ​​of each charging pile according to the parking lot, region, and overall domain dimensions to form a global status view. The global status view includes at least the charging pile availability status layer, the equipment health status layer, the timeliness reliability status layer, and the waiting cost status layer. Based on this global status view, the management platform can output a list of available charging piles, a list of high-risk devices, and a list of devices to be prioritized for scheduling in real time, and provide these lists to the subsequent resource filtering and priority ranking modules.

[0073] For example, within the same sampling time window, charging pile A in parking lot A is available and fault-free, charging pile B in parking lot B is occupied with a long remaining charging time, and charging pile C in parking lot C has a critical fault code. After the management platform completes field unification, time alignment, and conflict arbitration, it calculates the corresponding comprehensive status value for each, and marks the charging pile in parking lot A as a priority candidate, the charging pile in parking lot B as a delayed candidate, and the charging pile in parking lot C as an unavailable device in the global status view. Through this construction process, consistent expression and computable scheduling of status data across parking lots can be achieved.

[0074] In the same scenario, the management platform also simultaneously collects parking lot-level site information. Parking lot A uploads its location and main entrance / exit capacity; parking lot B uploads peak-hour congestion coefficients and parking space density; and parking lot C uploads zoned parking space layouts and information on temporarily closed passages. The management platform maps this site information into traffic efficiency and accessibility parameters, which are used to assess the ease of access and queuing risk for users from their current location to their target parking lot during the candidate selection and priority ranking stages.

[0075] During the parameter information collection phase, the management platform obtains charging power, charging interface type, and compatible vehicle model information from each charging pile terminal. Some devices in parking lot A support high-power DC interfaces, while some devices in parking lot B only support medium-power charging. The interface types of devices in parking lot C are more varied. The management platform matches the user's vehicle interface type with the device interface type, compares the expected charging power with the device's output power, and, combined with the compatible vehicle model information, eliminates devices with insufficient compatibility, thus forming a candidate set that meets both interface compatibility and power threshold constraints.

[0076] During the fee payment rule collection phase, the management platform obtains charging fee standards and payment method requirements by region. Region 1 uploads time-of-use electricity pricing rules and parking fee reduction rules; Region 2 uploads service fee surcharge rules and cross-regional clearing rules; and Region 3 uploads available payment channels and identity authentication requirements. When generating recommendation results, the management platform simultaneously incorporates these rules into the fee prediction and payment feasibility verification process to avoid situations where charging is feasible but settlement is not. Specific implementation details are provided in the following examples and will not be elaborated upon here.

[0077] For example, a user in a multi-location area submitted a request for rapid charging during the evening rush hour, with a limited budget. Based on the collected data, the management platform first filters out faulty and incompatible devices, then combines parking lot traffic efficiency and remaining charging time to screen feasible candidates. Subsequently, it calculates the comprehensive cost by incorporating regional time-of-use electricity pricing and payment rules, ultimately outputting a list of available charging stations and their corresponding ranking. This process enables unified perception, filtering, and settlement adaptation of resources across different parking lots.

[0078] As an optional embodiment, step 101, obtaining vehicle information and charging demand information submitted by users in multiple locations, includes: The system receives vehicle information and charging request information submitted by the user terminal. The vehicle information includes the area information corresponding to the vehicle's license plate, the vehicle's battery capacity, and the type of charging interface. The charging request information includes charging time, charging power, target parking area, and charging budget. With user authorization, the system performs character recognition on the license plate image and compares the recognized license plate information with a license plate database to verify the authenticity of the vehicle information. The system also performs a compatibility check on the user's submitted payment account and the payment rules for the target area. After successful verification, the system binds the user information to the payment account to generate a user charging identity identifier.

[0079] Specifically, in one embodiment, after a user accesses the charging service page through a user terminal, the terminal first receives the vehicle information and charging demand information filled in or selected by the user. The vehicle information includes at least the area information corresponding to the vehicle's license plate, the vehicle's battery capacity, and the type of charging interface. The charging demand information includes at least the planned charging time, desired charging power, target parking area, and charging budget. The user terminal uploads the above information to the management platform according to a preset field structure. The management platform performs a preliminary verification of the field completeness and format validity to form the business data to be verified.

[0080] With user authorization, the user terminal captures vehicle license plate images and uploads them to the management platform. The management platform uses a character recognition module to extract characters from the license plate image, obtaining the recognized license plate information. This information is then compared with the registered license plate information in the vehicle license plate database. If the recognition result matches the registered information, the vehicle information authenticity verification is considered successful. If inconsistencies exist or the recognition confidence level is below a preset threshold, a secondary verification process is triggered. This secondary verification may include re-capturing the license plate image or manual confirmation to improve the reliability of the vehicle information authenticity verification.

[0081] In one embodiment, after verifying the authenticity of vehicle information, the management platform performs a region compatibility check on the payment account submitted by the user. The management platform first obtains payment rule data corresponding to the target parking area. This payment rule data may include available payment channels, settlement currency requirements, identity authentication requirements, and clearing constraints. Subsequently, the management platform extracts the user's payment account's account type, channel attributes, authentication status, and availability status, and matches them with the target area's payment rules. If the payment account meets the target area's payment rules, the payment compatibility check is deemed successful. If not, the platform returns a reason for failure to the user's terminal and prompts them to switch to an available payment account.

[0082] In one embodiment, after both vehicle information authenticity verification and payment compatibility verification pass, the management platform performs user information binding with the payment account. The binding process includes establishing the association between the user identifier, vehicle identifier, payment account identifier, and regional rule identifier, and generating a corresponding user charging identity identifier. This user charging identity identifier can be used for subsequent reservation locking, charging unlocking, payment settlement, and record archiving, achieving unified identity recognition and consistent management across parking lots and regions.

[0083] For example, a user submits vehicle location information, battery capacity, interface type, evening charging plan, target area, and budget information on the terminal, along with an image of the license plate. The management platform verifies the vehicle's authenticity by recognizing the license plate characters and comparing them with the database. It then matches the user's linked payment account with the target area's payment rules to confirm the account is usable for payment in that area. Once verification is successful, the platform generates a user charging identity identifier and returns it to the user's terminal. Subsequently, when the user initiates a reservation at any participating parking lot, the system can quickly complete permission identification, reservation locking, and settlement processing based on this identity identifier.

[0084] Optionally, the process of verifying the compatibility of the user-submitted payment account with the payment rules of the target region includes: obtaining the payment method requirements, payment channel access rules, settlement currency requirements, and identity authentication requirements corresponding to the target region; extracting the account type, payment channel, settlement attribute, and authentication status of the user-submitted payment account; using the target region's payment rules as a reference sequence and the payment account attributes as a comparison sequence; employing a grey relational analysis method based on a multidimensional rule feature set to map the target region's payment rules to the reference sequence and the payment account's account type, channel, settlement attribute, and authentication status to the comparison sequence, and calculating the correlation degree between the payment account attributes and the target region's payment rules through a correlation coefficient matrix; wherein, the multidimensional rule feature set is constructed based on the payment method requirements, payment channel access rules, settlement currency requirements, and identity authentication requirements; determining a compatibility score based on the correlation degree, and determining that the payment account passes the compatibility verification when the compatibility score reaches a preset threshold.

[0085] Specifically, after receiving the payment account submitted by the user, the management platform first obtains the payment rule data corresponding to the target region. This payment rule data can be provided by a regional rule sub-library and includes at least payment method requirements, payment channel access rules, settlement currency requirements, and identity authentication requirements. The management platform performs structured encoding on the payment rule data to form a rule dimension vector, where each dimension corresponds to a rule constraint term, used for subsequent correlation calculations.

[0086] The management platform extracts and standardizes user payment account attributes. These attributes include at least account type, payment channel, settlement attribute, and authentication status. Account type indicates whether the account belongs to a bank card account, e-wallet account, or other settlement-enabled account type. Payment channel indicates the payment network or clearing network the account can access. Settlement attribute indicates the currency and settlement range. Authentication status indicates whether the account has completed the required real-name, risk control, or cross-regional identity authentication for the target region. The management platform maps these account attributes to attribute vectors isomorphic to the rule dimension vector, forming a comparison sequence.

[0087] The management platform uses the rule vector obtained by mapping the payment rules of the target area as a reference sequence and the attribute vector obtained by mapping the payment account attributes as a comparison sequence. It then constructs a multi-dimensional rule feature set based on payment method requirements, payment channel access rules, settlement currency requirements, and identity authentication requirements. This multi-dimensional rule feature set is used to define the feature dimensions and dimension weights in grey relational analysis. Subsequently, the management platform performs dimensionless processing on the reference and comparison sequences to eliminate the influence of differences in the value ranges of different dimensions on the correlation calculation results.

[0088] Optionally, when constructing the reference sequence and comparison sequence, the management platform first reads the target region rule template from the regional rule sub-library, establishes a set of rule fields according to payment method requirements, channel access rules, settlement currency requirements, and identity authentication requirements, and assigns a dimension number to each field. For discrete fields, enumeration encoding or one-hot encoding is used to form rule vector components. For interval fields, upper and lower limit interval mapping is used to form rule vector components. The management platform then extracts the account type, channel, settlement attribute, and authentication status from the user's payment account and generates attribute vectors according to the same dimension number. If there are missing fields, they are filled with preset default values ​​and a missing marker is recorded. Subsequently, dimensionless processing is performed on the reference sequence and comparison sequence. Discrete dimensions are uniformly normalized, and continuous dimensions are minimum-maximum normalized, so that the values ​​of each dimension are unified to the same numerical range to reduce the impact of dimensional differences on subsequent correlation calculations.

[0089] The management platform calculates the correlation coefficients of each dimension's features based on grey relational analysis, forming a correlation coefficient matrix. This matrix characterizes the matching degree of payment accounts across each rule dimension. The platform further aggregates the correlation coefficient matrix according to preset weights to obtain a comprehensive correlation degree. A higher comprehensive correlation degree indicates a higher consistency between the payment account and the target region's payment rules, and stronger settlement feasibility. In one embodiment, the platform calculates the absolute value of the difference between the dimensionless reference sequence and the comparison sequence dimension by dimension, obtaining the difference sequence for each dimension. The platform extracts the minimum and maximum difference values ​​across all dimensions and sets a resolution coefficient. Based on grey relational calculation rules, the correlation coefficients for each dimension are obtained, forming a correlation coefficient matrix. If multiple payment account candidates exist, they are organized by account identifier as matrix rows and rule dimensions as matrix columns. The platform then weights and aggregates the correlation coefficients for each dimension according to preset dimension weights to obtain the comprehensive correlation degree for each payment account.

