Electric vehicle cluster grid-connected load intelligent processing method based on space-time correlation features

Abstract

Embodiments of the present disclosure disclose a method for intelligent processing of grid-connected load of electric vehicle cluster based on space-time correlation features. A specific embodiment of the method includes: obtaining charging space-time correlation information and vehicle flow space-time correlation information corresponding to each charging station; for each charging station, using a station agent, performing the following steps: generating charging space-time prediction information; comparing the charging space-time prediction information with the charging space-time correlation information; calling a set of historical charging video clips; generating fuel vehicle occupancy information; adjusting the charging space-time correlation information to obtain charging space-time adjustment information; generating load supply information; and sending the load supply information to a regional agent for power supply of each charging station. This embodiment can accurately and efficiently determine the load power supply situation of the target urban public area by considering various factors affecting the load of the charging station based on the agent, thereby avoiding waste of power resources.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and more specifically to a method, apparatus, electronic device, and computer-readable medium for intelligent processing of grid-connected loads of electric vehicle clusters based on spatiotemporal correlation characteristics. Background Technology

[0002] Currently, with the continuous development of new energy sources, electric vehicles are increasingly used as daily transportation. How to achieve load management of charging stations has become a key technological direction. For load management of charging stations in public areas of a target city, the common approach is to investigate load changes at charging stations in public areas of the target city and artificially generate load management strategies for subsequent charging stations to ensure power supply to them.

[0003] However, when using the above method, the following technical problems often arise: The load changes of charging stations in public areas of the target city are affected by many factors, which makes it impossible to fully determine the load usage of charging stations. As a result, the formulation of load management strategies can lead to insufficient power supply at charging stations during peak periods.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure propose a method, apparatus, electronic device, and computer-readable medium for intelligent processing of grid-connected loads of electric vehicle clusters based on spatiotemporal correlation characteristics, in order to solve one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide an intelligent load processing method for electric vehicle cluster grid connection based on spatiotemporal correlation features, including: acquiring spatiotemporal correlation information of charging and spatiotemporal correlation information of vehicle flow for each charging station in a target city's public area during a target peak time period; for each charging station, using a corresponding deployed site agent, performing the following generation steps: generating spatiotemporal prediction information of charging for the charging station based on the spatiotemporal correlation information of vehicle flow; comparing the spatiotemporal prediction information of charging with the spatiotemporal correlation information of charging to query a target time period set in which the information difference meets the target difference condition; retrieving a set of historical charging video clips corresponding to the target time period set; generating occupancy information of fuel vehicles based on the set of historical charging video clips; adjusting the spatiotemporal correlation information of charging based on the occupancy information of fuel vehicles to obtain spatiotemporal adjustment information of charging; generating load supply information of the charging station based on the spatiotemporal adjustment information of charging; and sending the load supply information to the regional agent corresponding to the public area of ​​the target city to supply power to each charging station during future peak time periods.

[0008] Secondly, some embodiments of this disclosure provide an intelligent load processing device for electric vehicle clusters based on spatiotemporal correlation features, comprising: an acquisition unit configured to acquire spatiotemporal correlation information of charging and spatiotemporal correlation information of vehicle flow corresponding to each charging station in a target city's public area during a target peak time period; an execution unit configured to, for each charging station, utilize a corresponding deployed site agent to execute the following generation steps: generating spatiotemporal prediction information of charging corresponding to the charging station based on the vehicle flow spatiotemporal correlation information; comparing the spatiotemporal prediction information of charging with the spatiotemporal correlation information of charging to query a target time period set in which the information difference meets the target difference condition; retrieving a set of historical charging video clips corresponding to the target time period set; generating occupancy information of fuel vehicles based on the set of historical charging video clips; adjusting the spatiotemporal correlation information of charging based on the occupancy information of fuel vehicles to obtain spatiotemporal adjustment information of charging; and generating load supply information corresponding to the charging station based on the spatiotemporal adjustment information of charging; and a sending unit configured to send the load supply information to a regional agent corresponding to the target city's public area to supply power to each charging station during future peak time periods.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, such that when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any implementation of the first aspect.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method as described in any implementation of the first aspect.

[0011] The above-described embodiments of this disclosure have the following beneficial effects: By utilizing the intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics, based on some embodiments of this disclosure, and considering various factors affecting the load of charging stations, the load power supply situation in public areas of a target city can be accurately and efficiently determined, avoiding waste of power resources. Specifically, the reason for insufficient power supply during peak hours is that the load changes of charging stations in public areas of the target city are affected by many factors, making it impossible to fully determine the load usage of charging stations. This leads to insufficient power supply during peak hours when manually formulating load management strategies. Therefore, the intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics of some embodiments of this disclosure first obtains the spatiotemporal correlation information of charging and traffic flow for each charging station in the public area of ​​the target city during the target peak time period. Here, by obtaining the spatiotemporal correlation information of charging and traffic flow during the target peak time period, it is possible to subsequently determine the corresponding load supply information that meets the power demand of the charging station from the dimensions of actual charging and traffic flow. Then, for each charging station, the following steps are performed using the corresponding deployed site agent: First, based on the spatiotemporal correlation information of the traffic flow, accurate spatiotemporal prediction information for charging at the charging station can be generated to determine the possible charging status of the charging station based on the traffic flow. Second, the spatiotemporal prediction information for charging is compared with the spatiotemporal correlation information to query the target time period set where the information difference meets the target difference condition. Here, by comparing charging information differences, the time periods with significant differences between charging information and actual charging information under the traffic flow dimension can be preliminarily and accurately determined to determine whether there is a charging impact based on these time periods, thus improving the accuracy of the charging station's load prediction. Third, the historical charging video clip set corresponding to the target time period set is retrieved to serve as the data basis for judging the charging impact, facilitating the subsequent determination of the occupancy status of fuel vehicles. Fourth, fuel vehicle occupancy information can be accurately generated based on the historical charging video clip set. Here, by determining the fuel vehicle occupancy information, it is determined whether there is a serious fuel vehicle occupancy problem at the charging station, resulting in the available load of the charging station being lower than the actual usage. During peak hours, considering only actual usage, some charging stations may not be fully utilized due to occupancy by gasoline-powered vehicles, leading to insufficient power supply. The fifth step involves adjusting the charging time-space correlation information based on the gasoline-powered vehicle occupancy information to accurately generate charging time-space adjustment information while taking into account the gasoline-powered vehicle occupancy issue. The sixth step involves generating the load supply information corresponding to the charging station based on the charging time-space adjustment information.Here, based on accurate charging time and space adjustment information, load supply can be accurately predicted. Finally, the load supply information is sent to the regional agents corresponding to the public areas of the target city. This ensures that power supply to each charging station during future peak hours can fully utilize their charging capacity and avoid power shortages during peak periods. In summary, by considering various factors affecting charging station load and utilizing intelligent agents, the load supply situation in the public areas of the target city can be accurately and efficiently determined, avoiding waste of electrical resources. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics according to this disclosure; Figure 2 This is a schematic diagram of the structure of some embodiments of the intelligent load processing device for electric vehicle clusters based on spatiotemporal correlation characteristics according to this disclosure; Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] refer to Figure 1 The diagram illustrates a flow 100 of some embodiments of the intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics according to this disclosure. This intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics includes the following steps: Step 101: Obtain the spatiotemporal correlation information of charging and traffic flow for each charging station in the public area of ​​the target city during the target peak time period.

