Service Area Vehicle Charging Analysis and Processing Method Based on Video Analytics

Abstract

This invention relates to the field of charging analysis technology, specifically a service area vehicle charging analysis and processing method based on video analysis. The method includes: connecting vehicles and charging piles in a charging event to construct a charging status distribution map; determining whether the current charging event is a valid charging event based on the update status of each vehicle in the charging status distribution map, and tracking the status of the current vehicle; based on the results of the current vehicle status tracking, filtering vehicles and charging piles in the current charging pile area video for scheduling objects, and identifying at least one scheduling object; combining the scheduling objects by nodes in the charging status distribution map to determine the scheduling probability of each scheduling object; and sorting the scheduling objects based on their scheduling probabilities to determine the matching charging order for each vehicle; thereby improving the utilization rate and allocation effect of charging resources.

Description

Technical Field

[0001] This invention relates to the field of charging analysis technology, specifically a service area vehicle charging analysis and processing method based on video analysis. Background Technology

[0002] In recent years, electric vehicles have gained wider application due to their advantages such as energy saving, environmental protection, and high energy utilization. As a charging device for electric vehicles, charging piles, as a new type of charging equipment, are becoming increasingly popular among car owners, which also means more intense competition. The allocation method of charging piles, as devices used by users during charging, limits the efficiency of charging scheduling.

[0003] For example, Chinese Patent Publication No. CN119369973A discloses a mobile charging vehicle charging scheduling method and system, including: obtaining the charging posture corresponding to each vehicle to be charged; guiding each vehicle to be charged to adjust its corresponding charging posture based on the standard charging posture; obtaining the basic information corresponding to each vehicle to be charged; establishing several demand selection conditions for this charging based on the charging demand uploaded by the corresponding user; selecting the corresponding target mobile charging vehicle according to the demand selection conditions; constructing the guiding movement direction corresponding to the target mobile charging vehicle; controlling each target mobile charging vehicle to move forward according to the corresponding guiding movement direction; collecting the real-time video stream corresponding to each target mobile charging vehicle; guiding the target mobile charging vehicle to perform real-time obstacle avoidance; and scheduling nearby suitable mobile charging vehicles to perform mobile charging according to the user's charging demand.

[0004] For example, Chinese Patent Publication No. CN118478737A discloses an intelligent monitoring method for charging data based on wireless connection and its storage medium. This invention analyzes the characteristic charging data generated by the user during the process of charging the vehicle using the charging device, and judges and identifies the remaining battery power value of the vehicle that meets the user's usage habits and does not cause the vehicle to be over-discharged. Based on this, it realizes charging planning prompts for the user, and scientifically guides the user on how much battery power to charge the vehicle each time they choose a charging device, as well as the waiting time required to charge to the above value.

[0005] Existing technologies describe how to quantify the charging demand of vehicles based on remaining battery power and the location of charging stations, and guide charging accordingly; and how to extract historical intervals and charging records, and after eliminating risk warnings for abnormal distribution, arrange vehicles to corresponding charging areas. These methods explain the quantitative demand for charging, but in scenarios with multiple vehicles and multiple charging stations, these vehicle scheduling methods are prone to inaccurate scheduling targets, making it impossible to allocate each vehicle to the corresponding charging station in a timely manner, resulting in low efficiency in charging guidance and low utilization of charging resources. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a service area vehicle charging analysis and processing method based on video analysis, including: S1, identifying the vehicle, parking space and vehicle entry and exit path corresponding to each frame of the video in the charging pile area, and treating each vehicle as a charging event to be processed.

[0007] S2 connects the vehicle and charging pile in the current charging event based on the vehicle-side charging status and the charging pile-side charging status, and constructs the charging status distribution.

[0008] S3, based on the update status of each vehicle in the charging status distribution map, determines whether the current charging event is a valid charging event, and performs status tracking on the current vehicle.

[0009] S4. Based on the results of current vehicle status tracking, filter the vehicles and charging piles in the current charging pile area video to identify at least one object to be scheduled.

[0010] S5, combine the objects to be scheduled into nodes in the charging state distribution map, and determine the schedulable probability of each object to be scheduled.

[0011] S6 sorts the vehicles to be scheduled based on the schedulable probability and determines the charging order for each vehicle.

[0012] The beneficial effects of this invention are as follows: First, by labeling the parking space occupancy status, vehicle parking status, and vehicle entry and exit paths in the video of the charging pile area, this invention forms an associated link. Combined with the vehicle parking time, charging events are initially classified, and each parked vehicle with potential charging needs is identified as a charging event to be processed. Then, the changes in charging status and the vehicle-pile connection relationship are identified, and the vehicle charging time and the minimum time interval for state transition are synchronized. The vehicles are matched with the charging piles to construct a charging status distribution map, which ensures the matching of vehicles and charging status in the global scenario and provides a data foundation for subsequent data updates and filtering.