[0090] The management platform generates a suitability score based on comprehensive relevance and compares this score with a preset threshold. When the suitability score reaches the preset threshold, the payment account is deemed to have passed the suitability verification and is added to the set of available payment accounts. When the suitability score is below the preset threshold, the payment account is deemed to have failed the suitability verification, and the platform returns the reason for failure and suggested alternative payment methods to the user terminal. Reasons for failure include, but are not limited to, currency mismatch, unreachable channel, or authentication status not meeting regional rule requirements. Furthermore, to ensure settlement security, the management platform can set a hard constraint priority rule, meaning that if the settlement currency is mismatched, the channel is unreachable, or the authentication status does not meet the mandatory regional requirements, the account is directly deemed to have failed. Threshold determination is then performed based on the suitability score, provided the hard constraints are met. For accounts that pass verification, they are added to the set of available payment accounts and a validity period marker is generated. For accounts that fail verification, the management platform generates a failure reason code and a readable reason description, and recommends alternative payment accounts or channels to the user terminal in descending order of comprehensive relevance, so that the user can re-initiate verification after switching accounts.

[0091] For example, the target region rules require payment to be made through either Channel A or Channel B, settlement to be in Currency A, and cross-regional identity verification to be completed. A user submits a payment account with Channel A as the payment channel and Currency A as the settlement currency, but the cross-regional verification status is incomplete. The management platform constructs a reference sequence and a comparison sequence, then performs grey relational calculations, finding a high correlation coefficient between the payment method and currency dimensions, and a low correlation coefficient between the verification dimension. After weighted aggregation, the suitability score fails to reach the threshold, the system determines it as unsuccessful and prompts the user to supplement verification or switch to a verified payment account. Through this process, payment feasibility verification can be completed before the reservation, reducing the risk of settlement failure after charging.

[0092] Step 102: Based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and the charging and payment rules of the corresponding area, construct a cross-site charging pile resource matching model with the optimization objectives of charging pile resource utilization rate, user charging cost, and charging waiting time, and screen candidate charging piles to obtain at least one target charging pile.

[0093] The cross-site charging pile resource matching model is a multi-objective optimization decision-making model for collaborative scheduling of multiple parking lots. This model takes the user's current charging task as the decision input, the candidate charging pile set as the decision space, and resource utilization, overall cost, and waiting time as joint optimization objectives, outputting target charging piles and their ranking results that meet the constraints. This model is not a static recommendation model for a single parking lot, but rather a global matching model capable of simultaneously handling cross-site status differences, cross-regional billing rule differences, and reservation time period conflict constraints.

[0094] In practical applications, the model structure can be divided into a data layer, a constraint layer, an optimization layer, and a decision output layer. The data layer receives vehicle information, charging demand information, real-time charging pile status information, parameter information, and regional charging and payment rules. The constraint layer establishes interface matching constraints, reservation time period constraints, budget constraints, and accessibility constraints. The optimization layer performs multi-objective solving to obtain a set of candidate solutions. The decision output layer generates a list of target charging piles, priority scores, and recommendation reason fields for subsequent reservation and locking steps. Furthermore, model parameters include, but are not limited to: candidate parking lot identifiers, charging pile identifiers, reservation time period parameters, vehicle arrival time estimation parameters, route toll cost parameters, interface adaptation status parameters, time-of-use electricity price parameters, service fee parameters, parking fee parameters, and budget threshold parameters. Among these parameters, real-time status parameters can be dynamically updated according to the sampling period, rule parameters can be updated according to the regional rule version, and user preference parameters can be updated per task. This hierarchical parameter management ensures the maintainability and scalability of the model in cross-regional scenarios.

[0095] For example, in the feature construction stage, multi-source data is uniformly mapped into structured feature vectors. In the candidate generation stage, an initial candidate set is generated based on the target parking area and interface reachability. In the constraint encoding stage, hard and soft constraints are encoded separately. In the objective function definition stage, the overall resource utilization rate, user comprehensive cost, and expected waiting time are constructed into a joint optimization objective, providing a mathematical object for subsequent improvement of the genetic algorithm.

[0096] As an optional embodiment, in step 102, a candidate charging pile set is constructed based on vehicle information and charging demand information, combined with real-time status information of charging piles, site location information, parameter adaptation information and corresponding regional charging rules; the candidate charging pile set is screened using charging pile resource utilization rate, total user charging cost and user waiting time as multi-objective optimization indicators; and at least one target charging pile that meets the current charging demand is determined according to the screening results.

[0097] First, the candidate charging pile set can be constructed using a hard filtering followed by a soft scoring mechanism. The hard filtering stage first eliminates faulty devices, devices with incompatible interfaces, devices unavailable during certain time periods, and devices with significantly exceeded budgets. The soft scoring stage then calculates a comprehensive score for the remaining devices. This mechanism is adopted because eliminating infeasible solutions beforehand reduces the solution space and improves the efficiency of subsequent multi-objective optimization.

[0098] Optionally, a suitability evaluation factor can be added to the candidate set screening. This suitability evaluation factor may include at least interface suitability, vehicle-parking lot distance, and charging fee suitability. The management platform performs a weighted calculation on each candidate charging pile to obtain a suitability score, and then integrates this score with the multi-objective optimization solution results to obtain the final recommendation score. This mechanism can enhance user experience consistency while meeting optimization objectives, and reduce recommendation bias caused by the model only pursuing a single cost optimum.

[0099] The implementation process of step 102 is illustrated below with a specific example. In one example, the user's vehicle is a DC fast-charging model with a low current battery level. The user plans to recharge in the target area during the evening peak hours, but has a limited budget. The management platform first obtains real-time status and parameter information from three parking lots, with an initial candidate of twenty devices. Optionally, the selection of the initial candidate charging piles can employ a joint screening method of regional constraints and basic reachability constraints. For example, the management platform first extracts online charging piles from the set of parking lots within the area based on the user's submitted target parking area and planned charging time as the first candidate set. Then, based on the vehicle's current location and the parking lot location, accessibility is calculated, and devices whose estimated arrival time exceeds the user's acceptable time limit are eliminated, forming the initial candidate set. Furthermore, a coarse matching can be performed based on the vehicle interface type and charging power requirements to obtain an initial device pool for subsequent fine screening, thereby avoiding the introduction of devices that clearly do not meet the basic requirements into the optimization process.

[0100] Next, in the example above, after hard filtering, nine devices were eliminated due to interface mismatch, fault status, and time period conflicts, leaving eleven devices for optimization. Optionally, hard filtering can combine rule engine judgment with threshold verification. For example, the management platform sequentially performs fault status verification, interface compatibility verification, time period availability verification, and budget feasibility verification on each candidate device. Fault status verification eliminates devices in fault, offline, or maintenance states. Interface compatibility verification eliminates devices with mismatched interface types or protocol versions. Time period availability verification checks the overlap ratio between the candidate device's idle time period and the user's planned charging time; devices with an overlap less than a threshold are eliminated. Budget feasibility verification quickly estimates the minimum feasible cost based on the target area's charging rules; devices significantly exceeding the user's budget limit are directly eliminated. Through this hard filtering method, infeasible solutions can be eliminated before optimization, reducing the complexity of the feasible solution space and improving subsequent solution efficiency.

[0101] Finally, in the example above, the improved genetic algorithm iteratively solves for the remaining eleven devices within the feasible solution space, obtaining three non-dominated optimal solutions. These are then combined with a fitness evaluation factor to calculate the final score, outputting two target charging stations as recommendations along with suggested reservation times.

[0102] In this example, if a candidate charging station has a low charging cost but an estimated long waiting time, its score will be lowered by the waiting penalty in the multi-objective evaluation. If another candidate charging station is closer but its payment rules are not well-suited, it will be downgraded during the constraint repair or suitability fusion stage. The target charging stations that are ultimately retained typically meet the requirements of accessibility, availability, settlement, and cost control, and can more stably support subsequent reservation locking and billing settlement processes.

[0103] Through the above steps 102 and their embodiments, this application can achieve full-domain collaborative matching and dynamic optimization scheduling of charging resources in real-world scenarios involving cross-site, cross-regional, and heterogeneous rules. Compared with the single-site static screening method, this application has better engineering adaptability and application effects in terms of resource utilization, user waiting time control, and overall cost optimization, and provides high-quality input for subsequent steps such as reservation, billing, and settlement.

[0104] It is understood that the improved genetic algorithm in this embodiment is used to solve the cross-site charging pile resource matching problem. Under the premise of satisfying the charging interface matching constraint, reservation time period constraint, and user budget constraint, the algorithm jointly optimizes the utilization rate of charging pile resources across the entire region, the total user charging cost, and the user waiting time, and searches for the comprehensive optimal solution from the candidate charging pile set to obtain a target charging pile recommendation result with high executability and better user experience.