[0021] In some embodiments, the executing entity (e.g., an electronic device) of the above-described intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics can acquire, via wired or wireless means, the spatiotemporal correlation information of charging and traffic flow for each charging station in the target urban public area during the target peak time period. The target urban public area can be the urban public area for which charging station load supply forecasting is performed. Urban public areas typically refer to areas within the urban planning area that are open to the public at all times and unconditionally available for free use and activity by members of society. Depending on the area size of the urban public area, there can be at least one charging station. Each charging station has multiple deployed charging piles. These charging piles can support charging at different power levels. The target peak time period can be a peak time period preceding the current time. For example, the target peak time period can be any of the peak time periods within a month preceding the current time. For example, if the current time is May 1st, the corresponding target peak time period could be the peak charging time periods for each day of April. The spatiotemporal correlation information of charging can be correlation data in both the time and spatial dimensions related to the charging data corresponding to the charging station. In practice, the spatiotemporal correlation information of charging can be the charging status of the charging station during each time period. In practice, charging status can be reflected through charging load. That is, the spatiotemporal correlation information of charging can be time-series data. Each time period has a corresponding charging status. Specifically, the charging status includes the overall charging load. The time series here can be various time series within the target peak time period. Similarly, the spatiotemporal correlation information of traffic flow can be correlated data in both time and spatial dimensions related to the traffic flow data corresponding to the charging station. In practice, the spatiotemporal correlation information of traffic flow can be the traffic flow situation on roads near the charging station in each time period. In practice, traffic flow situation can be reflected through the number of vehicles per unit time. That is, the spatiotemporal correlation information of traffic flow can be time-series data. Each time period has a corresponding charging status. Specifically, the traffic flow situation includes the sum of the overall traffic flow. The time series here can be various time series within the target peak time period.

[0022] It should be noted that the spatiotemporal correlation information of charging can be obtained from real-time statistics of the relevant power system. The spatiotemporal correlation information of traffic flow can be obtained from the identification of relevant road camera devices.

[0023] Step 102: For each charging station, utilize the corresponding deployed site agent to perform the generation steps: Step 1021: Generate charging spatiotemporal prediction information corresponding to the charging station based on the traffic flow spatiotemporal correlation information.

[0024] In some embodiments, the aforementioned executing entity can generate charging spatiotemporal prediction information corresponding to the charging station based on the traffic flow spatiotemporal correlation information. The charging spatiotemporal prediction information can be the load status of the charging station during the predicted target peak time period. That is, the charging spatiotemporal prediction information is the load data within each time period of the target peak time period. The charging spatiotemporal prediction information can be in time series form. There is a time correspondence between each time period corresponding to the charging spatiotemporal prediction information and each time period corresponding to the charging spatiotemporal correlation information. The site intelligent agent can be an intelligent agent deployed at the charging station for intelligent processing of various events within the station. In practice, each station has a deployed site intelligent agent. The site intelligent agent is deployed with a multimodal large model for processing various events. The site intelligent agent can call relevant skill files to achieve intelligent processing of various events. For example, various events may include: safety and fire warning events, equipment operation and maintenance and fault handling events, station operation and order management events, macro-decision and owner service events, and peak period handling events. In specific scenarios, the intelligent agents within charging stations can adopt a "cloud-edge-device" collaborative architecture to achieve efficient, real-time data processing and business closed-loop. The cloud control center, deployed on a public or private cloud, is responsible for global policy management, traffic scheduling, big data analysis, and non-real-time business (e.g., global load forecasting, cross-regional resource scheduling, financial reconciliation). Edge computing nodes are deployed locally at the charging station or on the distribution network side (e.g., edge gateways, microservers). They handle terminal access and data preprocessing, running algorithms such as AI analysis, battery safety prediction, and orderly charging control to achieve "on-site perception of distribution network status and on-site regulation of charging power," significantly reducing cloud pressure and achieving millisecond-level response. The terminal device layer is deployed on the charging pile itself or on IoT hardware such as station cameras and ground locks, responsible for basic status collection, protocol integration, and lightweight command execution. Here, the intelligent agent within the station can be a distributed system containing five core technology layers. The core technology layers include: the inference engine layer, the memory system layer, the tool protocol layer, the orchestration and state machine layer, and the guardrail and observation layer. The inference engine layer, centered on a multimodal large model, is responsible for understanding context, recognizing intent, and planning. It supports model routing (using small models for simple tasks and large models for complex inference) and enforces structured output (e.g., JSON) for interface with backend devices. The memory system layer includes: working memory (current session), short-term memory (recent interactions and device state cache), and long-term memory (vector storage of user charging preferences and historical fault databases). The tool protocol layer encapsulates the charging station's hardware APIs and database queries into tools recognizable by the large model through standardized protocols (e.g., MCP), supporting autonomous invocation by intelligent agents. The orchestration and state machine layer can employ graph structure orchestration (e.g., DAG) to restrict the autonomy of intelligent agents within defined business boundaries and set checkpoints to prevent interruptions of long-running tasks.The guardrail and observation layer incorporate manual confirmation mechanisms for sensitive operations such as power outages and billing, and provide end-to-end tracking for later optimization. The multimodal large model can be a large model with a hybrid expert model (MoE) architecture. The multimodal large model can be trained on multiple training datasets corresponding to multiple events. Specifically, the training process can use 30% basic power plant knowledge and 70% multiple training datasets. The multimodal large model is trained periodically.

[0025] As an example, firstly, the aforementioned execution entity can input the spatiotemporal correlation information of traffic flow into the prediction prompt word template to obtain charging prediction prompt words. Then, the charging prediction prompt words are input into the multimodal large model in the site intelligent agent to obtain charging spatiotemporal prediction information.

[0026] As another example, firstly, the aforementioned executing entity can obtain a charging spatiotemporal prediction skill file (based on the sequence of prediction steps outlined in the prediction process). Then, based on the charging spatiotemporal prediction skill file and the spatiotemporal correlation information of traffic flow, the station intelligent agent generates charging spatiotemporal prediction information.

[0027] In some optional implementations of certain embodiments, the executing entity may generate charging spatiotemporal prediction information corresponding to the charging station based on the traffic flow spatiotemporal correlation information, including the following steps: The first step is to obtain the road information set corresponding to the charging station. This road information can include the names of roads near the charging station.

[0028] The second step, for each piece of road information, is to perform the following first processing step: Sub-step 1 involves extracting the spatiotemporal sub-information of traffic flow corresponding to the road information from the spatiotemporal correlation information of traffic flow. The spatiotemporal sub-information of traffic flow can be the traffic flow sequence of the road corresponding to the road information during the target peak time period.

[0029] Sub-step 2 involves generating charging spatiotemporal prediction sub-information based on the traffic flow spatiotemporal sub-information. This charging spatiotemporal prediction sub-information can be a sequence of the number of vehicles traveling on the roads corresponding to the road information and entering charging stations during the target peak time period. There is a one-to-one correspondence between the number of vehicles in the vehicle number sequence and the time periods in the target peak time period. These time periods constitute the target peak time period. Each time period can be divided based on historical experience.

[0030] As an example, firstly, the aforementioned executing entity can substitute the spatiotemporal sub-information of traffic flow into the prediction prompt word template to obtain charging prediction sub-prompt words. Then, the charging prediction sub-prompt words are input into the multimodal large model in the site agent to obtain charging spatiotemporal prediction sub-information.

[0031] As another example, firstly, the aforementioned executing entity can obtain a charging spatiotemporal prediction skill file (based on the sequence of prediction steps outlined in the prediction process). Then, based on the charging spatiotemporal prediction skill file and the spatiotemporal correlation sub-information of traffic flow, the station intelligent agent generates charging spatiotemporal prediction sub-information.