[0013] Second, this invention transforms node state changes into state sequences, verifies whether they conform to the charging process, and if they conform, illuminates a valid charging event and outputs the state tracking result; in case of anomalies, it initializes or re-matches, restores the state before the deviation, and synchronously identifies state changes, thereby realizing valid event judgment and anomaly correction; it avoids the chaotic state that occurs when matching vehicles and charging piles, and ensures that the state changes recorded in the charging state distribution map are a reliable data foundation.

[0014] Third, this invention statistically analyzes vehicle-charging pile matching data pairs by time, records the charging pile occupancy time of each data pair, sorts the data by timestamps, takes the intersection of adjacent times to determine the first priority, and then divides the data outside the intersection into quartiles based on their correlation weights to determine the second and third priorities, thus achieving priority filtering of the objects to be scheduled. First, it ensures that objects without time conflicts and with stable states are scheduled first, avoiding stable objects from waiting idly. Then, it selects objects with high correlation weights to obtain resources first, avoiding low-compatibility objects from wasting high-quality charging piles, forming a priority order of stability first and compatibility second, which establishes an orderly basis for scheduling objects for subsequent charging sequences and avoids scheduling chaos.

[0015] Fourth, this invention records the state probability based on the node records of the charging state distribution map, performs joint probability calculation according to the normal charging process, outputs the schedulable probability, and quantifies the scheduling feasibility of each schedulable object; then, it adjusts vehicles of different priorities, sorts them according to schedulable probability + waiting time, and determines the charging order; the charging order takes order, low risk and high utilization as the core idea, constructs the charging order in the current scenario, and achieves the goal of no waste of resources, no backlog of vehicles and no scheduling failure in multi-vehicle and multi-charging scenarios. Attached Figure Description

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] Figure 1 This is a flowchart illustrating a service area vehicle charging analysis and processing method based on video analytics.

[0018] Figure 2 This is a flowchart illustrating step S1 of the service area vehicle charging analysis and processing method based on video analytics.

[0019] Figure 3 This is a flowchart illustrating step S2 of the service area vehicle charging analysis and processing method based on video analytics.

[0020] Figure 4 This is a flowchart illustrating step S3 of the service area vehicle charging analysis and processing method based on video analytics.

[0021] Figure 5 This is a flowchart illustrating step S4 of the service area vehicle charging analysis and processing method based on video analytics.

[0022] Figure 6 This is a flowchart illustrating step S6 of the service area vehicle charging analysis and processing method based on video analytics. Detailed Implementation

[0023] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0024] See Figure 1 A service area vehicle charging analysis and processing method based on video analysis includes: S1, identifying the vehicle, parking space and vehicle entry and exit path corresponding to each frame of the video in the charging pile area, and treating each vehicle as a charging event to be processed.

[0025] S2 connects the vehicle and charging pile in the current charging event based on the vehicle-side charging status and the charging pile-side charging status, and constructs the charging status distribution.

[0026] S3, based on the update status of each vehicle in the charging status distribution map, determines whether the current charging event is a valid charging event, and performs status tracking on the current vehicle.

[0027] S4. Based on the results of current vehicle status tracking, filter the vehicles and charging piles in the current charging pile area video to identify at least one object to be scheduled.

[0028] S5, combine the objects to be scheduled into nodes in the charging state distribution map, and determine the schedulable probability of each object to be scheduled.

[0029] S6 sorts the vehicles to be scheduled based on the schedulable probability and determines the charging order for each vehicle.

[0030] The current solution involves deploying cameras in the charging pile area to cover the charging pile area, parking spaces, and vehicle entry and exit paths. The collected video data is then timestamped with the real-time current, voltage, and power data collected by the charging pile to form a complete charging event record. The identified vehicles can be identified using the YOLO7 algorithm, which extracts bounding boxes to identify parked vehicles and vehicles entering in real time.

[0031] Recorded charging events can include identifiers such as valid charging, invalid parking space occupancy, short charging with long pauses, incomplete charging, and failed plug-in events. These events are identified through a series of charging status changes, such as valid charging (vehicle parked → successful plug-in → charging at the charging station → normal termination); invalid parking space occupancy (vehicle parked → charging not started); short charging with long pauses (vehicle parked → successful plug-in → charging at the charging station → short charging time + long parking time); and incomplete charging (charging in progress or charging interruption). A failed plug-in event indicates that the vehicle did not transition to charging after plugging in. In this case, charging interruption, failed plug-in, and invalid parking are scenarios that need to be avoided, as they reduce the utilization rate of parking spaces near the charging station.