[0105] Specifically, the management platform receives vehicle information, charging demand information, real-time charging pile status information, site location information, and regional charging rule information. It eliminates faulty equipment, incompatible interface equipment, and equipment unavailable during specific time periods, forming a candidate charging pile set, and simultaneously establishes interface constraints, time period constraints, budget constraints, and accessibility constraints. Furthermore, the management platform encodes a feasible scheduling scheme as a chromosome individual. This chromosome includes at least candidate parking lot identifier gene segments, charging pile identifier gene segments, reservation time period gene segments, route toll cost parameter gene segments, and interface adaptation status parameter gene segments. This multi-segment encoding structure allows for the simultaneous expression of site selection, equipment selection, and time period selection within a single entity, reducing suboptimal problems caused by step-by-step decision-making. Next, the management platform constructs initial individuals based on interface matching constraints, idle time period constraints, and budget constraints, retaining only individuals that meet the constraints in the initial population. Individuals near the constraint boundaries can be marked for repair, allowing for priority repair and re-evaluation in subsequent iterations. Finally, a fitness function is constructed, and the initial evaluation is completed. The management platform constructs a comprehensive fitness function based on charging costs, parking costs, cross-site travel costs, estimated waiting time, reservation conflict penalties, and constraint violation penalties. Each indicator is first normalized and then aggregated according to preset weights to form a comprehensive fitness value. A higher comprehensive fitness value indicates a better balance between economy, timeliness, and feasibility. Finally, the management platform generates a new generation of individuals in the order of selection, crossover, and mutation. An adaptive update strategy is used for crossover and mutation probabilities, improving global search capabilities in the early stages of iteration and enhancing local convergence capabilities in the later stages to reduce the risk of getting trapped in local optima and improve solution stability. The management platform performs feasibility checks on the new generation of individuals, prioritizing repairs for individuals that violate interface constraints, time period constraints, or budget constraints. Repair methods include time period shifting, site replacement, and power downgrading matching; individuals failing repair are discarded. Simultaneously, the management platform directly replicates high-fitness individuals from the current generation to the next generation according to the elite retention ratio to avoid losing high-quality genes. Finally, termination conditions are determined, and candidate optimal solutions are output. The management platform terminates iteration when it reaches the maximum number of iterations or when the optimal fitness change is below the convergence threshold for multiple consecutive generations, and outputs at least one target charging station and alternative solutions. The output may include recommended appointment times, estimated total cost, estimated waiting time, and the ranking of alternative solutions.

[0106] Furthermore, the management platform sends the target charging pile results to the reservation module for reservation locking, and transmits the estimated cost and time period information to the billing and settlement module. Through the above process, the improved genetic algorithm not only achieves joint optimization solutions across site resources, but also directly supports subsequent reservation, billing, and settlement processes, improving overall business execution efficiency and user charging experience.

[0107] Further optionally, in the above embodiments, the candidate charging pile set is screened using charging pile resource utilization rate, total user charging cost, and user waiting time as multi-objective optimization indicators. This includes: constructing a multi-objective optimization function with the objectives of maximizing the utilization rate of charging pile resources across the entire region, minimizing user charging costs, and minimizing charging waiting time; and solving the multi-objective optimization function using an improved genetic algorithm. The improved genetic algorithm adopts a multi-segment coded chromosome structure oriented towards cross-site charging reservation scenarios. The multi-segment coded chromosome includes at least candidate parking lot identifiers, charging pile identifiers, reservation time periods, path toll cost parameters, and connection parameters. The system considers various factors, including: interface matching parameters; population initialization; generating feasible solutions based on interface matching constraints, charging pile idle time constraints, and user charging budget constraints; fitness evaluation; constructing a fitness function by considering charging costs, parking costs, cross-site travel costs, expected waiting time, and reservation conflict penalties; introducing elite retention and infeasible solution repair mechanisms during genetic iteration to repair or remove individuals that violate interface matching constraints, reservation time constraints, or budget constraints; and adaptively adjusting crossover and mutation probabilities based on the number of iterations during the solution process to reduce the impact of local optima on the selection results.

[0108] Specifically, in one embodiment, optimization objective modeling is performed first. The management platform establishes three optimization objectives for the same user task, corresponding to the overall charging pile resource utilization rate, the total user charging cost, and the user waiting time, respectively. Resource utilization rate reflects whether the charging resources across the entire region are used evenly; the total user charging cost reflects the overall cost to the user in this task; and the user waiting time reflects the timeliness of charging availability after reservation. The management platform unifies the dimensions of the three objectives before proceeding to the subsequent solution process to avoid any single objective dominating the result due to its large numerical range.

[0109] A multi-segment coding chromosome is adopted for cross-site reservations. Each individual contains at least the candidate parking lot identifier, charging pile identifier, reservation time period, route cost parameters, and interface adaptation status parameters. The principle of this coding method is to put site selection, equipment selection, and time period selection into the same decision unit, avoiding local optima caused by selecting the site first and then the charging pile. For example, an individual can be represented as parking lot B, charging pile number six, reservation time period from 7:00 PM to 7:30 PM, route cost medium, and interface adaptation set to "pass".

[0110] The initialization phase of the population does not involve random full generation; instead, a set of feasible solutions is constructed based on constraints. The management platform first checks if the interface matches, then checks if the scheduled time slot overlaps with the idle time slot, and finally checks if the predicted cost exceeds the user's budget. Only individuals satisfying these three types of constraints are retained in the initial population. This approach reduces the proportion of infeasible solutions and improves early iteration efficiency. For example, if the initial candidates are twenty stalls, seven are removed due to interface mismatch, four due to time slot conflicts, and two due to budget overruns, leaving seven stalls to generate the initial population.

[0111] The fitness assessment phase employs a comprehensive evaluation function. The management platform incorporates charging costs, parking fees, cross-site travel costs, estimated waiting time, and reservation conflict penalties into the fitness calculation. The principle is to quantify economy, timeliness, and feasibility simultaneously, avoiding excessively long queues or reservation conflicts caused by solely selecting based on lower electricity costs. For example, among two candidate options, Option 1 has lower electricity costs but requires a 30-minute wait and has a high risk of conflict, while Option 2 has slightly higher electricity costs but almost no waiting time. Therefore, Option 2 may have a better overall fitness rating.

[0112] An elite retention mechanism is introduced during the genetic iteration phase. In each generation, the top 10% of individuals with the highest fitness are directly retained for the next generation, preventing the loss of high-quality solutions during crossover and mutation. The principle behind this is to stabilize the convergence path and improve the reproducibility of the solution results. For example, the top 10% of individuals are retained in each generation, while the remaining individuals participate in selection, crossover, and mutation to generate new solutions.

[0113] Furthermore, an infeasible solution repair mechanism is introduced simultaneously. Instead of directly deleting all individuals that violate constraints, they are repaired first. The repair order can be set to repair interface constraints first, then time period constraints, and finally budget constraints. If repair fails, the individual is then removed. The principle is to preserve valid gene fragments within the individual, improving population diversity. For example, if an individual only has a conflict during a reservation time period, the system can shift its time period to the nearest available window and recalculate its fitness.

[0114] The crossover and mutation probabilities employ an adaptive adjustment strategy. In the early stages of iteration, the crossover and mutation probabilities are increased to enhance the global search, while in the mid-to-late stages, the mutation probability is gradually decreased to stabilize convergence. The principle behind this is to balance exploration and convergence capabilities, reducing the risk of getting trapped in local optima. For example, the crossover probability can be set to a higher range in the early stages, gradually decreasing in the mid-stages, while the mutation probability gradually decreases from a medium level to a lower level.

[0115] The system outputs results upon reaching the termination condition. The termination condition can be reaching the maximum number of iterations or the optimal fitness change being below a threshold for multiple consecutive generations. The management platform outputs at least one target charging station and can simultaneously output alternative charging stations and their corresponding reservation time slots. For example, the final output might be charging station three in parking lot A as the first choice, charging station six in parking lot B as an alternative, along with a recommended reservation window and estimated total cost.

[0116] For example, a user's current battery level is 30%, and the goal is to increase it to 80%, aiming to recharge during the evening rush hour with a budget of no more than 80 yuan. After obtaining real-time status data from three parking lots, a population of feasible solutions is constructed. After multiple generations of genetic iteration, the preferred solution is a fast-charging station in parking lot B. Although this solution does not have the lowest electricity price, it has lower access costs, shorter waiting times, and lower risk of reservation conflicts, resulting in the highest overall fitness. Therefore, the improved genetic algorithm described above can obtain more feasible target charging station selection results under cross-site and multi-constraint conditions.

[0117] Further optionally, in the above embodiments, the screening of the candidate charging pile set further includes: constructing a charging pile adaptability evaluation factor based on charging interface adaptability, distance between the vehicle and the parking lot, and charging adaptability; performing weighted calculation on each candidate charging pile to obtain a corresponding adaptability score; and determining at least one target charging pile based on the adaptability score and the solution result of the multi-objective optimization function.

[0118] In an optional embodiment, after constructing the candidate charging pile set and solving the multi-objective optimization function, the management platform further introduces a charging pile suitability evaluation mechanism to enhance the business executability and user experience consistency of the candidate solutions. This suitability evaluation mechanism uses charging interface suitability, the distance between the vehicle and the parking lot, and the degree of fee suitability as core evaluation dimensions to form a suitability evaluation factor set, and calculates the suitability score for each candidate charging pile accordingly. By integrating the suitability score with the multi-objective optimization solution results, the recommendation bias caused by relying solely on the optimization function can be reduced, improving the availability and user acceptability of the final target charging pile.

[0119] The management platform first calculates the charging interface compatibility factor for each candidate charging pile. Specifically, it reads the vehicle's interface type, protocol version, and acceptable charging power range, then reads the candidate charging pile's interface type, protocol capability, and rated output capability. If the physical interface types are inconsistent, the candidate charging pile's interface compatibility factor is set to the lowest value, and it is either downweighted or eliminated. If the interface types are consistent, the platform further assesses protocol compatibility and power matching to obtain tiered values ​​for the interface compatibility factor. This allows for a hierarchical determination from physical access feasibility to charging efficiency feasibility.

[0120] The management platform calculates distance factors using path reachability distance or path reachability time, rather than simple straight-line distance. Specifically, based on the vehicle's current location and the locations of each candidate parking lot, it calls the route planning service to obtain the actual travel distance and estimated travel time. Then, it maps the distance or time to a distance adaptation value according to a preset normalization rule. The shorter the distance or the shorter the travel time, the higher the corresponding distance adaptation value. This approach more accurately reflects the convenience of arrival in cross-site scenarios.

[0121] When calculating the pricing fit factor, the management platform first predicts the total cost of candidate plans based on regional pricing rules, and then matches it with the user's budget and pricing preferences. The total cost may include charging fees, parking fees, service fees, and cross-regional settlement surcharges. If the predicted total cost exceeds the budget threshold, the pricing fit factor decreases. If the total cost is within the budget and the rate structure aligns with the user's preferences, the pricing fit factor increases. This factor helps mitigate the risk of user non-payment and inconsistent user experience during the settlement phase, which is crucial during the screening stage.