[0032] Sub-step 3 involves obtaining the charging conversion coefficient corresponding to the road information. The charging conversion coefficient can be the degree of deviation between the predicted number of charges based on road traffic flow and the actual number of charges. A higher charging conversion coefficient indicates a higher degree of deviation. Each piece of road information has a unique charging conversion coefficient. The charging conversion coefficient for each piece of road information is updated periodically. The charging conversion coefficient can be obtained by averaging a series of conversion difference values ​​over a certain period.

[0033] The third step is to generate charging spatiotemporal prediction information based on the obtained charging spatiotemporal prediction sub-information set and charging conversion coefficient set.

[0034] As an example, firstly, for each charging spatiotemporal prediction information, each value in the corresponding numerical sequence is multiplied by the corresponding charging conversion coefficient to generate a multiplication result sequence. Then, the multiplication result sequences are added together according to their sequence positions to obtain an added numerical sequence, which serves as the charging spatiotemporal prediction information.

[0035] In addressing the technical problems mentioned in the background section, when employing technical solutions to solve the problems encountered in the application scenario—where charging stations are located in peripheral areas (e.g., suburban areas with numerous residential buildings) and the target audience is relatively fixed—the following technical issues often arise: Since the target audience for charging stations in peripheral areas is already established, the current situation is not considered in the relevant spatiotemporal predictions, leading to significant deviations in the predicted spatiotemporal information. Considering the following requirements for this application scenario: a stable target audience in peripheral areas, we have decided to adopt the following solution: In some optional implementations of certain embodiments, the executing entity may generate charging spatiotemporal prediction information corresponding to the charging station based on the traffic flow spatiotemporal correlation information, including the following steps: The first step is to obtain the urban location information and value information corresponding to the charging station. The urban location information can be the deployment location of the charging station in a public area of ​​the target city. The value information can be the price of the charging piles at the charging station at different time periods.

[0036] Here, the price and target audience vary depending on the location of the charging station. For example, charging stations in the suburbs are often cheaper, targeting nearby residents, taxi drivers, and ride-hailing users. Charging stations in the city center are often more expensive, targeting office workers or nearby residents. However, the target audience for office workers is uncertain; given the higher price, office workers might choose a cheaper charging station instead.

[0037] The second step is to determine whether to perform basic traffic flow information prediction based on the urban area location information and the value information. This basic traffic flow information prediction can be a prediction of the number of frequent customers who regularly charge at charging stations.

[0038] As an example, if the urban location information is in the suburbs and the value information represents a low price, it is determined to perform basic traffic flow information prediction. If the urban location information is in the non-suburbs and the value information represents a high price, it is determined not to perform basic traffic flow information prediction (because the users of charging stations in the non-suburbs are often not fixed, so basic traffic flow information prediction is not performed).

[0039] The third step involves retrieving historical charging videos for each peak time period in response to the determination of basic traffic flow information prediction. Each peak time period has a corresponding historical charging video. A peak time period can be the time period of peak charging demand that occurs before the target peak time period. Historical charging videos can be videos of charging occurring during those peak time periods.

[0040] The fourth step involves extracting license plate information from each historical charging video and time-annotating each extracted license plate to obtain an annotated information sequence set. The license plate information can be the license plate of the charging vehicle. Time annotation involves adding the time the license plate information corresponds to the time the vehicle appeared at the charging station. The annotated information can include: the license plate information and the corresponding charging station appearance time.

[0041] As an example, a multimodal large model can be used to extract license plate information from each of the historical charging videos and to time-annotate each extracted license plate information to obtain a set of annotated information sequences.

[0042] The fifth step is to deduplicate the annotation information in each annotation information sequence in the annotation information sequence set to obtain a deduplicated information sequence set. Deduplication can be achieved by removing annotation information with identical license plate information from the annotation information sequences.

[0043] Step 6: Based on the deduplicated information sequence set, generate basic traffic flow prediction information and basic traffic flow fluctuation information. The basic traffic flow prediction information can be the predicted basic traffic flow situation at charging stations, that is, the predicted number of vehicles frequently charging at charging stations. The basic traffic flow fluctuation information can be a fluctuation range of the number of frequently charging vehicles.

[0044] As an example, firstly, the aforementioned execution entity can filter out vehicle information that appears more frequently than a first frequency in each deduplicated information sequence set to obtain a first vehicle information set. Then, the number of vehicles corresponding to the first vehicle information set is determined as the basic traffic flow prediction information. Next, vehicle information that appears more frequently than a second frequency in each deduplicated information sequence set is filtered out to obtain a second vehicle information set. The second frequency is higher than the first frequency. Then, vehicle information that appears more frequently than a third frequency in each deduplicated information sequence set is filtered out to obtain a third vehicle information set. The third frequency is lower than the first frequency. The number of vehicles corresponding to the third vehicle information set and the number of vehicles corresponding to the second vehicle information set are combined to generate a vehicle number variation range, which serves as the basic traffic flow fluctuation information.

[0045] Step 7: The basic traffic flow prediction information is sent to the vehicle management system corresponding to the public area of ​​the target city for verification. The vehicle management system verifies each vehicle corresponding to the basic traffic flow prediction information to determine whether it is a taxi, ride-hailing vehicle, or a vehicle belonging to the user residing nearby. The verification result indicates whether each vehicle corresponding to the basic traffic flow prediction information is a taxi, ride-hailing vehicle, or a vehicle belonging to the user residing nearby. The vehicle management system can be a system that manages vehicles operating within the public area of ​​the target city.

[0046] Step 8: In response to the obtained traffic flow verification result indicating no errors, actual traffic flow spatiotemporal correlation information and basic charging spatiotemporal correlation information are generated based on the traffic flow spatiotemporal correlation information and the basic traffic flow prediction information. The actual traffic flow spatiotemporal correlation information can be the traffic flow spatiotemporal correlation information after removing the basic vehicle prediction information. That is, vehicle information in the basic traffic flow prediction information is removed from the time series corresponding to the traffic flow spatiotemporal correlation information, resulting in the removed time series, which serves as the actual traffic flow spatiotemporal correlation information. The basic charging spatiotemporal correlation information can be the charging spatiotemporal correlation information corresponding to each vehicle during charging, based on the basic traffic flow prediction information. In other words, the basic charging spatiotemporal correlation information is the charging spatiotemporal correlation information for vehicles that are frequently charging.

[0047] As an example, firstly, the vehicle information set corresponding to the basic traffic flow prediction information is determined. Then, the spatiotemporal correlation information corresponding to the vehicle information set is removed from the traffic flow spatiotemporal correlation information to obtain the actual traffic flow spatiotemporal correlation information. Finally, the charging spatiotemporal correlation sub-information corresponding to the vehicle information set is extracted from the charging spatiotemporal correlation information to serve as the basic charging spatiotemporal correlation information.

[0048] Step nine: Based on the actual traffic flow spatiotemporal correlation information, generate charging spatiotemporal prediction information corresponding to the charging station. This charging spatiotemporal prediction information can be a prediction of load conditions during future peak periods. That is, the charging spatiotemporal prediction information can be a supply-load sequence during future peak periods. The future peak period can be the period following a previous peak period. The charging spatiotemporal prediction information can be in time series format.

[0049] As an example, the aforementioned execution entity can use a multimodal large model to predict future time series information based on the actual traffic flow spatiotemporal correlation information, and generate the charging spatiotemporal prediction information corresponding to the charging station.

[0050] In some optional implementations of certain embodiments, comparing the charging spatiotemporal prediction information with the charging spatiotemporal association information to query a target time period set whose information differences satisfy the target difference condition includes the following steps: The first step is to adjust the basic charging spatiotemporal correlation information based on the existing information to obtain new charging spatiotemporal correlation information. This new information can be in the form of a time-series sequence. The new charging spatiotemporal correlation information can characterize the charging status of newly added charging vehicles in each time period.