[0032] The charging events identified at this point represent abnormal and normal charging events in the overall process. These events will affect the scheduling of the overall charging process. At this time, it is necessary to identify the charging events in the current charging pile area by using video and specific power data, and synchronizing the data of the parking space where the charging pile is located.

[0033] As for the relative time between successful and failed plug-in attempts, sensors can be installed near the charging station. After the camera detects the plug-in action, the sensor can be used to determine whether the connection was successful, which will facilitate subsequent updates to the charging status of the charging station and the vehicle.

[0034] like Figure 2 As shown, the implementation of step S1 includes: S11, using image segmentation, marking the occupancy status of each parking space and the parking status of the vehicle in the charging pile area video, and forming an association link between the vehicle, parking space and vehicle entry and exit path; the occupancy status of the parking space indicates whether there is a vehicle in the corresponding parking space, and the parking status of the vehicle records the parking time of the vehicle and the timestamp of its appearance in the charging pile area video, recording the driving process of each vehicle in the service area.

[0035] During image segmentation, a semantic segmentation algorithm, specifically the U-Net algorithm, is used to label parking spaces. The input image resolution is set to 640×640, and the occupancy status is determined by the pixel mask of the labeled parking space area. The YOLO7 algorithm is used for vehicle recognition to ensure that the position of each vehicle in each frame of the image is identified.

[0036] S12, based on the acquired association links and the parking time of each vehicle, performs preliminary event classification for each vehicle and filters out the charging events to be processed for each vehicle.

[0037] When performing initial event classification, vehicles that do not require charging, such as fuel vehicles, vehicles that pass by temporarily and stop for ≤3 seconds, and vehicles that are not near charging stations, will be removed first. Then, the charging events to be processed will be described according to whether the vehicle stops, the stopping time, and the parking space occupancy status, and the usage status of the current charging station area will be recorded.

[0038] When recording the time a vehicle is parked, 3 seconds can be selected as the judgment value. A vehicle that stays in the corresponding parking space for 3 seconds without moving is considered parked, otherwise it is considered a temporary passing vehicle. As for vehicles that are not close to the charging pile, they are not considered vehicles that belong to the current charging pile area. These are not the basic data for the current analysis of charging behavior.

[0039] In one embodiment of the present invention, during step S2, the charging status of the vehicle side and the charging side will be recorded, and the current status of the vehicle and the corresponding charging pile will be synchronized to facilitate subsequent screening of schedulable vehicles.

[0040] The charging status on the vehicle side can be represented as waiting to charge, plugging in, charging, charging completed and waiting to unplug, and occupying the space without charging intention; the charging status on the charging pile side can be represented as idle, charging, about to complete, and fault.

[0041] like Figure 3 As shown, the implementation of step S2 also includes: S21, determining whether there is a change in charging status in the current charging event; when there is a change in charging status, identifying the connection relationship between each vehicle and the charging pile; when a change in charging status is determined, focusing on vehicles that are waiting, vehicles that are about to complete charging and generate schedulable charging piles, and vehicles occupying positions, to obtain the positions of vehicles that need to be scheduled and charging piles.

[0042] S22, obtain the charging pile associated with each vehicle, synchronize the charging time and remaining power of each vehicle, and retrieve the minimum time interval for each vehicle under the charging state transition; the minimum time interval is used to record the changes of the vehicle or charging pile under the conditions of waiting, charging, about to complete charging, leaving and generating a schedulable charging pile and abnormal occupancy, as well as the minimum value of the time interval for each change, to quantify the update status of relevant data for each vehicle.

[0043] S23, in response to the minimum time interval of the charging state transition, record the time difference between the time point when the vehicle's charging state changes and the time point when the charging pile's charging state changes; use this time difference as the association weight between the corresponding vehicle and the charging pile.

[0044] This step quantifies the time response of changes in vehicle charging status to changes in charging pile charging status, and uses the time response as the association weight between the vehicle and the charging pile. The higher the weight, the higher the matching efficiency between the vehicle and the charging pile. When multiple vehicles are waiting for multiple charging piles, the vehicle-pile combination with the higher weight is matched first.

[0045] For example, if the current vehicle status is waiting, and the charging station status changes from charging to idle, then the calculated difference represents the length of time the vehicle is waiting to charge. Or, if the current vehicle is charging and the charging station is also charging, then the difference represents the time difference between the two completing data synchronization. This value also represents the response status of the corresponding location. Based on the weights obtained from these situations, we can promptly understand the relative status of the current vehicle when waiting to charge, charging, and after charging is completed and the vehicle has left, thus achieving matching and scheduling of vehicles and charging stations.