[0122] The management platform calculates a weighted average of the three factors to obtain a suitability score for the candidate charging piles. The calculation logic can be described as follows: weights are assigned to the interface suitability factor, distance factor, and charging suitability factor, respectively. These weights can be preset by the system or dynamically adjusted according to the scenario. The three factors are then linearly aggregated to form the suitability score. To avoid any single factor dominating the score, the management platform normalizes each factor before aggregation and can set upper and lower limits for each factor to ensure score stability.

[0123] The management platform integrates the fit score with the results of the multi-objective optimization function to form the final decision score. The integration process can employ a two-stage mechanism. The first stage obtains a candidate optimization set based on the multi-objective optimization results. The second stage sorts the candidates within the optimization set according to their final decision scores. The final decision score is obtained by proportionally integrating the multi-objective optimization value and the fit score. The principle behind this mechanism is to first ensure the global optimization direction, and then improve the business feasibility of a single recommendation through fit correction.

[0124] The following example illustrates the solution and calculation process. Assume a user's vehicle is a DC fast-charging model, and there are three candidate charging stations A, B, and C within the target area. Interface testing shows that A and B are fully compatible, while C is only partially compatible. Route planning estimates the travel time to A at 12 minutes, B at 8 minutes, and C at 10 minutes. Cost prediction shows the total cost to be 68 yuan for A, 75 yuan for B, and 62 yuan for C, with the user's budget capped at 70 yuan. Based on this, the management platform obtains factor evaluation results: A is superior in terms of interface and cost; B is optimal in terms of distance but slightly more expensive; and C has lower cost but weaker protocol compatibility. The platform then performs a weighted calculation according to preset weights to obtain a suitability score for each candidate charging station, and merges and sorts this score with the multi-objective optimization output. If A has the highest score after merging, A is selected as the target charging station, and B is output as a backup option.

[0125] In this example, even if B has the shortest arrival time, its final score may still decrease due to exceeding the budget. Similarly, even if C has the lowest cost, insufficient interface compatibility may affect its feasibility. Through the above calculation logic, the system can strike a balance between cost, efficiency, and feasibility, ultimately outputting target charging pile results that better meet the actual usage needs across different sites. This solution enhances screening stability and reduces the probability of reservation failures and settlement conflicts without altering the main optimization framework.

[0126] Step 103: Based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and the reservation lock duration of charging piles, predict the estimated time for vehicles in multiple locations to arrive at each target parking lot, and prioritize at least one target charging pile by combining the estimated time, the idle time of the target charging pile, and the traffic efficiency of the parking lot.

[0127] Optionally, the network can be specifically selected based on the needs of time-series prediction scenarios involving vehicles traveling across regions. If a balance between long-term prediction accuracy and low computational complexity is required, either the Informer network or the Autoformer network can be selected. The Informer network is a long-series time-series prediction network based on probabilistic sparse attention, while the Autoformer network is a time-series prediction network based on sequence decomposition and autocorrelation mechanisms. If the ability to capture the global correlation of multi-source time-series features needs to be enhanced, the Transformer network can be selected. The Transformer network is a time-series prediction network based on a self-attention mechanism. If the periodic changes in road conditions and historical travel trajectories need to be accurately captured, the FEDformer network can be selected. The FEDformer network is a time-series prediction network based on a frequency domain enhancement decomposition mechanism. Alternatively, two or more of the aforementioned basic time-series prediction networks can be combined, and prediction robustness can be improved through weighted fusion.

[0128] In this embodiment, the estimated arrival time prediction in step 103 is a multi-source time-series prediction task. The input features include at least the vehicle's current location sequence, real-time traffic condition sequence, historical travel duration sequence, and reservation lock duration sequence. The output is the estimated arrival time of the vehicle at each target parking lot. To adapt to different workloads and prediction accuracy requirements, Transformer, Informer, Autoformer, and FEDformer can be selected specifically or combined and fused.

[0129] In one embodiment, the Transformer is a temporal modeling network based on a self-attention mechanism, capable of establishing global dependencies at any position in the input sequence. Its advantage lies in its strong ability to model the correlations between multi-source heterogeneous features, making it suitable for scenarios requiring the joint characterization of the coupling relationship between current location changes, traffic condition changes, and appointment time slots. Its limitation is that computational and storage costs increase with increasing sequence length. Therefore, when the goal is to enhance the global correlation capture capability of multi-source features and computational resources are relatively sufficient, the Transformer can be preferentially selected as the base prediction network.

[0130] Informer improves upon standard attention by implementing sparsity for long-sequence prediction scenarios. It reduces computational complexity through a probabilistic sparse attention mechanism while maintaining high prediction performance under long-sequence input conditions. Its advantage lies in processing long-term historical travel data with lower computational overhead, making it suitable for scenarios involving cross-regional travel that require utilizing historical information over a longer time window. Therefore, when business requirements align with both long-term prediction accuracy and online inference efficiency, Informer is the preferred choice.

[0131] Autoformer enhances its ability to model long-term trends and cyclical patterns through sequence decomposition and autocorrelation mechanisms. Its core feature is the separate modeling of trend and cyclical components in the sequence, enabling it to maintain relatively stable predictive performance under medium- to long-term traffic patterns. For scenarios involving cross-regional travel with distinct commuting rhythms and time-specific congestion patterns, Autoformer can reduce the impact of unnecessary high-frequency noise while maintaining accuracy. Therefore, when the scenario emphasizes joint modeling of long-term trends and cyclical patterns, Autoformer is a preferred choice.

[0132] FEDformer uses a frequency domain enhancement mechanism for time series modeling, enabling it to more effectively extract frequency structure information from sequences and demonstrating good expressive power for both periodic and fluctuating signals. For scenarios with significant real-time traffic fluctuations and historical traffic trajectories exhibiting multi-period superposition characteristics, FEDformer excels in fluctuation capture. Therefore, when the goal is to accurately model traffic fluctuations and periodic variation characteristics, FEDformer should be the preferred choice.

[0133] In one embodiment, the model selection process can be executed in three stages: data feature evaluation, resource constraint evaluation, and offline validation evaluation. The data feature evaluation stage statistically analyzes historical sequence length, periodic significance, fluctuation intensity, and correlation of multi-source features. The resource constraint evaluation stage assesses the inference latency, memory, and computing power limits of the target deployment environment. The offline validation stage compares the performance of candidate models on the same training and validation sets in terms of predicted arrival time error, stability metrics, and inference time metrics, and determines the final model accordingly.

[0134] If a single model cannot simultaneously meet the requirements of accuracy and robustness, two or more base networks can be fused. A weighted fusion method can be used, where the prediction results of each model are first output separately, and then fusion weights are set according to the model validation performance to obtain the final estimated arrival time. The weights can be inversely allocated based on the validation set error and updated periodically. This fusion mechanism can reduce the prediction bias of a single model under sudden road conditions or sample distribution drift, and improve the system's stability in cross-regional scenarios.

[0135] For example, in a cross-regional travel scenario where the historical travel sequence is long and low online inference latency is required, the platform can first compare Informer and Autoformer models. If the verification results show that Informer inference is faster and the error meets the threshold, Informer is selected as the main model. If the error increases during peak hours, FEDformer is introduced as an auxiliary model and weighted fusion is performed to enhance the ability to capture cyclical fluctuations. If the scenario subsequently adds multi-source behavioral features and requires enhanced global correlation modeling, Transformer can be further introduced and the weights recalibrated. Through the above dynamic selection and fusion process, step 103 can be guaranteed to have usable prediction accuracy and computational efficiency at different business stages.

[0136] As an optional embodiment, in step 103, based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and the reservation lock duration of charging stations, the estimated arrival time of vehicles in multiple locations at each target parking lot is predicted. This includes: selecting at least one of Transformer, Informer, Autoformer, and FEDformer as the basic time series prediction network; and constructing a multi-source time series network composed of location trajectory subsequences, traffic fluctuation subsequences, historical traffic duration subsequences, and reservation lock duration subsequences based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and the reservation lock duration of charging stations. The feature matrix is ​​processed by location encoding and time encoding. The encoded multi-source time-series feature matrix is ​​then weighted and optimized by combining parking lot access node weight parameters, cross-site path length parameters, and reservation time period correction factors. The access node weight parameters are set according to the road segment congestion frequency, the path length parameters are converted according to the actual mileage, and the reservation time period correction factor is dynamically adjusted according to the remaining reservation lock time to obtain the weighted optimized feature matrix. The weighted optimized feature matrix is ​​input into the basic time-series prediction network, and the estimated time for vehicles from multiple locations to arrive at each target parking lot is output through time-series dependency modeling and feature mapping.

[0137] Specifically, step 103 consists of six stages: model selection, feature construction, encoding processing, scene enhancement, network inference, and result calibration. It is used to predict the estimated arrival times of vehicles from multiple locations at various target parking lots in cross-regional driving scenarios. The core improvement of the above embodiment lies in introducing a multi-source temporal feature matrix and scene parameter weighted optimization mechanism before the basic temporal prediction network. This allows the prediction process to simultaneously perceive road condition fluctuations, path length, and reservation locking constraints, thereby improving usability in real-world business scenarios.

[0138] During the model selection phase, the management platform chooses the base time-series prediction network based on task-side constraints. If the task requires maintaining low inference complexity under long historical sequence conditions, Informer or Autoformer can be selected. If the task emphasizes modeling global dependencies between multi-source features, Transformer can be selected. If the task emphasizes the ability to capture periodic fluctuations, FEDformer can be selected. For high-uncertainty road condition scenarios, two or more networks can be used for weighted fusion, and the fusion weights can be dynamically updated based on the validation set error. This selection strategy represents a scenario-specific improvement of the base network, avoiding performance degradation of a single model across all scenarios.

[0139] During the feature construction phase, the management platform generates a multi-source temporal feature matrix using a unified time window sampling method. This multi-source temporal feature matrix consists of location trajectory subsequences, traffic condition fluctuation subsequences, historical travel duration subsequences, and reservation lock-in duration subsequences. The location trajectory subsequence reflects the spatial displacement of vehicles, the traffic condition fluctuation subsequence reflects changes in road segment congestion, the historical travel duration subsequence provides a baseline for path experience duration, and the reservation lock-in duration subsequence introduces constraints related to business time periods. This multi-source joint input avoids prediction biases caused by relying solely on single trajectory information.