[0051] As an example, the aforementioned execution entity can remove the basic charging spatiotemporal correlation information from the charging spatiotemporal correlation information to obtain the newly added charging spatiotemporal correlation information. Specifically, the newly added charging spatiotemporal correlation information can be obtained by subtracting the sequence elements of the time sequence corresponding to the charging spatiotemporal correlation information from the time sequence corresponding to the basic charging spatiotemporal correlation information.

[0052] The second step involves using the newly added charging spatiotemporal correlation information as the charging spatiotemporal correlation information, and comparing the charging spatiotemporal prediction information with the charging spatiotemporal correlation information to query the target time period set where the information differences meet the target difference conditions. Details will not be elaborated further.

[0053] The above content, as another inventive point of this disclosure, solves another technical problem: "Charging stations located in peripheral areas have a certain audience base, so the current situation is not considered in the relevant charging spatiotemporal predictions, resulting in a large deviation in the predicted charging spatiotemporal information, leading to a large deviation in the subsequent power supply for each charging station." Based on this, this disclosure first determines the urban location and value information corresponding to the charging station to determine whether to perform basic traffic flow prediction (i.e., determine whether the charging station is located in a peripheral area and has a stable audience). Second, after determining to perform basic traffic flow prediction, basic traffic flow prediction information and basic traffic flow fluctuation information are generated based on historical charging videos during various peak time periods. Then, based on the vehicle management system, the traffic flow-related information is verified. Next, after confirming that the verification is correct, actual traffic flow spatiotemporal correlation information and basic charging spatiotemporal correlation information can be accurately generated based on the traffic flow spatiotemporal correlation information and the basic traffic flow prediction information, resulting in more accurate charging spatiotemporal prediction information. On this basis, through the basic charging spatiotemporal correlation information, new charging spatiotemporal correlation information can be accurately generated to query the target time period set where the information difference meets the target difference condition.

[0054] In addressing the technical problems mentioned in the background section, the application scenario of predicting charging time and space when the occupied area of ​​a charging station is large or small often presents the following challenges: When the occupied area is large, there is a high incidence of gasoline-powered vehicles. When the occupied area is small, significant fluctuations in traffic flow lead to unstable and inaccurate charging predictions. Considering the specific requirements of this application scenario—generating charging time and space prediction information based on occupied areas—we have decided to adopt the following solution: In some optional implementations of certain embodiments, the executing entity may generate charging spatiotemporal prediction information corresponding to the charging station based on the traffic flow spatiotemporal correlation information, including the following steps: The first step is to determine the occupancy zone corresponding to the charging station based on the urban area location information and the corresponding handling plan for gasoline vehicles at the charging station. The gasoline vehicle handling plan can be a scheme for handling gasoline vehicles currently occupying parking spaces corresponding to charging piles at the charging station. The occupancy zone can be the size of the occupancy zone for each charging pile within the charging station.

[0055] As an example, the aforementioned implementing entity can query the urban area location information and the corresponding occupied intervals of the fuel vehicle processing plan by querying the relationship table. The relationship table represents the mapping relationship between the type of urban area location information, the plan identifier of the fuel vehicle processing plan, and the occupied interval.

[0056] The second step involves generating a first processing method that represents the spatiotemporal adjustment of charging based on the occupancy information of gasoline vehicles, in response to the maximum value of the occupancy interval being not less than the target value. Here, the maximum value of the occupancy interval can be the largest value within that interval. The target value can be a threshold used to measure whether a large number of gasoline vehicles are occupying the parking spaces corresponding to the charging piles. In practice, the target value can be set based on historical experience. The first processing method can represent a method that corrects the spatiotemporal correlation information of charging by determining the occupancy information of gasoline vehicles and then predicts the load during future peak periods.

[0057] The third step, in response to the first processing method, is to generate charging spatiotemporal prediction information corresponding to the charging station based on the traffic flow spatiotemporal correlation information. The specific implementation method will not be elaborated further.

[0058] Optionally, the steps also include: The first step involves generating a second processing method to represent future load prediction based on traffic flow and charging spatiotemporal information, in response to the maximum value of the occupied interval being less than the target value. This second processing method can be either predicting future load supply based on traffic flow or predicting future load supply based on charging spatiotemporal information. When the maximum value of the occupied interval is less than the target value, it indicates that there are relatively few gasoline vehicles occupying the corresponding charging spaces. Such charging stations are typically located in busy areas, and their corresponding processing solutions are usually more advanced.

[0059] The second step, in response to the second processing method, involves generating future traffic flow spatiotemporal correlation information for future peak periods based on the traffic flow spatiotemporal correlation information. Here, the future peak period can be a peak period following the current peak period. The future traffic flow spatiotemporal correlation information can represent the changes in traffic flow during the future peak period. In practice, the future traffic flow spatiotemporal correlation information can be in the form of a time-series sequence. The future peak period can include multiple future peak sub-periods. The future traffic flow spatiotemporal correlation information can include multiple future traffic flow information. Each future traffic flow information corresponds to a future peak sub-period.

[0060] As an example, based on the spatiotemporal correlation information of traffic flow, the temporal prediction capability corresponding to the multimodal large model is used to generate the spatiotemporal correlation information of future traffic flow during future peak periods.

[0061] The third step involves generating first future charging spatiotemporal correlation information for the future peak time period based on the aforementioned charging spatiotemporal correlation information. This first future charging spatiotemporal correlation information can represent changes in charging status within the future peak time period. In practice, the first future charging spatiotemporal correlation information can be in the form of a time-series sequence. The first future charging spatiotemporal correlation information can include multiple future charging information sets. Each future charging scenario in Western Sydney corresponds to a specific future peak time period.

[0062] As an example, based on the charging spatiotemporal correlation information, the temporal prediction capability corresponding to the multimodal large model is used to generate the first future charging spatiotemporal correlation information for the future peak time period.

[0063] The fourth step is to generate charging spatiotemporal conversion information for the target peak time period based on the traffic flow spatiotemporal correlation information. This charging spatiotemporal conversion information can be the charging spatiotemporal information for the target peak time period predicted based on the traffic flow spatiotemporal correlation information.

[0064] As an example, various charging conversion coefficients can be used to convert the spatiotemporal correlation information of traffic flow into spatiotemporal charging information to obtain charging spatiotemporal conversion information.

[0065] Fifth, based on the charging spatiotemporal transformation information, generate second future charging spatiotemporal correlation information for the future peak time period. This second future charging spatiotemporal correlation information can be various future charging information within the future peak time period predicted based on the charging spatiotemporal transformation information. Similarly, the second future charging spatiotemporal correlation information can be information in time-series format. The second future charging spatiotemporal correlation information can include multiple future charging information entries. Each future charging information entry can be the charging status for various future time periods predicted based on the charging spatiotemporal transformation information.

[0066] Step 6: Based on the future traffic flow spatiotemporal correlation information, generate third future charging spatiotemporal correlation information for the future peak time period. This third future charging spatiotemporal correlation information can be charging spatiotemporal correlation information transformed from the future traffic flow spatiotemporal correlation information. The specific implementation method is not detailed here; please refer to the generation of charging spatiotemporal conversion information.

[0067] Step 7: Based on the target peak time period and the urban area location information, obtain the corresponding spatiotemporal correlation coefficients for the first, second, and third future charging spatiotemporal correlation information. The spatiotemporal correlation coefficient can represent the importance and usefulness of the corresponding charging spatiotemporal correlation data. In practice, the charging spatiotemporal correlation coefficient can be a value between 0 and 1. A higher value indicates greater importance and higher usefulness of the corresponding charging spatiotemporal correlation data.