[0046] Preferably, when constructing the charging status distribution map, it is necessary to identify the specific behavior of the current charging event, such as whether the charging gun is successfully plugged in / out, charging is interrupted, and the duration of the current charging event, in order to determine the charging events that the vehicle-side charging status and the charging pile-side charging status can represent.

[0047] The implementation of step S21 also includes: for each vehicle's charging status change, obtaining the triggering conditions for each vehicle's charging status change, and filling each node in the charging status distribution map with data according to three dimensions: equipment failure, user active operation, and time-limited resource pressure, to obtain multiple nodes corresponding to the triggering conditions.

[0048] Based on multiple nodes corresponding to the triggering conditions, the vehicle-side charging status and charging pile-side charging status of the corresponding nodes are determined, and after being tagged, they are synchronized to the charging status distribution map.

[0049] Triggering conditions can include: 1. Sudden change in charging pile status, from normal to faulty. This indicates an anomaly between the currently matched vehicle and the charging pile, requiring re-matching. 2. Changes in user operation, such as canceling charging or changing parking spaces. This represents a user-initiated change, indicating an unconventional charging situation, requiring synchronized updates to the distribution map data for this triggering condition. It should be noted that user operation changes often occur after prior reservations or entry scheduling, where users adjust their parking spaces accordingly. 3. Time-based load fluctuations, such as during peak hours. This indicates changes in the load of different charging piles, preventing vehicles from being preferentially matched to high-load charging piles. Here, the load status of charging piles during use needs to be recorded, and subsequent scheduling of charging piles should be based on load to prevent charging pile overload or other abnormal situations. These three triggering conditions represent equipment failure, user-initiated operation, and charging status under time-based resource pressure, supplementing information on changes in vehicle status and preventing abnormal data from interfering with subsequent analysis and decision-making.

[0050] It should be noted that in the charging status distribution map constructed in step S2, each node represents a set of vehicle and charging pile inputs, such as (vehicle: charging; pile: charging). Then, according to the data on each node, it is represented using a state sequence or other data form. The set association weight is used to identify the matching situation of vehicle and charging pile, and to prevent errors in the monitored data, which would cause the incorrect identification of the state evolution of vehicle and charging pile.

[0051] Preferably, in order to determine whether the currently matched vehicle and charging pile belong to the compatible part, the matching data between each charging pile and the vehicle will be determined according to the compatibility between the charging pile and the corresponding vehicle model, the historical interaction situation and the real-time load of the charging pile.

[0052] This involves analyzing whether each vehicle's model is compatible with the corresponding charging station, the historical matching success rate, the charging station's load rate, and the ratio of the corresponding time difference to the maximum allowable time difference. After standard post-processing, these data are weighted and summed with weights of 0.2, 0.2, 0.1, and 0.5 to illustrate the association weight between each vehicle and its corresponding charging state. This emphasizes whether the currently selected vehicle and charging state belong to a matchable combination. When there are multiple vehicles waiting to be charged and available charging stations, the association weight can quickly filter out the optimal vehicle-charging station combination, avoiding inefficiency or failure caused by blind matching. At this time, associating the data with the corresponding nodes facilitates subsequent matching of related vehicles and timely guidance and analysis for vehicle charging.

[0053] The values ​​for whether each vehicle model is compatible with the corresponding charging station, as explained above, will be expressed as logical values: 1 for compatible and 0 for otherwise. The historical matching success rate represents the ratio of the number of times this vehicle model and the corresponding charging station have been successfully matched in historical data to the total number of matches. As for the charging station load rate, it is to avoid using high-load parts. In this case, the calculation will use 1 minus the real-time load rate to prioritize matching low-load charging stations. Finally, the maximum allowable time difference value represents the maximum allowable time difference when the charging station and the vehicle's charging status change, representing the maximum allowable time for the corresponding charging station and vehicle status response.

[0054] The maximum allowable time difference is determined based on the type of charging pile. For example, the maximum allowable time difference for DC fast charging piles is 30 seconds, or it can be calculated by statistically analyzing the 95th percentile time of historical state synchronization for that type of pile. This 95th percentile represents the time value at the 95th percentile after sorting the state synchronization time from smallest to largest under normal circumstances. This value is close to the maximum allowable time difference for state updates under normal circumstances. For AC slow charging piles, it is 60 seconds. This value can be retrieved in real time through the charging pile controller, or it can be obtained by using the 95th percentile time of historical state synchronization.

[0055] The implementation of step S2 also includes: S24, matching each vehicle with the charging pile according to the association weight, using the matched data as nodes, connecting the charging status changes of each node, and constructing a charging status distribution map.

[0056] The progress of the final charging status distribution diagram can be explained using the following example: waiting → charging; vehicle-side status: plugging in → charging; pile-side status: idle → charging; the pile has current output, at which point the node changes from the waiting combination to the charging combination.