[0140] During the encoding process, the management platform performs positional encoding and temporal encoding on the multi-source temporal feature matrix. Positional encoding preserves the positional relationships of the sequences, while temporal encoding injects time-specific and time-period attribute information. For samples spanning multiple days or time periods, temporal encoding can further include weekday attributes and peak / off-peak / valley time period identifiers. This processing enables the network to simultaneously perceive sequence order and temporal context during learning, improving its ability to identify peak / congestion transition boundaries.

[0141] In the scenario enhancement phase, the management platform incorporates parking lot access node weight parameters, cross-site path length parameters, and reservation time slot correction factors into the encoded feature matrix to obtain a weighted optimized feature matrix. Access node weight parameters are set based on road segment congestion frequency, with nodes experiencing higher congestion frequencies receiving higher impact weights. Path length parameters are calculated from actual drivable mileage and used to correct the basic access costs for different target parking lots. The reservation time slot correction factor is dynamically adjusted based on the remaining reservation lock time, increasing the time-sensitive penalty when the remaining time is short. This weighting mechanism is a key algorithmic improvement in this embodiment, explicitly injecting traffic layer and business layer constraints into the model input, enhancing the predictive results' support for reservation execution.

[0142] During the network inference phase, the management platform inputs the weighted optimized feature matrix into the selected base time-series prediction network. Through time-series dependency modeling and feature mapping, the estimated arrival time of each target parking lot is obtained. If a single model is used, the prediction result is directly output. If multiple models are fused, the outputs of each model are obtained separately and then weighted according to the fusion weights to obtain the final prediction result. The fusion weights can be back-allocated by the offline validation error and updated periodically to enhance the model's adaptability to distribution drift.

[0143] During the results calibration phase, the management platform verifies the consistency between the model output and the current traffic events. If a sudden congestion or temporary road closure is detected, the system can invoke the short-term correction module to incrementally correct the estimated time. The calibrated results serve as input for subsequent priority ranking of target charging stations, used in conjunction with idle time matching and parking lot traffic efficiency for joint decision-making.

[0144] For example, a user in a multi-location vehicle initiates a reservation at 6 PM. Their current location is at the boundary of the area, and the system's candidate parking lots are Lot A, Lot B, and Lot C. The management platform selects Informer as the primary model and introduces FEDformer to assist in fusion. The system constructs a multi-source temporal feature matrix within the most recent one-hour window, and after encoding, overlays scene parameters. Due to the high frequency of congestion at key nodes leading to Lot A, the weight parameters of these nodes are high, resulting in a stronger congestion impact on the route-related features of Lot A after weighting. Lot B has a shorter route but a shorter remaining reservation time; the reservation time adjustment factor increases the time constraint. Lot C has a longer route but stable road conditions; the route length parameter provides a stable toll cost adjustment. After model inference, the estimated arrival times for the three parking lots are obtained. The fusion result shows that Lot B has the fastest arrival time but a smaller remaining time, while Lot C has a slightly later arrival time but higher reservation feasibility. The system passes this result to the ranking module, ultimately providing the preferred and alternative charging piles.

[0145] In this example, a multi-source temporal feature matrix replaces the single trajectory input, enhancing the ability to represent complex traffic conditions. Contextual enhancement of the input side is achieved through traffic node weights, path lengths, and reservation correction factors, making the model output closer to the reservation execution target. A switchable single-model or fusion model mechanism improves robustness, reducing error amplification caused by single-model instability under sudden road conditions. Through the above implementation process, step 103 can provide more stable estimated arrival time prediction results under conditions of cross-regional, multi-site, and time-limited constraints.

[0146] As an optional embodiment, in step 103, prioritizing at least one target charging pile by combining the estimated time, the idle time period of the target charging pile, and the parking lot traffic efficiency includes: obtaining the idle time period, the remaining reservation lock time, and the corresponding parking lot traffic efficiency information for each target charging pile; using a Gaussian kernel time-series matching function to calculate the time period matching degree between the estimated time and the idle time period of each target charging pile, and using a multi-dimensional weighted TOPSIS algorithm to generate a priority evaluation value for each target charging pile by integrating parking lot traffic efficiency, time period matching degree, and charging cost parameters; prioritizing at least one target charging pile according to the priority evaluation value, and pushing the ranking result to the user terminal.

[0147] In an optional embodiment, the priority ranking in step 103 can be completed in four stages: data acquisition, time period matching calculation, multi-dimensional evaluation fusion, and ranking output. First, the management platform acquires idle time periods, remaining reservation lock duration, and parking lot traffic efficiency information for each target charging pile. The idle time periods are obtained by combining the charging pile status stream and the reservation table; the remaining reservation lock duration is calculated from the current time and the lock end time; and the parking lot traffic efficiency is generated by combining the entrance queue length, lane passage speed, and recent congestion index. Subsequently, the platform receives the estimated arrival time output from the first half of step 103 as the core input for the time period matching calculation.

[0148] During the time-segment matching calculation phase, the management platform uses a Gaussian kernel time-series matching function to calculate the matching degree between the expected arrival time and the idle time period. Specifically, the deviation between the expected arrival time and the center time of the idle window can be used as an input variable. The smaller the deviation, the higher the matching degree; the larger the deviation, the matching degree decays smoothly in a Gaussian form. For cases exceeding the idle window boundary, the system introduces a remaining reservation lock duration correction term to impose additional penalties on schemes that may cause lock failure. Compared with the linear interpolation method, this method is more sensitive to small deviations and more stable to large deviations, reducing frequent sorting reversals caused by boundary jitter.

[0149] In the multi-dimensional evaluation and fusion phase, the management platform constructs an evaluation matrix, with evaluation dimensions including at least parking lot traffic efficiency, time-period matching degree, and charging cost parameters. Subsequently, a multi-dimensional weighted TOPSIS algorithm is used for comprehensive evaluation. Specifically, the process involves dimensionless processing of each indicator, generating a weighted decision matrix based on business-side weights, constructing positive and negative ideal solutions, and calculating the distances of each candidate charging pile to both solutions to obtain the proximity score as a priority evaluation value. A higher proximity score indicates that the candidate solution is superior in terms of traffic efficiency, time-period feasibility, and cost controllability.

[0150] In terms of algorithm improvement, this embodiment has two enhancements compared to conventional sorting methods. First, the time-slot matching is changed from the traditional hard threshold judgment to Gaussian kernel continuous matching, which transforms the relationship between the expected arrival time and the idle window from discrete feasibility / infeasibility to a continuous and comparable score, improving the stability of the sorting. Second, TOPSIS introduces parameters of time-slot matching degree and charging cost under the influence of the remaining reservation lock time, forming a coordinated decision-making process that considers traffic efficiency, time feasibility, and economy, avoiding single-dimensional suboptimal results caused by relying solely on distance or cost.

[0151] For example, suppose there are three target charging stations A, B, and C under the same user task. The system predicts that the user's expected arrival time is 19:12. A's idle window is from 19:10 to 19:30, with 18 minutes remaining in the lockout, high parking efficiency, and an estimated cost of 68 yuan. B's idle window is from 19:00 to 19:15, with 5 minutes remaining in the lockout, medium efficiency, and an estimated cost of 62 yuan. C's idle window is from 19:20 to 19:45, with 25 minutes remaining in the lockout, low efficiency, and an estimated cost of 60 yuan. After Gaussian kernel timing matching, A achieves a high matching degree because its arrival time is close to the center of the window. B, although having a lower cost, suffers a penalty due to insufficient lockout margin. C's matching degree decreases because its arrival time is earlier than the start of the window. After weighted TOPSIS fusion of traffic efficiency and cost indicators, A receives the highest priority evaluation value, C is the second best, and B is the third best. Based on this, the system pushes the sorting results to the user's terminal and prioritizes the reservation and locking of the time period corresponding to A.

[0152] In engineering implementation, weights can be dynamically configured according to regional congestion characteristics. During peak hours, the weights for traffic efficiency and time-of-day matching can be increased, while during off-peak hours, the weights for cost can be appropriately increased. The Gaussian kernel bandwidth parameter can be adaptively updated according to the historical arrival error distribution to maintain the matching function's adaptability to traffic fluctuations in different regions. Through the above mechanisms, this embodiment can output more executable priority ranking results in cross-site reservation scenarios.

[0153] Step 104: Receive the reservation request submitted by the user for the sorted target charging pile, and send a reservation instruction to the terminal corresponding to the target charging pile to lock the usage time of the target charging pile.

[0154] As an optional embodiment, in step 104, the reservation lock duration corresponding to the target charging pile is determined based on the vehicle's expected arrival time; a reservation instruction is sent to the charging pile terminal in the target parking lot to lock the usage rights of the target charging pile during the corresponding time period; after the lock is completed, the real-time status of the target charging pile is updated, and a reservation success reminder is sent to the user terminal. The reservation success reminder includes the location of the charging pile, the unlocking method, and the timeout reminder rules.

[0155] Specifically, after receiving a reservation request from a user for a sorted target charging station, the management platform first reads the estimated arrival time of the corresponding vehicle and, combined with the current road condition fluctuation range and parking lot traffic efficiency parameters, determines the reservation lock duration and lock start and end times corresponding to the target charging station. Subsequently, the management platform issues a reservation instruction to the charging station terminal in the target parking lot. This reservation instruction includes at least the target charging station identifier, the reservation time period, the user's charging identity identifier, and the lock expiration time, ensuring that the target charging station is only accessible to that user during the corresponding time period.

[0156] After the charging station terminal reports a successful lock, the management platform synchronously updates the real-time status information of the target charging station, changing it from an idle state to a reserved lock state, and records the lock timestamp and associated user identifier. After completing the status update, the management platform sends a reservation success reminder to the user terminal. This reminder includes the parking lot location of the target charging station, the unlocking method upon arrival, and timeout reminder rules. The timeout reminder rules may include the reservation effective time, the approaching timeout reminder time, and the timeout release conditions, used to remind the user to arrive on time and complete the charging unlocking process.

[0157] Furthermore, if a user fails to unlock the charging station within the locked time period, the management platform can automatically unlock the station and restore the target charging station to a reservationable state according to the timeout reminder rules, thereby reducing invalid occupancy and improving the turnover efficiency of charging stations.