[0068] It should be noted that the spatiotemporal correlation coefficients for each charging station differ depending on the target peak time period and the location information within the urban area. Here, different urban locations and time periods result in different vehicle characteristics for the charging vehicles; therefore, each charging station has a unique spatiotemporal correlation coefficient.

[0069] Step 8: Based on the aforementioned spatiotemporal correlation coefficients for each charging station, generate the load supply information corresponding to that charging station. This load supply information can be for the charging station's load supply during future peak hours.

[0070] As an example, for each future peak sub-period within a future peak period, firstly, the first, second, and third future charging information corresponding to the future peak sub-period are extracted from the first, second, and third future charging spatiotemporal correlation information, respectively. Then, the first, second, and third future charging information, along with each charging spatiotemporal correlation coefficient, are weighted and fused to obtain future fused charging information. The values ​​of each charging information in the future fused charging information sequence are then increased to obtain the load supply information for the future peak period.

[0071] The above-mentioned content, as another inventive point of this disclosure, solves another technical problem: "When the occupied area corresponding to the charging station is large, there will be a large number of fuel vehicles occupying the space. When the occupied area is small, there will be large fluctuations in traffic flow, leading to unstable and inaccurate charging predictions." Based on this, this disclosure, for cases where the maximum value of the occupied area is not less than a target value, uses a first processing method to accurately generate charging spatiotemporal prediction information. Secondly, when the maximum value of the occupied area is less than the target value, through multi-faceted prediction of traffic flow and multi-faceted conversion prediction between traffic flow and charging spatiotemporal correlation information, accurate generation of load supply information can be achieved even under unstable traffic flow conditions.

[0072] Step 1022: Compare the charging spatiotemporal prediction information with the charging spatiotemporal association information to query the target time period set in which the information differences meet the target difference conditions.

[0073] In some embodiments, the aforementioned executing entity can compare the charging spatiotemporal prediction information with the charging spatiotemporal association information to query a target time period set where the information difference satisfies the target difference condition. Here, each predicted value in the charging spatiotemporal prediction information and each load value in the charging spatiotemporal association information have a one-to-one correspondence at the same time. The target difference condition can be that the difference between the predicted value and the load value at the same time is higher than a target threshold. The target threshold can be set based on historical experience. The target threshold can be a periodically updated threshold. That is, after satisfying the target difference condition, it indicates that there is a significant difference between the predicted load information and the actual load information within the corresponding time period, possibly due to the target vehicle occupying the charging pile, resulting in the charging pile being idle. Here, the target vehicle can be a vehicle that does not charge but occupies the charging position corresponding to the charging pile. For example, the target vehicle can be a gasoline vehicle. Here, considering the difficulty of parking during peak hours in public areas of the target city, some gasoline vehicles or electric vehicles that do not charge will occupy the parking space corresponding to the charging pile, resulting in a waste of charging resources at the charging station. Therefore, determining whether the information difference satisfies the target difference condition determines whether the charging pile is occupied by the target vehicle. The target time periods within the target time period set can be any time period within the target peak time period. Target time periods can be those where there is a significant difference between the predicted load and the actual load, potentially resulting in many charging stations being underutilized.

[0074] In some optional implementations of certain embodiments, the executing entity may compare the charging spatiotemporal prediction information with the charging spatiotemporal association information to query a target time period set in which the information differences satisfy the target difference condition, including the following steps: For each time period within the target peak time period, perform the following comparison steps: Sub-step 1 involves filtering out first charging prediction sub-information corresponding to the time period from the charging spatiotemporal prediction information. Here, each of the first charging prediction sub-information constitutes the charging spatiotemporal prediction information.

[0075] Sub-step 2 involves filtering out second charging prediction sub-information corresponding to the time period from the charging spatiotemporal correlation information. Here, each piece of second charging prediction sub-information constitutes the charging spatiotemporal correlation information.

[0076] Sub-step 3 involves determining whether the charging difference information between the first charging prediction sub-information and the second charging prediction sub-information satisfies the difference condition. For example, the difference condition could be whether the numerical difference between the first charging prediction sub-information and the second charging prediction information is higher than a target threshold.

[0077] Sub-step 4: In response to the determination that the condition is met, the time period is determined as the target time period.

[0078] Step 1023: Retrieve the set of historical charging video clips corresponding to the target time period set.

[0079] In some embodiments, the executing entity may retrieve a set of historical charging video clips corresponding to the target time period set. The target time period in the target time period set and the historical charging time clips in the historical charging time clip set have a corresponding relationship within the same time frequency band. For example, the target time period is the time period between 12:00 and 13:00 on July 1st. The corresponding historical charging video clip within the same time frequency band is the time period between 12:00 and 13:00 on July 8th. A historical charging video clip can be a video clip of charging occurring at a charging station during a historical time period.

[0080] It should be noted that the charging videos within the charging station can be temporarily stored.

[0081] Step 1024: Generate fuel vehicle occupancy information based on the historical charging video clip set.

[0082] In some embodiments, the aforementioned executing entity can generate gasoline vehicle occupancy information based on the historical charging video clip set. This gasoline vehicle occupancy information can be the occupancy ratio of each charging pile and corresponding parking space shown in the historical charging video clip set, indicating that gasoline vehicles occupy each space. That is, the gasoline vehicle occupancy information can be the occupancy ratio of charging stations under each target time period within a target time period set. In other words, the gasoline vehicle occupancy information includes multiple occupancy ratios. Each occupancy ratio corresponds to a specific time period.

[0083] As an example, the aforementioned implementing entity can use a multimodal large model to identify the license plates of the vehicles in the parking spaces corresponding to the charging piles in each historical charging video segment, in order to determine the occupancy ratio of fuel vehicles in the target time period and obtain each occupancy ratio as fuel vehicle occupancy information.

[0084] In some optional implementations of certain embodiments, after step 1024, the steps further include: The first step is to generate a corresponding charging solution based on the occupancy information of the fuel-powered vehicles. This solution can be a modified version of the occupancy status of the fuel-powered vehicles. In practice, the solution can address situations where a charging station's parking space is occupied by a target vehicle, tailored to different fuel-powered vehicle occupancy information. For example, the solution could be a linked charging system that only lowers the lock when a vehicle starts charging or a reservation is successful, automatically locking or refusing to lower the lock when not charging. Another solution could be to provide a 15-30 minute free relocation buffer period after charging is complete, after which charges will be applied per minute (e.g., 1 yuan / minute, with no upper limit), with strong reminders via App / SMS. Yet another solution could be to use cameras to identify whether a vehicle is plugged in, whether someone is present, and the duration of occupancy, automatically identifying it as a "suspected target vehicle" and sending the information to maintenance personnel. Here, the target vehicle can be a vehicle occupying a charging station's parking space without charging.

[0085] The second step is to perform a simulation process for the aforementioned charging scheme. This simulation process can be a method where the site agent predicts the charging state after the charging scheme is executed. Specific simulation methods will not be detailed here.

[0086] The third step involves executing the charging process according to the simulation results indicating that the occupancy reduction ratio has reached the target ratio. The target ratio can be a threshold used to measure whether the proportion of the target vehicle is relatively low. The target ratio can be determined based on historical experience.

[0087] In some optional implementations of certain embodiments, the historical charging video clips in the historical charging video clip set include multiple sub-videos from multiple shooting angles. Each sub-video has a different shooting angle.