[0057] Charging in progress → Charging complete; Vehicle-side status: Charging in progress → Charging complete; Charging pile-side status: Charging in progress → Charging complete; Battery level: Vehicle SOC ≥ 95%, Node combination from charging in progress → Combination complete.

[0058] Charging in progress → Abnormal interruption, vehicle-side status unchanged, but charging pile-side status: Charging in progress → Fault; current drops sharply to 0, no gun disconnection action; node combination from charging in progress → Abnormal combination; these data are used to record the status changes of each matched vehicle and charging pile to complete subsequent vehicle tracking and processing.

[0059] Preferably, when constructing the charging status distribution map, the implementation method further includes: traversing the association weights of all vehicles and charging piles, taking the maximum sum of the total association weights as the objective, and treating the combination of vehicles and charging piles under the objective as the data of the corresponding nodes in the charging status distribution map.

[0060] At this point, the Hungarian algorithm will be used to match multiple charging piles and multiple vehicles. This process is based on the node settings for waiting vehicles and idle charging piles. For situations that have already occurred, such as charging in progress or abnormal occupancy, the relevant charging piles and vehicles will be directly combined into nodes of the charging status distribution map to present the charging status of each vehicle and charging status. Then, by tracking the data of the charging status distribution map, the scheduling objects will be filtered and processed to complete the selection of the charging order under the charging path.

[0061] In one embodiment of the present invention, a valid charging event in step S3 represents a portion of the complete charging process, such as waiting to charge → plugging in → charging → charging complete → leaving. A valid charging event here at least completes the plugging in and charging stages, and the vehicle-charging pile status displays are synchronized, with no other obvious charging interruptions or charging faults. Then, based on the updates to the valid charging events, the vehicle's status is tracked. Conversely, an invalid charging event indicates problems such as plugging in failure, charging interruption, or invalid occupancy, representing resource waste or charging pile malfunction. In such cases, the vehicle's trajectory and charging status changes need to be cross-updated according to the results of the valid charging events, and the updated results are output to complete the vehicle's status tracking.

[0062] like Figure 4 As shown, the implementation of step S3 includes: S31, converting the charging state changes of each node in the charging state distribution diagram into a state sequence, and determining whether the current state sequence conforms to the charging process.

[0063] S32, if the conditions are met, the corresponding node in the current vehicle charging status distribution map is lit up as a valid charging event according to the updated status of the current vehicle charging status.

[0064] S33, and according to the multiple vehicles corresponding to the valid charging event, outputs the state tracking result according to the location of the charging pile where each vehicle is located.

[0065] S34 If the condition is not met, retrieve the corresponding vehicle charging status and perform cross-updates based on the charging status, then synchronize the updated data to the status tracking results.

[0066] During cross-updates, the essence is to continuously record the charging status of the corresponding vehicle and charging pile, record abnormal scenarios, and continuously update the status tracking results in a traceable form. This allows for direct determination of the current usage status of the charging pile and the vacancy status of the relevant parking spaces during subsequent scheduling and allocation.

[0067] When performing cross-update in step S34, the implementation method includes: if there is an anomaly in the current charging event, after initializing or rematching the current vehicle and the corresponding charging pile, the charging state corresponding to the current charging event is restored to the charging state before the deviation occurred, and the changes in the charging state between the current vehicle and the charging pile are identified simultaneously.

[0068] During cross-updates, the approach to handling anomalies is to quickly recover and release resources that were in the event of an anomaly, so that these resources can be used for subsequent viewing and matching.

[0069] In one embodiment of the present invention, the objects to be dispatched include all vehicles waiting to be charged and charging piles that are idle or about to complete charging.

[0070] In these cases, vehicles waiting to be charged are found during the status tracking process in steps S2 and S3; vehicles that need to be re-matched are found in the abnormal handling part of step S3, which indicates that the charging gun failed / charging was interrupted; and vehicles that have been waiting again after being actively canceled by the user in step S2 are found based on the trigger conditions. Vehicles that meet any of these conditions are considered to be dispatched.

[0071] As for the charging piles to be scheduled, these represent those that are idle, about to finish charging, and those that are idle after users cancel charging. These are the charging piles that are currently the main ones to be scheduled. However, those that are about to finish charging will release the corresponding resources according to the estimated completion time and the actual departure time, and perform relevant scheduling adaptation. At this time, the scheduling process is a pre-estimation process to predict the portion that can be scheduled at the corresponding time, as well as the charging order of these vehicles and charging piles.