[0158] Furthermore, after locking the usage period of the target charging station, it can be determined whether the target charging station is activated by the corresponding vehicle within the locked period. If the target charging station is not used within the locked period and the preset duration is exceeded, the lock on the target charging station is released. After the lock is released, the status of the target charging station is updated to idle, and an unlock reminder and rescheduling suggestion are sent to the user terminal.

[0159] Specifically, after locking the target charging station, the management platform enters the post-lock monitoring process. This monitoring process determines whether the target charging station has been actually used by the corresponding vehicle during the locked period, and automatically releases it if it has not been used and the preset time has elapsed. The goal of this embodiment is to reduce resource idleness caused by reservations and improve charging station turnover efficiency.

[0160] Upon successful locking, the management platform records the lock start time, lock failure time, user charging identity identifier, and target charging pile identifier, and establishes a locking session. Subsequently, the management platform receives session status reports from the target charging pile terminal according to a preset polling cycle or event-triggered method. The session status may include whether identity verification is complete, whether the charging gun is plugged in, whether charging has started, and the current power output status. Based on these statuses, the management platform determines whether the corresponding vehicle has activated the target charging pile.

[0161] The management platform determines whether a charging station is not enabled based on the locked session state. If no identity verification or charging start event is detected continuously during the locked period, the target charging station is determined to be not enabled by the corresponding vehicle. For short-term communication jitter scenarios, the management platform can set a continuous confirmation threshold, triggering the disabled conclusion only when no enable event is detected for multiple consecutive sampling cycles, thereby reducing the probability of false judgment.

[0162] The management platform sets a preset release duration and performs timeout checks. The preset release duration can be dynamically adjusted based on the congestion level of the charging station and historical arrival errors. If the current time exceeds the lock start time plus the preset release duration and the lock is still not activated, automatic unlocking is triggered. When unlocking, the management platform sends an unlock command to the charging pile terminal. After the terminal confirms the command, it returns the execution result. The management platform simultaneously closes the lock session and records the release reason.

[0163] After unlocking, the management platform updates the target charging station's status. This status update includes at least changing the target charging station from a reserved / locked state to an idle state, clearing the associated user's lock icon, resetting the available-for-reservation flag, and updating the timestamp. If there are queued candidate tasks, the management platform can re-add the charging station to the candidate set for the next round of allocation, thereby shortening the idle time.

[0164] The management platform sends unlock reminders and rebooking suggestions to user terminals. Unlock reminders may include the release time, reason for release, and an overview of currently available charging stations. Rebooking suggestions can be generated based on the user's current location, current traffic conditions, available charging station availability times, and cost levels, and may directly include a one-click rebooking entry to reduce the cost of repeated operations for users.

[0165] For example, a user successfully locks onto a target charging pile A in parking lot A at 7:05 PM. The system is set to lock out at 7:25 PM, with a preset release time of ten minutes. The management platform continuously receives the session status of pile A from 7:06 PM to 7:15 PM, but detects no successful identity verification or charging start events. By 7:15 PM, the current time has exceeded the lock start time plus the preset release time. The system determines that the timeout has occurred and sends an unlock command to pile A. After pile A successfully unlocks, the platform updates pile A's status to idle and sends a reminder to the user, including that the reservation has been released, the reason for release is that the timeout occurred, and suggestions to re-reserve at nearby available charging piles B and C. The user can then directly select pile B on the terminal to complete the second reservation.

[0166] In another example, if a user completes identity verification and starts charging within the preset release time, the management platform will lock the session status to enabled and terminate the release process, without performing automatic unlocking. Therefore, this embodiment can ensure users arrive at the site at a reasonable time while preventing charging stations from being occupied ineffectively for extended periods, thus improving cross-site resource scheduling efficiency and overall service availability.

[0167] Step 105: After the vehicles from multiple locations arrive at the target parking lot and start charging, obtain the charging process data during the charging process, and calculate the cost based on the charging payment rules and charging time information of the target area.

[0168] As an optional embodiment, in step 105, the charging monitoring unit installed in the charging pile collects charging time, charging amount, charging voltage, charging current and battery temperature in real time to generate corresponding charging process data; based on the corresponding billing period division, unit electricity price standard and service fee rules in the target area charging rules, and combined with the peak, normal or valley time type of the current charging period, a dynamic billing model is constructed; the charging process data is input into the dynamic billing model to calculate the basic charging fee corresponding to the current charging process; the parking billing rules corresponding to the target parking lot are obtained, and the parking fee is calculated based on the vehicle parking time.

[0169] It is understandable that step 105 can consist of five stages: data collection, data verification, dynamic billing, parking billing, and fee merging. After vehicles from multiple locations arrive at the target parking lot and complete identity unlocking, the charging monitoring unit in the charging pile begins to collect charging duration, charging amount, charging voltage, charging current, and battery temperature according to a preset sampling period, and writes the collection results into the charging process data stream with timestamps. After receiving the data stream, the management platform first performs data integrity checks and timing consistency checks, and then sends the data that passes the checks into the dynamic billing model.

[0170] In one embodiment, the input to the dynamic billing model includes at least charging process data, target area billing period division, unit electricity price standard, service fee rules, and current time period type information. The management platform automatically divides the charging process into multiple billing segments based on the billing period boundaries. Each billing segment corresponds to one of the following types: peak, normal, or off-peak hours, and calculates electricity and service fees separately for each segment. This processing method avoids discrepancies caused by charging across time periods being billed as a single time period, ensuring that the billing results are consistent with the regional rules.

[0171] In one embodiment, the calculation of the basic charging fee can be implemented by first estimating the segment's electricity consumption based on power and sampling time interval, then calculating the segment fee according to the unit electricity price and service rate corresponding to the time period to which the segment belongs, and finally summing all segment fees to obtain the basic charging fee. Parking fees are calculated separately by parking billing rules, which may include free time, segmented pricing, capped pricing, or nighttime differential pricing. The management platform calculates the parking duration based on the vehicle's entry time and the current settlement time, and obtains the parking fee according to the parking lot rules.

[0172] The management platform then merges the basic charging fee and parking fee to form the combined fee structure for this order. This combined fee structure can include separate electricity charges, service fees, parking fees, and a rule version identifier, used for subsequent payment settlement and dispute resolution. If the target area's payment rules require time-of-day details, the platform can simultaneously output peak, normal, and off-peak hour fee information.

[0173] For example, a user starts charging at 6:40 PM and finishes at 7:25 PM, charging a total of 24 kWh. The target area rules define 6:00 PM to 7:00 PM as peak hours and 7:00 PM to 10:00 PM as normal hours. The peak hour electricity price is 1.20 yuan per kWh, the normal hour price is 0.90 yuan per kWh, and the service fee is 0.25 yuan per kWh. After the system divides the time into segments, the charging amount from 6:40 PM to 7:00 PM is 9 kWh, and the charging amount from 7:00 PM to 7:25 PM is 15 kWh. Therefore, the electricity cost is 9 x 1.20 + 15 x 0.90, the service fee is 24 x 0.25, and the basic charging cost is 29.3 yuan. If the parking rules are free for the first 30 minutes, and 5 yuan for every 30 minutes thereafter, with a parking fee of 10 yuan for 90 minutes, the final combined cost is 39.3 yuan.

[0174] Regarding algorithm improvements, compared to the conventional static billing method, this embodiment adopts an automatic time-bound boundary segmentation and segmented integral billing mechanism to improve billing accuracy in cross-time-bound charging scenarios. It also employs an event-driven boundary correction mechanism to instantly redraw billing segments when charging interruptions, power surges, or time-bound switching occur, reducing accumulated errors. Furthermore, a robust filtering mechanism for sampled data is used to denoise or interpolate abnormal voltage and current points, reducing the impact of sensor jitter on cost calculation. Fourth, parking fees and charging fees are merged within the same order using standardized rules to avoid inconsistencies in settlement caused by separate calculations across multiple systems.

[0175] The management platform can also calibrate the parameters of the dynamic billing model based on historical order error feedback, such as rolling updates to the sampling period compensation coefficient and the time period switching delay compensation coefficient, to continuously improve the stability of cost calculation. Through the above implementation process, step 105 can achieve interpretable, traceable, and settleable cost calculation results under conditions of cross-regional rule differences and multi-time period charging.

[0176] Further optionally, in step 105, after acquiring the charging data during the charging process, anomalies can be judged based on preset thresholds for charging current and battery temperature; when overcurrent or overtemperature anomalies occur, an early warning signal is generated; the early warning signal is sent to the user terminal and / or management platform, and the target charging pile is controlled to suspend charging.

[0177] After acquiring charging process data in step 105, the management platform enters the charging safety monitoring sub-process. This sub-process is used to perform real-time anomaly detection of charging current and battery temperature, and to execute early warning and protection controls when overcurrent or overtemperature occurs, thereby reducing equipment and battery safety risks. This sub-process consists of five stages: data acquisition, threshold determination, alarm generation, linkage control, and recovery processing.

[0178] The charging station terminal reports charging current and battery temperature data according to a preset sampling period. The management platform performs timestamp alignment and validity verification on the reported data. To reduce false alarms caused by momentary fluctuations, the platform can perform moving average or median filtering on the most recent sampling points to obtain the current and temperature values ​​for judgment. The judgment values ​​are compared with preset thresholds to form an anomaly detection result.

[0179] In one embodiment, an anomaly detection can employ a combined strategy of dual thresholds and duration. The first criterion is that the current used for detection exceeds an overcurrent threshold. The second criterion is that the temperature used for detection exceeds an overtemperature threshold. If either condition is met and continues for a preset duration, the anomaly is determined to be valid. This strategy avoids false triggering caused by short-term spikes while ensuring that persistent anomalies can be identified promptly. For different vehicle models or different battery states, the thresholds can be set in stages according to regional safety rules and equipment capability configurations.

[0180] Once a valid anomaly is identified, the management platform generates an early warning signal and executes a tiered response. The early warning signal may include at least the anomaly type, anomaly level, time of occurrence, current current value, current temperature value, and associated charging pile identifier. When the early warning signal is sent to the user terminal, it can display risk warnings and handling suggestions. When sent to the management platform's operations and maintenance side, it can trigger a work order or remote review. If the anomaly level reaches the protection level, the platform issues a pause charging command to the target charging pile, and the terminal executes the charge stop and sends back the execution result.