[0088] Optionally, the executing entity can generate fuel vehicle occupancy information based on the historical charging video clip set, including the following steps: For each historical charging video clip, perform the following information generation steps: Sub-step 1: For each sub-video corresponding to the historical charging video segment, perform the following second processing step: The first sub-step involves coarse-grained frame extraction from the sub-video to obtain a first video frame sequence. Coarse-grained frame extraction can be performed with a longer frame duration. Considering that the target peak time period often lasts several hours, corresponding to sub-videos of tens of minutes, and that the available computing resources at charging stations are limited during peak periods, fine-grained frame extraction and subsequent video frame analysis might waste significant computing resources. Furthermore, vehicle speeds are often slow during peak periods, and fine-grained frame extraction contains a large amount of redundant information. Therefore, coarse-grained frame extraction not only reduces the waste of computing resources but also removes a large amount of redundant video content, improving vehicle recognition efficiency. Here, the frame extraction duration for coarse-grained frame extraction is longer than that for fine-grained frame extraction. For example, the frame extraction duration for coarse-grained frame extraction could be 1 second. Each video frame in the first video frame sequence is a sub-video frame after frame extraction.

[0089] The second sub-step involves marking each first video frame with a charging pile frame according to the charging pile layout information corresponding to the charging station, resulting in marked video frames and a sequence of marked video frames. The charging pile layout information can be the location layout of each charging station within the charging station. In practice, the charging pile layout information may include the coordinates of each charging pile in the coordinate system corresponding to the charging station. The charging pile frame marking involves marking each charging pile in the first video frame with a charging pile frame and a charging pile identifier. That is, each charging pile is outlined in the first video frame, and its identification is marked. The marked video frame is the video frame that displays the identifier of each charging pile and shows the presence of a charging pile frame.

[0090] The third sub-step involves grouping sub-images in the marked video frame sequence that are labeled with the same charging station into one category, resulting in a sub-image sequence set. Each sub-image in this set represents a different charging station. The sub-image sequences are ordered according to the video frame duration. Each sub-image sequence corresponds to a different charging station.

[0091] Sub-step 2 involves fusing the obtained multiple sub-image sequence sets to obtain a fused image sequence set. This sub-image sequence fusion can involve fusing sub-image sequences belonging to the same charging station. The fused image sequence is obtained by fusing at least one sub-image sequence. Each fused image sequence corresponds to a specific charging station.

[0092] Sub-step 3: For each fused image sequence, based on the fused image sequence, use a multimodal large model to generate initial vehicle model information and vehicle model confidence score for the parked charging station. The initial vehicle model information can be the preliminary identification of the vehicle model of the vehicle parked at the charging station. The vehicle model confidence score can be the accuracy of the initial vehicle model information. The vehicle model confidence score can be a value between 0 and 1. The higher the value, the more accurate the initial vehicle model information.

[0093] As an example, firstly, decision prompts can be generated to identify the vehicle model based on the fused image sequence. Then, these prompts are input into a multimodal large model to obtain initial vehicle model information and vehicle model confidence scores.

[0094] Sub-step 4: In response to the existence of at least one initial vehicle model in the initial vehicle model information set with a confidence level lower than the target confidence level, the at least one initial vehicle model is re-verified to obtain a verification result. The target confidence level can be a confidence threshold used to measure whether the initial vehicle model information is sufficiently accurate. The verification result can be a verification result regarding the accuracy of at least one initial vehicle model. In addition, the verification result may also include: corrected vehicle model information.

[0095] Sub-step 5: Based on the verification results and the initial vehicle model information set, generate fuel vehicle occupancy sub-information. This fuel vehicle occupancy sub-information can be the percentage of each fuel vehicle appearing in the sub-video.

[0096] As an example, firstly, the aforementioned execution entity can adjust the initial vehicle information in the initial vehicle information set based on the corrected vehicle information corresponding to the verification results and the verification sub-results indicating no problems, thus obtaining an adjusted vehicle information set. Then, it generates the fuel vehicle occupancy sub-information corresponding to the adjusted vehicle information set.

[0097] Optionally, the executing entity may re-verify the at least one initial vehicle model information to obtain a verification result, including the following steps: The first step, for each initial vehicle model information, is to perform the third processing step: Sub-step 1: Determine the fused image sequence corresponding to the initial vehicle model information as the target fused image sequence.

[0098] Sub-step 2 involves selecting the target fused image with the highest confidence level from the target fused image sequence to output the initial vehicle model information. The confidence level can be defined as the probability that the vehicle model displayed in the target fused image matches the initial vehicle model information.

[0099] Sub-step 3 involves determining the capture time corresponding to the target fused image. The capture time can be a specific video time point corresponding to the target fused image.

[0100] Sub-step 4: For each of the multiple sub-videos, select a video segment with a target duration centered on the shooting time. The target duration can be a pre-set duration. Here, the target duration is determined based on the average vehicle speed during peak hours. The target duration can be determined based on a mapping relationship corresponding to the average vehicle speed.

[0101] Sub-step 5: For each video segment, perform the fourth processing step: The first sub-step involves performing fine-grained frame extraction on the video segment to obtain a first video frame sequence. Here, the frame extraction duration for fine-grained extraction is shorter than that for coarse-grained extraction. For example, the frame extraction duration for fine-grained extraction could be 0.5 seconds.

[0102] The second sub-step involves cropping the image content of the charging pile corresponding to the initial vehicle model information from each first video frame to obtain a cropped image.

[0103] Sub-step 6 involves performing multi-view image fusion on the obtained multiple cropped image sequences to obtain a multi-view fused image sequence. This multi-view image fusion can involve fusing images of the same charging station from multiple perspectives. The multi-view fused image can be the result of fusing images from multiple perspectives.

[0104] Sub-step 7: Based on the multi-vision fusion image sequence, generate the vehicle model verification confidence score corresponding to the initial vehicle model information. The vehicle model verification confidence score reflects the accuracy of verifying the initial vehicle model information. The vehicle model verification confidence score can be a value between 0 and 1; a higher value indicates a more successful verification of the initial vehicle model information.

[0105] As an example, based on the multi-vision fusion image sequence, a multimodal large model can be used to verify the vehicle model information of the initial vehicle model, and obtain the vehicle model verification confidence and the corrected vehicle model information.

[0106] The second step is to generate the verification result corresponding to the initial vehicle model information based on the vehicle model verification confidence level.

[0107] As an example, in response to a vehicle model verification confidence level less than a predetermined confidence threshold, a verification result is generated indicating that the initial vehicle model information failed verification and that the corrected vehicle model information is identified as the initial vehicle model information. In response to a vehicle model verification confidence level not less than the predetermined confidence threshold, a verification result is generated indicating that the initial vehicle model information passed verification.

[0108] Step 1025: Based on the occupancy information of the fuel vehicle, adjust the charging time-space correlation information to obtain charging time-space adjustment information.

[0109] In some embodiments, the aforementioned executing entity can adjust the charging spatiotemporal correlation information based on the fuel vehicle occupancy information to obtain charging spatiotemporal adjustment information. The correlation information adjustment can be an adjustment of the charging load in various time periods. The charging spatiotemporal adjustment information can be the charging spatiotemporal correlation relationship after load adjustment for the charging load in various target time periods.

[0110] As an example, firstly, for each occupancy percentage in the fuel vehicle proportion information, the first step is to determine the load adjustment information corresponding to the occupancy percentage that has a mapping relationship. Secondly, the load content of the second charging prediction sub-information in the charging spatiotemporal correlation information that has the same target time period as the proportion is adjusted to obtain the adjustment result. Finally, each second charging prediction sub-information in the charging spatiotemporal correlation information is replaced with the respective adjustment result to obtain the charging spatiotemporal adjustment information.

[0111] As another example, based on the occupancy information of the fuel vehicles, a multimodal large model is used to re-predict and adjust the spatiotemporal correlation information of charging to obtain spatiotemporal adjustment information.