[0072] like Figure 5 As shown, the implementation of step S4 includes: S41, for each identified object to be scheduled, according to the number identified at each time, statistically analyze the data pairs that match the vehicles and charging piles in the objects to be scheduled; the method of forming data pairs is consistent with the method of generating corresponding nodes in the charging status distribution map, and the corresponding vehicles and charging piles are matched by using the method with the largest total association weight.

[0073] S42, based on each identified data pair, records the time each data pair occupies at the corresponding charging station in a time period distribution format.

[0074] S43, sort all the occupied time according to the corresponding timestamp, record the distribution of the occupied time of the objects to be scheduled at each time, and retrieve the intersection of the processing at adjacent time points, and take the objects to be scheduled corresponding to the intersection as the first priority.

[0075] S44. Select the data outside the intersection and sort them from largest to smallest according to their correlation weight. The data before the third quartile after sorting are regarded as the second priority, and the remaining data are regarded as the third priority. Output the corresponding scheduled objects according to the priority.

[0076] When filtering vehicles to be scheduled, the system first analyzes the available vehicles and charging stations at each time point. This time point can be set to a 5-minute interval, or other time intervals can be selected, such as a 15-minute interval during off-peak hours when fewer vehicles enter to prevent waste of system resources; and a 1-minute interval for special scenarios where malfunctions occur. Alternatively, the system can filter vehicles for scheduling when they enter and select a 10-second or 30-second time interval to process real-time vehicles, quickly update the matching of each charging station with the corresponding vehicle, and avoid vehicles waiting idly to count the available vehicles and charging stations at each time point.

[0077] Secondly, the extracted data pairs will be compared and arranged according to the situation acquired at each time point. If the occupancy time of some vehicles after matching with the charging pile is the same in the prediction of adjacent time points, it means that this part of the data has no conflict and belongs to the priority scheduling target, and can be guided to charge first.

[0078] For other priority parts, this data will be matched and processed in subsequent moments, and the corresponding vehicles will be assigned according to the priority. At the same time, the adjacent moments of the current identification represent the processing status of two consecutive moments, and so on to complete the identification and processing of their intersection.

[0079] At this point, the first priority indicates vehicles that can be directly matched with the corresponding charging pile, can normally enter the charging process, and have no overlap in subsequent occupancy time; the second priority indicates that there is a certain waiting time, and the corresponding vehicle is compatible with the charging pile at the corresponding location. As for the third quartile, it represents the meaning of 75%, and this value can distinguish vehicles that can charge at the corresponding charging pile location in most cases; as for the third priority, it represents non-standard vehicles, which generally represent vehicles that rarely charge at the current charging pile model. These vehicles have few normal charging times in historical data, and their vehicle compatibility with the corresponding charging pile may be insufficient.

[0080] In one embodiment of the present invention, step S5 is implemented by: based on the node of the object to be scheduled in the charging state distribution map, recording the state probability of the node when the charging state changes, each node represents the change in its charging process, and at this time, the probability values ​​of waiting → plugging in, plugging in → charging, charging → completed, and completed → driving away are recorded respectively.

[0081] The probability of the state from "waiting" to "plugging in" is represented by the proportion of the number of times the same type of vehicle has been charged normally at the corresponding type of charging pile in recent historical data. At the same time, since the time difference between the vehicle and the charging pile in the corresponding charging state is taken into account when the state from "plugging in" to "charging", the ratio of the time difference value obtained in step S23 to the maximum allowable time difference value is multiplied by the proportion of the number of normal charging times in the total number of times, and the result of the multiplication is regarded as the probability of the state from "plugging in" to "charging".

[0082] Then, during charging → after completion, the stability of the charging process needs to be considered. This means that the proportion of normal charging times out of the total number of times is considered as the probability of the current state after excluding the number of faults.

[0083] Finally, during the process of charging completion and driving away, it is more important to focus on the speed of user settlement. This means setting a probability value based on whether the user occupies the charging station for a long time. For example, if we choose the average time spent at the corresponding charging station after charging is completed in historical data, we can use this average time as the standard to check whether the user has exceeded the time limit. When it exceeds the average time, the corresponding vehicle is considered to have exceeded the time limit. We can also calculate the ratio of the number of times the current model did not exceed the time limit in historical data to the total number of times. This ratio is regarded as the probability of the current state.

[0084] The state probabilities are calculated jointly according to the normal charging process, and the joint probabilities are output as the schedulable probability for each schedulable object.

[0085] The joint probability requires calculating four probability values: waiting → inserting the gun, inserting the gun → charging, charging → completed, and completed → departing. Then, based on the weights assigned to each transition, the cumulative value of the weighted probability values ​​is used as the output joint probability.