[0181] After charging is paused, the system enters a recovery assessment process. The management platform continuously monitors the current and temperature drop. When the assessment parameters remain stable within a safe range for the preset recovery time, the platform can prompt the user to choose to continue charging or end charging according to the strategy. If the anomaly is not recovered or is repeatedly triggered, the platform maintains the charging stop status and recommends replacing the charging station, while recording the abnormal event for subsequent equipment health assessment and threshold calibration.

[0182] For example, after a user starts DC fast charging in the target parking lot, the system samples and reports data on a second-by-second basis. At the seventeenth minute of charging, the current value continuously exceeds the overcurrent threshold and persists for an extended period, triggering an overcurrent anomaly on the platform. The system immediately generates an overcurrent warning signal and pushes it to the user terminal and management platform, then issues a pause charging command to the charging pile. After the charging pile stops charging, it sends a successful status update, and the platform marks this session as a safe interruption and saves the anomaly log. Five minutes later, the parameters return to normal, and the system prompts the user to either restart charging or select a nearby backup charging pile. If the user chooses to continue, the platform performs a second safety self-check before resuming charging.

[0183] In another example, the battery temperature continued to rise and exceeded the over-temperature threshold during the later stages of charging. The system triggered an over-temperature warning and stopped charging, while simultaneously prompting the user to check the vehicle's cooling status. The management platform also added the charging station and corresponding session to a priority monitoring list. Through this process, the system can achieve closed-loop processing from identification to control when an anomaly occurs, reducing safety risks and ensuring the safety of users and equipment.

[0184] Step 106: After charging is completed, settlement is performed based on the user's bound payment method and fee calculation results, generating the corresponding charging record and updating the status information of the target charging pile.

[0185] As an optional embodiment, after charging is completed, in step 106, the management platform first receives the basic charging fee and parking fee output in step 105, and merges the fees according to the unified settlement statement structure to generate a combined charging and parking fee. The combined fee may include electricity fee items, service fee items, parking fee items, total amount, rule version number, and timestamp information to ensure traceability of subsequent settlements.

[0186] In one embodiment, the management platform performs payment method matching after generating the merged fee. The platform reads the user's bound payment method set and the target region's payment requirements, and filters for executable payment channels. If multiple executable channels exist, the platform can select the preferred channel based on channel availability, historical success rate, and settlement latency, while reserving alternative channels for rollback in case of failure. This process can improve the success rate of cross-regional settlement and reduce the probability of users needing to perform secondary operations.

[0187] Furthermore, the management platform initiates a deduction request to the priority payment channel. This request may include the order number, combined fee amount, user identification, and signature verification fields. If the deduction is successful, the platform updates the order status to "settled" and generates a payment voucher. If the deduction fails, the platform switches to an alternative channel and retryes according to a preset rollback strategy. If the retry still fails, the platform marks the order status as "pending" and sends a processing prompt to the user's terminal, allowing the user to manually confirm the payment method or supplement their payment capabilities.

[0188] In one embodiment, the management platform generates a charging record after settlement. The charging record may include at least the vehicle identifier, charging pile identifier, charging start time, charging end time, charging amount, basic charging fee, parking fee, combined fee, payment channel, payment result, exception event marker, and rule version identifier. The management platform writes the record to the order database and user profile database for subsequent queries, reconciliation, and scheduling optimization.

[0189] Furthermore, the management platform updates the target charging pile status information after the record is entered into the database. Status updates may include changing the target charging pile from occupied to available and reservable, clearing the current session lock flag, resetting the session billing context, and refreshing the list of available resources in the global status view. If the charging pile experiences an abnormal interruption or alarm during the session, the platform can mark it as limited availability and trigger an operation and maintenance review process.

[0190] For example, a user completes a charging session at a target parking lot. After the session ends, the system calculates the basic charging fee to be 31 yuan and the parking fee to be 10 yuan. The management platform combines these two fees to generate a combined fee of 41 yuan and matches it with an available payment channel already linked to the user. After the platform initiates the deduction and returns a success message, the system generates a payment voucher and complete order details, and then writes the session into the charging record. Once the record is entered into the database, the target charging station's status changes from occupied to available and reservable, and the user's terminal simultaneously receives a settlement success notification and fee details.

[0191] In another example, if the priority payment channel fails to deduct payment, the system automatically switches to the backup channel to complete the deduction and updates the order status to settled. The user only receives one successful result and does not need to initiate payment again. Through the above process, step 106 can achieve closed-loop processing of fee merging, cross-channel settlement, record accumulation, and status recovery after charging is completed, thereby improving settlement efficiency and resource turnover efficiency.

[0192] Optionally, in step 106, based on the payment requirements of the target area, a suitable payment channel is determined from the multiple payment methods already bound by the user; a unified charging settlement bill and charging record are generated and pushed to the user terminal; after the user completes the settlement, a payment voucher and charging report are sent to the user terminal; and the status of the target charging pile is updated to an available and reservable status.

[0193] For example, after charging is complete, the management platform first reads the payment requirements for the target area and filters available channels from the user's multiple linked payment methods. Filtering dimensions may include payment channel availability, currency consistency, authentication status, and current channel success rate. If multiple available channels exist, the platform selects the primary channel according to a preset priority strategy and retains backup channels for fallback in case of failure. Subsequently, the platform generates a unified settlement bill based on the basic charging fee and parking fee, simultaneously generating a charging record and pushing it to the user's terminal. After the user completes settlement, the platform sends a payment voucher and charging report to the user's terminal and updates the target charging station status to "available and bookable."

[0194] For example, after a user completes charging in the target area, the system receives a combined charge of 42 yuan. This user has linked accounts A, B, and C. The target area's rules require the use of channels that have completed cross-regional authentication; after screening, the platform determines account B as the primary channel. The platform pushes a unified bill to the user's device, and the user confirms and completes the payment. After successful payment, the platform sends back a payment voucher and a charging report, which includes charging duration, charging volume, breakdown of charges, and time period information. Simultaneously, the platform updates the target charging station's status from occupied to available and reservable, and releases session resources for subsequent users to reserve.

[0195] Furthermore, after the user completes the settlement, the charging time, charging amount, and user preference information during this charging process can be added as feedback data to the training samples of the cross-site charging pile resource matching model; the cross-site charging pile resource matching model can be iteratively updated based on the feedback data to improve the accuracy of subsequent resource matching; and the user's charging records and settlement information can be archived to form a corresponding user charging profile.

[0196] For example, after a user completes payment, the management platform generates feedback data for this session and feeds it back into the model for training. This feedback data includes at least charging duration, charging amount, and user preference information, such as preferred time periods, acceptable waiting time range, and cost preference intervals. The platform writes the feedback data into a training sample pool and incrementally trains or iteratively updates the cross-site resource matching model at preset intervals. After the update, the model can more accurately assess the compatibility between user needs and candidate resources in the next round of matching, thereby improving recommendation accuracy and booking success rate.

[0197] The management platform archives user charging records and settlement information to create user charging profiles. These profiles may include regional travel habits, charging frequency, average charging volume, cost sensitivity, and common payment preferences. User profiles can be used for subsequent personalized recommendations, abnormal order identification, and service strategy optimization. Continuous archiving and updating ensure that the profiles can track changes in user behavior.

[0198] Furthermore, statistical analysis can be performed on the charging pile operation data of each parking lot and vehicle charging data in multiple locations; idle charging piles and high-frequency charging areas can be identified; and resource scheduling suggestions can be output to parking lot operators based on the statistical analysis results to optimize the layout and operation and maintenance plan of charging piles.

[0199] For example, in a further embodiment, the management platform can also perform statistical analysis on parking lot operation data. The analysis can include charging pile utilization, time-of-day load, fault frequency, average waiting time, and vehicle charging distribution across multiple locations. Based on the statistical results, the platform identifies idle charging piles and high-frequency charging areas and outputs resource scheduling suggestions to the operator. These suggestions may include pile relocation, power structure optimization, peak-hour charging strategies, and maintenance inspection priorities. Through this closed loop, the system not only completes a single settlement but also continuously optimizes overall resource allocation and maintenance plans.

[0200] This application embodiment can realize intelligent matching and dynamic reservation scheduling of charging pile resources for vehicles in multiple locations across multiple parking lots, and obtain target charging pile allocation results that are compatible with vehicle charging needs, arrival time and regional charging rules. This reduces the waiting time and overall charging cost for users charging across locations, improves the utilization rate of charging pile resources and the efficiency of cross-regional charging management, brings users a more convenient and efficient charging experience, and reduces resource waste caused by charging pile mismatch, reservation conflicts and invalid occupancy.

[0201] The above describes a parking lot charging pile management method in the embodiments of this application. The parking lot charging pile management system (e.g., server) that implements the above parking lot charging pile management method will be described below.

[0202] See Figure 3 ,like Figure 3 The diagram shows a structural schematic of a parking lot charging pile management system 30. The parking lot charging pile management system 30 in this embodiment can achieve the functions described above. Figure 2 The steps of the parking lot charging pile management method executed in the corresponding embodiments are described above. The functions implemented by the parking lot charging pile management system 30 can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions, and the modules can be software and / or hardware. The parking lot charging pile management system 30 may include an input / output module 301 and a processing module 302. The functional implementation of the processing module 302 and the input / output module 301 can be found in [reference]. Figure 2 The operations performed in the corresponding embodiments will not be described in detail here. For example, the processing module 302 can be used to control the sending, receiving, and acquiring operations of the input / output module 301.

[0203] The input / output module 301 is configured to acquire vehicle information and charging demand information submitted by users in multiple locations, and to acquire real-time status information, parameter information, and corresponding payment rules of charging piles in multiple parking lots. The processing module 302 is configured to: construct a cross-site charging pile resource matching model based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and the corresponding area's payment rules; optimize charging pile resource utilization, user charging costs, and charging waiting time; filter candidate charging piles to obtain at least one target charging pile; construct a vehicle arrival time prediction model based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and charging pile reservation lock duration; predict the estimated arrival time of vehicles at each target parking lot; prioritize at least one target charging pile by combining the idle time of the target charging pile and the parking lot's traffic efficiency; receive reservation requests submitted by users for the ranked target charging piles; send reservation instructions to the corresponding terminals of the target charging piles to lock the usage time of the target charging piles; after vehicles in multiple locations arrive at the target parking lot and start charging, acquire charging process data during the charging process and calculate fees based on the payment rules of the target area and charging time information; after charging is completed, perform settlement processing based on the user's bound payment method and fee calculation results, generate corresponding charging records, and update the status information of the target charging piles.