[0112] In some optional implementations of certain embodiments, after step 1025, the steps further include: The first step is to generate charging vehicle model ratio information based on the charging time-space adjustment information. This information can be the proportion of each vehicle model being charged. In practice, the charging vehicle model ratio information can be compared according to the vehicle nameplate and model number. The charging vehicle model ratio information can be in time series format; that is, it can include the proportion of each charging vehicle model in different time periods.

[0113] As an example, firstly, the charging time-space adjustment information can be extracted to identify the various charging vehicle models involved in different time periods. Then, the number of each charging vehicle model corresponding to each model can be determined for comparison, thus obtaining the charging vehicle model ratio information.

[0114] The second step involves generating a charging mismatch coefficient based on the aforementioned charging vehicle proportion information. This coefficient characterizes the degree of mismatch between vehicles parked in the corresponding charging spaces and those not actually charging. The charging mismatch coefficient can be a value between 0 and 1; a higher value indicates a higher proportion of parked vehicles not being charged. The charging mismatch coefficient can also be presented in a time-series format, meaning it can include various mismatch coefficients for different time periods.

[0115] As an example, for each charging vehicle ratio in the charging vehicle ratio information, firstly, the percentage of each charging vehicle whose charging protocol does not meet the charging requirements of the charging station is retrieved from the charging vehicle ratio, and this percentage is used as the charging mismatch coefficient.

[0116] The third step is to readjust the charging time-space adjustment information based on the charging mismatch coefficient to obtain the charging time-space readjustment information, which is used as the charging time-space adjustment information.

[0117] Step 1026: Generate load supply information corresponding to the charging station based on the charging time and space adjustment information.

[0118] In some embodiments, the aforementioned executing entity can generate load supply information corresponding to the charging station based on the charging spatiotemporal adjustment information. The load supply information can be the load supply situation of the charging station at the same time in the future. Here, "at the same time" can be the same hour segment within the same week of a future time period. The load supply information can be in the form of a time series.

[0119] As an example, the aforementioned implementing entity can use a multimodal large model to predict the load time series for the same period in the future based on the charging time-space adjustment information, and obtain the future load supply information sequence as the load supply information.

[0120] Step 103: Send the load supply information of each item to the regional intelligent agent corresponding to the public area of ​​the target city, so as to supply power to each charging station during future peak periods.

[0121] In some embodiments, the aforementioned executing entity may send the respective load supply information to the regional intelligent agent corresponding to the public area of ​​the target city, so as to ensure power supply to each charging station during future peak periods. The regional intelligent agent may be an intelligent agent built specifically for the public area of ​​the target city to support the processing of various matters within the area. The regional intelligent agent can support intelligent decision-making for various matters within the area.

[0122] The above-described embodiments of this disclosure have the following beneficial effects: By utilizing the intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics, based on some embodiments of this disclosure, and considering various factors affecting the load of charging stations, the load power supply situation in public areas of a target city can be accurately and efficiently determined, avoiding waste of power resources. Specifically, the reason for insufficient power supply during peak hours is that the load changes of charging stations in public areas of the target city are affected by many factors, making it impossible to fully determine the load usage of charging stations. This leads to insufficient power supply during peak hours when manually formulating load management strategies. Therefore, the intelligent load processing method for electric vehicle clusters based on spatiotemporal correlation characteristics of some embodiments of this disclosure first obtains the spatiotemporal correlation information of charging and traffic flow for each charging station in the public area of ​​the target city during the target peak time period. Here, by obtaining the spatiotemporal correlation information of charging and traffic flow during the target peak time period, it is possible to subsequently determine the corresponding load supply information that meets the power demand of the charging station from the dimensions of actual charging and traffic flow. Then, for each charging station, the following steps are performed using the corresponding deployed site agent: First, based on the spatiotemporal correlation information of the traffic flow, accurate spatiotemporal prediction information for charging at the charging station can be generated to determine the possible charging status of the charging station based on the traffic flow. Second, the spatiotemporal prediction information for charging is compared with the spatiotemporal correlation information to query the target time period set where the information difference meets the target difference condition. Here, by comparing charging information differences, the time periods with significant differences between charging information and actual charging information under the traffic flow dimension can be preliminarily and accurately determined to determine whether there is a charging impact based on these time periods, thus improving the accuracy of the charging station's load prediction. Third, the historical charging video clip set corresponding to the target time period set is retrieved to serve as the data basis for judging the charging impact, facilitating the subsequent determination of the occupancy status of fuel vehicles. Fourth, fuel vehicle occupancy information can be accurately generated based on the historical charging video clip set. Here, by determining the fuel vehicle occupancy information, it is determined whether there is a serious fuel vehicle occupancy problem at the charging station, resulting in the available load of the charging station being lower than the actual usage. During peak hours, considering only actual usage, some charging stations may not be fully utilized due to occupancy by gasoline-powered vehicles, leading to insufficient power supply. The fifth step involves adjusting the charging time-space correlation information based on the gasoline-powered vehicle occupancy information to accurately generate charging time-space adjustment information while taking into account the gasoline-powered vehicle occupancy issue. The sixth step involves generating the load supply information corresponding to the charging station based on the charging time-space adjustment information.Here, based on accurate charging time and space adjustment information, load supply can be accurately predicted. Finally, the load supply information is sent to the regional agents corresponding to the public areas of the target city. This ensures that power supply to each charging station during future peak hours can fully utilize their charging capacity and avoid power shortages during peak periods. In summary, by considering various factors affecting charging station load and utilizing intelligent agents, the load supply situation in the public areas of the target city can be accurately and efficiently determined, avoiding waste of electrical resources.

[0123] Further reference Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of an intelligent load processing device for electric vehicle clusters based on spatiotemporal correlation characteristics. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, this intelligent load processing device for electric vehicle clusters based on spatiotemporal correlation characteristics can be specifically applied to various electronic devices.

[0124] like Figure 2 As shown, an intelligent load processing device 200 for electric vehicle clusters based on spatiotemporal correlation features includes: an acquisition unit 201, an execution unit 202, and a transmission unit 203. The acquisition unit 201 is configured to acquire the spatiotemporal correlation information of charging and the spatiotemporal correlation information of traffic flow for each charging station in the public area of ​​the target city during the target peak time period. The execution unit 202 is configured to, for each charging station, use the corresponding deployed site agent to perform the following generation steps: generate spatiotemporal prediction information of charging for the charging station based on the spatiotemporal correlation information of traffic flow; compare the spatiotemporal prediction information of charging with the spatiotemporal correlation information of charging to query the target time period set in which the information difference meets the target difference condition; retrieve the set of historical charging video clips corresponding to the target time period set; generate fuel vehicle occupancy information based on the set of historical charging video clips; adjust the spatiotemporal correlation information of charging based on the fuel vehicle occupancy information of charging to obtain spatiotemporal adjustment information of charging; and generate load supply information corresponding to the charging station based on the spatiotemporal adjustment information of charging. The sending unit 203 is configured to send the load supply information to the regional agent corresponding to the public area of ​​the target city to ensure power supply to each charging station during the future peak time period.

[0125] It is understandable that the units described in the intelligent load processing device 200 for electric vehicle clusters based on spatiotemporal correlation characteristics are related to the reference... Figure 1The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method are also applicable to the intelligent load processing device 200 for electric vehicle clusters based on spatiotemporal correlation features and the units contained therein, and will not be repeated here.