[0086] At this point, the weighted calculation of probability values ​​can also emphasize the impact of each state transition on the overall charging allocation by using the weights as exponents. Finally, each probability value is multiplied in turn to complete the calculation of the joint probability. The weights of these four stages can be selected as 0.15, 0.4, 0.3, and 0.15, respectively. This emphasizes the process of plugging in → charging → completion. This part is the main part of the overall verification to identify whether the vehicle is charging normally after being dispatched to the corresponding charging station, and whether there is a relative situation of charging interruption. As for the stages of leaving and plugging in, the impact of problems is relatively small. Leaving only affects the release of the charging station position and does not affect the current charging scheduling. As for the plugging in stage, it can be retried and a charging station can be obtained again through the waiting scheduling method. The impact of problems at this time is relatively small.

[0087] The higher the joint probability obtained later, the more reliable the current scheduling method is, and the charging can be carried out using the current scheduling order or location. If the probability value is lower, it means that it is unreliable, indicating that it is difficult to charge the corresponding vehicle at the relevant location, and other methods are needed to intervene, which is a part that needs to be optimized.

[0088] In one embodiment of the present invention, when sorting the objects to be scheduled in step S6, the sorting process is based on the priority obtained in step S4. Vehicles corresponding to the first priority are allocated first. The first priority is the part that has no time conflict in continuous time and can be directly scheduled to the corresponding charging pile. Then, the second priority and third priority parts need to be identified according to the schedulable probability. The schedulable probability is mainly to solve the conflict handling of the same priority. For example, for vehicles waiting in the second priority, if there is a time conflict when the scheduling object is extracted, they are allocated according to the schedulable probability, and the part with the higher schedulable probability is processed first. The remaining vehicles will be updated with the third priority vehicles, and the updated data will be redistributed to the first priority and second priority to schedule and configure the vehicles one by one. Finally, the charging sequence configuration in the multi-vehicle and multi-charging pile scenario will be presented.

[0089] like Figure 6As shown, the implementation of step S6 includes: S61, obtaining the priority of the object to be scheduled, and allocating the current object to be scheduled according to the priority; the priority of the object to be scheduled is divided into first priority, second priority and third priority.

[0090] S62 directly assigns charging order to vehicles with the highest priority, thus obtaining the charging order under the highest priority.

[0091] S63, when allocating vehicles with the second priority, if there is a time conflict between the vehicles to be scheduled, the charging order of the other vehicles to be scheduled is reduced by one based on the vehicle with the highest schedulable probability, so as to obtain the charging order of the second priority vehicles.

[0092] S64, when allocating vehicles of the third priority, updates the data of the third priority vehicles and the vehicles of the second priority that have not been allocated, and obtains the updated charging order based on the schedulable probability and waiting time.

[0093] When using schedulable probability and waiting time as sorting criteria, the charging order at each charging station is obtained by first sorting according to schedulable probability from largest to smallest. If the schedulable probabilities are the same, the charging order is obtained by sorting according to waiting time from largest to smallest.

[0094] Because each vehicle is matched with a charging station and its occupancy time is recorded when setting priorities, the charging order only needs to be recorded for the parts with time conflicts under different priorities. The acquired data list is then updated, and these data are updated again considering the schedulable probability and waiting time. This allows the remaining vehicles of the second priority and the vehicles of the third priority to be re-updated to the positions of the first priority, etc. This makes the real-time effect of priority allocation processing stronger, more suitable for reserved parking and real-time parking, avoids conflicts and invalid allocation, and finally obtains a relatively stable charging order.

[0095] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A service area vehicle charging analysis processing method based on video analysis, characterized by, Comprise: S1, identify the vehicle, parking space and vehicle access path corresponding to each frame image in the charging pile area video, and regard each vehicle as a to-be-processed charging event; S2, connect the vehicle and the charging pile in the charging event according to the vehicle-side charging state and the charging pile-side charging state of the current charging event, and construct a charging state distribution diagram; S3, determine whether the current charging event is a valid charging event according to the update of each vehicle in the charging state distribution diagram, and track the state of the current vehicle; S4, based on the result of the current vehicle state tracking, screen the to-be-scheduled objects of the vehicle and the charging pile in the current charging pile area video, and determine at least one to-be-scheduled object; S5, combine the to-be-scheduled objects in the nodes in the charging state distribution diagram, and determine the schedulable probability of each to-be-scheduled object; S6, sort the to-be-scheduled objects based on the schedulable probability, and determine the charging sequence matched by each vehicle; The implementation of step S2 further comprises: S21, determining whether there is a charging state change in the current charging event, and identifying the connection relationship between each vehicle and the charging pile when there is a charging state change; S22, obtaining the charging pile associated with each vehicle, synchronizing the charging time and the remaining power of each vehicle, and calling the minimum time interval of each vehicle under the charging state conversion; S23, in response to the minimum time interval under the charging state conversion, recording the time difference between the time point of the vehicle charging state change and the time point of the charging pile charging state change at the corresponding time; taking the time difference as the association weight of the corresponding vehicle and the charging pile; S24, matching each vehicle and the charging pile according to the association weight, taking the matched data as the node, connecting the charging state change of each node, and constructing the charging state distribution diagram; The implementation of step S5 comprises: based on the node of the to-be-scheduled object in the charging state distribution diagram, recording the state probability of the node when the charging state changes; calculating the joint probability according to the normal charging process, and outputting the joint probability as the schedulable probability of each to-be-scheduled object.