[0204] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0205] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0206] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, apparatuses, or modules, and may be electrical, mechanical, or other forms.

[0207] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0208] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0209] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0210] The computer program product includes one or more computer instructions. When the computer program is loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state disk (SSD)).

[0211] The technical solutions provided in the embodiments of this application have been described in detail above. Specific examples have been used in the embodiments of this application to illustrate the principles and implementation methods of the embodiments of this application. The description of the above embodiments is only for the purpose of helping to understand the methods and core ideas of the embodiments of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the embodiments of this application. Therefore, the content of this specification should not be construed as a limitation on the embodiments of this application.

Claims

1. A method for managing charging piles in parking lots, characterized in that, The method, applicable to scenarios involving vehicle charging across multiple locations and sites, includes: It obtains vehicle information and charging demand information submitted by car users in multiple locations, and obtains real-time status information, parameter information, and corresponding regional charging and payment rules for charging piles in multiple parking lots. Based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and the charging and payment rules of the corresponding area, a cross-site charging pile resource matching model is constructed with the optimization objectives of charging pile resource utilization rate, user charging cost and charging waiting time. Candidate charging piles are screened to obtain at least one target charging pile. Based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and the reservation lock duration of charging stations, the estimated time for vehicles in multiple locations to arrive at each target parking lot is predicted. Then, the estimated time, the idle time of the target charging station, and the traffic efficiency of the parking lot are combined to prioritize at least one target charging station. The system receives reservation requests from users for sorted target charging stations and sends reservation instructions to the terminal corresponding to the target charging station to lock in the usage time of the target charging station. After vehicles from multiple locations arrive at the target parking lot and start charging, the charging process data is acquired, and the cost is calculated based on the charging payment rules and charging time information of the target area. After charging is completed, the system will process the settlement based on the user's linked payment method and the fee calculation result, generate the corresponding charging record, and update the status information of the target charging station.

2. The parking lot charging pile management method according to claim 1, characterized in that, The acquisition of vehicle information and charging demand information submitted by users in multiple locations includes: The system receives vehicle information and charging demand information submitted by the user terminal. The vehicle information includes the area information corresponding to the vehicle license plate, the vehicle battery capacity and the charging interface type. The charging demand information includes charging time, charging power, target parking area and charging budget. With user authorization, the authenticity of the vehicle information is verified by recognizing characters in the license plate image and comparing the recognized license plate information with the license plate database. Perform compatibility verification on the payment account submitted by the user and the payment rules of the target region; After verification, user information is bound to payment account to generate user charging identity identifier.

3. The parking lot charging pile management method according to claim 2, characterized in that, The compatibility verification of the user-submitted payment account with the payment rules of the target region includes: Obtain the payment method requirements, payment channel access rules, settlement currency requirements, and identity authentication requirements for the target region; Extract the account type, payment channel, settlement attributes, and authentication status of the payment account submitted by the user; The target area payment rules are used as a reference sequence, and the payment account attributes are used as a comparison sequence. A grey relational analysis method based on a multidimensional rule feature set is adopted to map the payment rules of the target area into a reference sequence, and to map the account type, channel, settlement attribute, and authentication status of the payment account into a comparison sequence. The correlation degree between the payment account attributes and the payment rules of the target area is calculated through a correlation coefficient matrix. The multidimensional rule feature set is constructed based on payment method requirements, payment channel access rules, settlement currency requirements, and identity authentication requirements. The compatibility score is determined based on the correlation, and when the compatibility score reaches a preset threshold, the payment account is deemed to have passed the compatibility verification.

4. The parking lot charging pile management method according to claim 1, characterized in that, Based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and corresponding regional payment rules, a cross-site charging pile resource matching model is constructed with the optimization objectives of charging pile resource utilization, user charging cost, and charging waiting time. This model is used to screen candidate charging piles, including: Based on vehicle information and charging demand information, combined with real-time status information of charging piles, site location information, parameter adaptation information and corresponding regional charging rules, a set of candidate charging piles is constructed. The candidate charging pile set is screened using charging pile resource utilization rate, total user charging cost and user waiting time as multi-objective optimization indicators; Based on the screening results, at least one target charging station that meets the current charging needs is identified.

5. The parking lot charging pile management method according to claim 4, characterized in that, The process of filtering the candidate charging pile set using charging pile resource utilization rate, total user charging cost, and user waiting time as multi-objective optimization indicators includes: Construct a multi-objective optimization function with the goals of maximizing the utilization rate of charging pile resources across the entire region, minimizing user charging costs, and minimizing charging waiting time; An improved genetic algorithm is used to solve the multi-objective optimization function. The improved genetic algorithm adopts a multi-segment encoded chromosome structure for cross-site charging reservation scenarios. The multi-segment coded chromosome includes at least candidate parking lot identifiers, charging pile identifiers, reservation time periods, path travel cost parameters, and interface adaptation status parameters. During the population initialization phase, a feasible solution set is generated based on charging interface matching constraints, charging pile idle time period constraints, and user charging budget constraints. During the fitness evaluation phase, charging costs, parking costs, cross-site travel costs, expected waiting time, and reservation conflict penalties are jointly used to construct a fitness function. Furthermore, an elite retention mechanism and an infeasible solution repair mechanism are introduced during the genetic iteration process to repair or remove individuals that violate interface adaptation constraints, reservation time period constraints, or budget constraints. During the solution process, the crossover probability and mutation probability are adaptively adjusted according to the number of iterations to reduce the impact of local optima on the screening results.

6. The parking lot charging pile management method according to claim 5, characterized in that, The process of filtering the candidate charging pile set also includes: Based on the compatibility of charging interfaces, the distance between vehicles and parking lots, and the degree of charging compatibility, an evaluation factor for the compatibility of charging piles is constructed. The corresponding compatibility score is obtained by weighting each candidate charging pile; Based on the fit score and the solution results of the multi-objective optimization function, at least one target charging pile is determined.

7. The parking lot charging pile management method according to claim 1, characterized in that, The method predicts the estimated arrival time of vehicles from multiple locations at each target parking lot based on the vehicle's current location, real-time traffic conditions, historical traffic data, and the reservation lock duration for charging stations. This includes: At least one of Transformer, Informer, Autoformer and FEDformer is selected as the basic temporal prediction network; Based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and the reservation and lock-in duration of charging piles, a multi-source time-series feature matrix is ​​constructed, consisting of location trajectory subsequence, traffic condition fluctuation subsequence, historical traffic duration subsequence, and reservation and lock-in duration subsequence. The multi-source temporal feature matrix is ​​subjected to position encoding and time encoding processing; The multi-source temporal feature matrix after encoding is weighted and optimized by combining parking lot access node weight parameters, cross-site path length parameters, and reservation time period correction factors. The access node weight parameters are set according to the road segment congestion frequency, the path length parameters are converted according to the actual mileage, and the reservation time period correction factor is dynamically adjusted according to the remaining reservation lock time to obtain the weighted optimized feature matrix. The weighted optimized feature matrix is ​​input into the basic time-series prediction network, and the estimated time for vehicles from multiple locations to arrive at each target parking lot is output through time-series dependency modeling and feature mapping.

8. The parking lot charging pile management method according to claim 7, characterized in that, The process of prioritizing at least one target charging station by combining the estimated time, the idle time of the target charging station, and the parking lot traffic efficiency includes: Obtain the available time periods, remaining reservation time, and traffic efficiency information of each target charging station; The Gaussian kernel time-series matching function is used to calculate the time-series matching degree between the estimated time and the idle time period of each target charging pile. The parking lot traffic efficiency, time-series matching degree and charging cost parameters are combined and the multi-dimensional weighted TOPSIS algorithm is used to generate the priority evaluation value corresponding to each target charging pile. At least one target charging pile is prioritized according to the priority evaluation value, and the ranking result is pushed to the user terminal.

9. The parking lot charging pile management method according to claim 1, characterized in that, After vehicles from multiple locations arrive at the target parking lot and begin charging, the system acquires charging process data and calculates costs based on the target area's charging rules and charging time information, including: The charging monitoring unit installed in the charging pile collects charging time, charging amount, charging voltage, charging current and battery temperature in real time, and generates corresponding charging process data. Based on the corresponding billing period division, unit electricity price standard and service fee rules in the target area charging rules, and combined with the peak, normal or valley time type of the current charging period, a dynamic billing model is constructed. Input the charging process data into the dynamic billing model to calculate the basic charging cost corresponding to the current charging process. Obtain the parking fee rules for the target parking lot and calculate the parking fee based on the vehicle's parking duration.

10. A parking lot charging pile management system, characterized in that, The system includes: The input / output module is configured to obtain vehicle information and charging demand information submitted by users in multiple locations, and to obtain real-time status information, parameter information, and corresponding payment rules of charging piles in multiple parking lots. The processing module is configured to: construct a cross-site charging pile resource matching model based on the vehicle information, charging demand information, real-time status information of charging piles, parameter information, and the corresponding area's payment rules; optimize charging pile resource utilization, user charging costs, and charging waiting time; filter candidate charging piles to obtain at least one target charging pile; construct a vehicle arrival time prediction model based on the current location of vehicles in multiple locations, real-time traffic conditions, historical traffic data, and charging pile reservation lock duration; predict the estimated arrival time of vehicles at each target parking lot; and prioritize at least one target charging pile by combining the idle time of the target charging pile and the parking lot's traffic efficiency; receive reservation requests submitted by users for the ranked target charging piles; send reservation instructions to the corresponding terminals of the target charging piles to lock the usage time of the target charging piles; after vehicles in multiple locations arrive at the target parking lot and start charging, acquire charging process data during the charging process and calculate the cost based on the payment rules of the target area and the charging time information; after charging is completed, perform settlement processing based on the user's bound payment method and the cost calculation result, generate the corresponding charging record, and update the status information of the target charging pile.

Images