[0126] The following is for reference. Figure 3 It shows a schematic diagram of the structure of an electronic device (e.g., an electronic device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0127] like Figure 3 As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0128] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0129] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0130] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0131] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0132] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs. When the electronic device executes the aforementioned one or more programs, the electronic device causes the following steps: It acquires charging spatiotemporal correlation information and vehicle flow spatiotemporal correlation information for each charging station in the target city's public area during a target peak time period; for each charging station, it utilizes a corresponding deployed site agent to perform the following generation steps: Based on the vehicle flow spatiotemporal correlation information, it generates charging spatiotemporal prediction information corresponding to the charging station; it compares the charging spatiotemporal prediction information with the charging spatiotemporal correlation information to query a target time period set whose information differences satisfy the target difference condition; it retrieves a set of historical charging video clips corresponding to the target time period set; it generates fuel vehicle occupancy information based on the historical charging video clip set; it adjusts the charging spatiotemporal correlation information based on the fuel vehicle occupancy information to obtain charging spatiotemporal adjustment information; it generates load supply information corresponding to the charging station based on the charging spatiotemporal adjustment information; and it sends the load supply information to the regional agent corresponding to the target city's public area to ensure power supply to each charging station during future peak time periods.

[0133] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0134] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0135] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including an acquisition unit, an execution unit, and a transmission unit. The names of these units do not necessarily limit the unit itself; for example, the acquisition unit may also be described as "a unit that acquires the spatiotemporal correlation information of charging stations and the spatiotemporal correlation information of traffic flow corresponding to each charging station in a target city's public area during a target peak time period."

[0136] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0137] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for intelligent load processing of electric vehicle clusters connected to the grid based on spatiotemporal correlation characteristics, comprising: Obtain the spatiotemporal correlation information of charging and the spatiotemporal correlation information of vehicle flow for each charging station in the public area of ​​the target city during the target peak time period; For each charging station, the generation steps are performed using the corresponding deployed site agent: Based on the spatiotemporal correlation information of the traffic flow, the spatiotemporal prediction information of the charging station is generated. The charging spatiotemporal prediction information is compared with the charging spatiotemporal association information to query the target time period set in which the information difference meets the target difference condition; Retrieve the set of historical charging video clips corresponding to the target time period set; Based on the historical charging video clip set, fuel vehicle occupancy information is generated; Based on the occupancy information of the fuel vehicles, the charging time-space correlation information is adjusted to obtain charging time-space adjustment information; Based on the charging time-space adjustment information, the load supply information corresponding to the charging station is generated; The load supply information is sent to the corresponding regional intelligent agent in the public area of ​​the target city to ensure power supply to each charging station during future peak periods.

2. The method according to claim 1, wherein, After generating the gasoline vehicle occupancy information based on the historical charging video clip set, the method further includes: Based on the occupancy information of the fuel vehicles, a corresponding charging solution is generated; Perform a scheme simulation process for the aforementioned charging scheme; In response to the simulation results indicating that the occupancy reduction ratio has reached the target ratio, the corresponding content of the charging processing scheme is executed.

3. The method according to claim 1, wherein, The step of generating charging spatiotemporal prediction information corresponding to the charging station based on the vehicle flow spatiotemporal correlation information includes: Obtain the road information set corresponding to the charging station; For each road information, perform the following first processing step: Extract the spatiotemporal sub-information of traffic flow corresponding to the road information from the spatiotemporal correlation information of traffic flow; Based on the aforementioned traffic flow spatiotemporal sub-information, charging spatiotemporal prediction sub-information is generated; Obtain the charging conversion coefficient corresponding to the road information; Based on the obtained charging spatiotemporal prediction sub-information set and charging conversion coefficient set, charging spatiotemporal prediction information is generated.

4. The method according to claim 1, wherein, The step of comparing the charging spatiotemporal prediction information with the charging spatiotemporal association information to query a target time period set in which the information differences satisfy the target difference condition includes: For each time period within the target peak time period, perform the following comparison steps: Filter out the first charging prediction sub-information corresponding to the time period from the charging spatiotemporal prediction information; Filter out the second charging prediction sub-information corresponding to the time period from the charging spatiotemporal correlation information; Determine whether the charging difference information between the first charging prediction sub-information and the second charging prediction sub-information satisfies the difference condition; In response to the determination that the condition is met, the time period is defined as the target time period.

5. The method according to claim 1, wherein, The historical charging video clips in the historical charging video clip set include: multiple sub-videos from multiple shooting angles; and The step of generating fuel vehicle occupancy information based on the historical charging video clip set includes: For each historical charging video clip, perform the following information generation steps: For each sub-video corresponding to the historical charging video segment, perform the following second processing step: The sub-video is subjected to coarse-grained frame extraction to obtain the first video frame sequence; Based on the charging pile layout information corresponding to the charging station, each first video frame is marked with a charging pile frame to obtain marked video frames and a marked video frame sequence. Sub-images in the marked video frame sequence that are marked with the same charging pile are grouped into one category to obtain a set of sub-image sequences. The obtained multiple sub-image sequence sets are fused to obtain a fused image sequence set; For each fused image sequence, based on the fused image sequence, the initial vehicle model information and vehicle model confidence of the charging pile parking are generated using a multimodal large model; In response to the existence of at least one initial vehicle information in the initial vehicle information set whose corresponding vehicle confidence score is lower than the target confidence score, the at least one initial vehicle information is re-verified to obtain a verification result; Based on the verification results and the initial vehicle model information set, fuel vehicle occupancy sub-information is generated.

6. The method according to claim 5, wherein, The step of re-verifying the at least one initial vehicle model information to obtain a verification result includes: For each initial vehicle model information, perform the third processing step: The fused image sequence corresponding to the initial vehicle model information is determined as the target fused image sequence; The target fused image with the highest confidence level corresponding to the initial vehicle model information is selected from the target fused image sequence; Determine the capture time corresponding to the target fused image; For each of the plurality of sub-videos, a video segment with a target duration centered on the shooting time is selected from the sub-videos; For each video segment, perform the fourth processing step: Fine-grained frame extraction is performed on the video segment to obtain the first video frame sequence; For each first video frame, the image content of the charging pile corresponding to the initial vehicle model information is cropped from the first video frame to obtain a cropped image; Multi-view image fusion is performed on the obtained multiple cropped image sequences to obtain a multi-view fused image sequence; Based on the multi-vision fusion image sequence, generate the vehicle verification confidence level corresponding to the initial vehicle information; Based on the vehicle model verification confidence level, a verification result corresponding to the initial vehicle model information is generated.

7. The method according to claim 1, wherein, After adjusting the charging time-space correlation information based on the fuel vehicle occupancy information to obtain charging time-space adjustment information, the method further includes: Based on the charging time-space adjustment information, the charging vehicle ratio information is generated; Based on the charging vehicle model ratio information, a charging mismatch coefficient is generated; Based on the charging mismatch coefficient, the charging time-space adjustment information is adjusted again to obtain charging time-space readjustment information, which is used as the charging time-space adjustment information.

8. An intelligent load processing device for electric vehicle clusters based on spatiotemporal correlation characteristics, comprising: The acquisition unit is configured to acquire the spatiotemporal correlation information of charging and the spatiotemporal correlation information of traffic flow for each charging station in the public area of ​​the target city during the target peak time period. The execution unit is configured to, for each charging station, utilize the corresponding deployed site agent to perform the generation step: generating charging spatiotemporal prediction information corresponding to the charging station based on the vehicle flow spatiotemporal correlation information; The charging spatiotemporal prediction information is compared with the charging spatiotemporal association information to query the target time period set whose information differences meet the target difference conditions; the historical charging video clip set corresponding to the target time period set is retrieved; based on the historical charging video clip set, fuel vehicle occupancy information is generated; based on the fuel vehicle occupancy information, the charging spatiotemporal association information is adjusted to obtain charging spatiotemporal adjustment information. Based on the charging time-space adjustment information, the load supply information corresponding to the charging station is generated; The sending unit is configured to send the respective load supply information to the regional intelligent agent corresponding to the public area of ​​the target city, so as to supply power to each charging station during future peak periods.

9. An electronic device, comprising: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

10. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images