2. The service area vehicle charging analysis processing method based on video analysis according to claim 1, characterized by, The implementation of step S1 comprises: S11, by image segmentation, label the occupancy state of each parking space and the parking state of the vehicle in the charging pile area video, and form an association link with the vehicle, the parking space and the vehicle access path; S12, based on the obtained association link, combine the parking time of each vehicle, and preliminarily classify each vehicle, and screen out the charging event to be processed for each vehicle.

3. The service area vehicle charging analysis processing method based on video analysis according to claim 1, characterized by, The implementation of step S21 further comprises: For the charging state change of each vehicle, obtain the trigger condition of the charging state change of each vehicle, fill the trigger condition into each node in the charging state distribution diagram according to three dimensions of equipment failure, user active operation and time period resource pressure, and obtain a plurality of nodes corresponding to the trigger condition; Based on the plurality of nodes corresponding to the trigger condition, determine the vehicle-side charging state and the charging pile-side charging state of the corresponding node, set a label for it, and synchronize to the charging state distribution diagram.

4. The service area vehicle charging analysis processing method based on video analysis according to claim 3, characterized by, When constructing the charging state distribution diagram, the implementation further comprises: The association weight of all vehicles and charging piles is traversed to maximize the sum of the total association weight, and the combination of the vehicle and the charging pile under the target is regarded as the data of the corresponding node in the charging state distribution diagram.

5. The service area vehicle charging analysis processing method based on video analysis according to claim 1, characterized by, The implementation of step S3 includes: S31, the charging state change of each node in the charging state distribution diagram is converted into a state sequence, and it is judged whether the current state sequence conforms to the charging process; S32, if it conforms, the corresponding node of the current vehicle in the charging state distribution diagram is lit as an effective charging event according to the update of the charging state of the current vehicle; S33, and according to the corresponding multiple vehicles under the effective charging event, the state tracking result is output according to the position of the charging pile where each vehicle is located; S34, if it does not conform, the charging state of the corresponding vehicle is called, and the charging state is cross-updated, and the updated data is synchronized to the state tracking result.

6. The service area vehicle charging analysis processing method based on video analysis according to claim 5, characterized in that, When cross-updating in step S34, the implementation includes: If there is an exception in the current charging event, the charging state corresponding to the current charging event is restored to the charging state before the deviation occurs after initialization or re-matching the operation of the current vehicle and the corresponding charging pile, and the charging state change between the current vehicle and the charging pile is identified synchronously.

7. The service area vehicle charging analysis processing method based on video analysis according to claim 1, characterized by, The implementation of step S4 includes: S41, for each identified to-be-scheduled object, the number of to-be-scheduled objects matched by the vehicle and the charging pile is counted according to the number of each time point; S42, according to each identified data pair, the occupation time of each data pair at the corresponding charging pile is recorded in the form of time period distribution; S43, sort all occupation times according to the corresponding time stamp, record the distribution of the to-be-scheduled object about the occupation time at each time point, and call the intersection of the adjacent time points, and take the to-be-scheduled object corresponding to the intersection as the first priority; S44, select the data outside the intersection, and sort the data from large to small according to the association weight, take the data located before the third quartile point after sorting as the second priority, and take the remaining data as the third priority, and output the corresponding to-be-scheduled object according to the priority.

8. The service area vehicle charging analysis processing method based on video analysis according to claim 1, characterized by, The implementation of step S6 includes: S61, the priority of the to-be-scheduled object is obtained, and the current to-be-scheduled object is distributed according to the priority; the priority of the to-be-scheduled object includes the first priority, the second priority and the third priority; S62, the first priority vehicle is directly distributed to obtain the charging sequence under the first priority; S63, when the second priority vehicle is distributed, if there is a time conflict in the to-be-scheduled object, the charging sequence of other to-be-scheduled objects is reduced by one based on the to-be-scheduled object with the maximum schedulable probability, and the charging sequence of the second priority is obtained; S64, when the third priority vehicle is distributed, the third priority vehicle and the vehicle in the second priority which has not been distributed are updated, and the updated charging sequence is obtained according to the schedulable probability and the waiting time as the sorting reference.

Images