Electric vehicle charger safe waiting management method and system based on improved sequential network

Abstract

The invention discloses an electric vehicle charger safety waiting management method and system based on an improved sequential network. The method comprises the following steps: S1, recording vehicle remaining capacity, accessory power consumption, a thermal risk tag and service pressure; s2, dividing active units according to a queuing order and service pressure, constructing an energy consumption and heat dissipation double-layer topology, and updating a weight in real time; s3, writing the double-layer topology and the historical state into a time sequence buffer area, traversing edge weight iteration vehicle electric quantity and battery temperature, and calculating a safety window; s4, calculating the waiting time of each vehicle according to the safety window and the queuing data, and generating a risk sequence; s5, the vehicles of the same unit are locally rearranged preferentially according to the urgency degree, and a correction queue is formed after fairness and connectivity verification; and S6, intercepting the correction queue according to idle stations, completing vehicle and station mapping, and issuing a charging instruction after reachability and power verification. The method has great superiority in microscopic safety evaluation, dynamic topology increment maintenance and real-time rearrangement giving consideration to safety and fairness.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charger safety management technology, and in particular to an electric vehicle charger safety waiting management method and system based on an improved timing network. Background Technology

[0002] With the rapid growth in the number of new energy vehicles, large and medium-sized public energy replenishment stations generally exhibit high vehicle entry density, long queue lengths, and fast workstation turnover in high-frequency operation scenarios such as urban distribution, sanitation, and commuting.

[0003] The industry still largely uses the first-come, first-served charging scheduling method based on arrival time or simple reservation order. This approach ignores the additional energy consumption of vehicles during the queuing process, the continuous power consumption of accessories such as air conditioning, and the micro-safety factors such as the obstruction of battery heat dissipation under extreme weather conditions. This leads to risks such as vehicle power outages, thermal imbalances, and even queue congestion during peak periods.

[0004] For example, existing literature on charging station training simulation methods has pointed out that training simulations can be conducted based on existing fault types and causes of chargers, providing theoretical support for electric vehicle charging equipment operation and maintenance training platforms. Their technical service reports further summarize the impact of factors such as ambient temperature, operating methods, single charging duration, and equipment service life on charger reliability, forming a relatively systematic list of fault factors. These findings demonstrate that to truly achieve safe, efficient, and fair operation of chargers, it is essential to integrate vehicle battery degradation mechanisms, queuing topology changes, and charger fault influencing factors into a single online decision-making logic.

[0005] Existing queuing systems often only focus on throughput or average waiting time, lacking dynamic topology incremental maintenance and safety-priority reordering strategies. They also fail to feed back the fault impact weights obtained from simulations to the scheduling level in real time, resulting in lagging safety assessments, frequent queue-jumping disputes, and unavoidable path connectivity conflicts.

[0006] Therefore, there is an urgent need to develop a new electric vehicle charger safety waiting management method that can collect multi-source data in a short time scale, dynamically update vehicle-site topology, quantify safety margin, and take into account both high-risk priority and first-come-first-served fairness, so as to provide reliable support for intelligent scheduling and equipment operation and maintenance of charging stations. Summary of the Invention

[0007] To address the aforementioned issues, this invention discloses a method and system for safe waiting management of electric vehicle chargers based on improved temporal networks. Addressing the shortcomings of existing technologies—such as the inability to fine-grainedly assess the combined risks of power decay and heat dissipation obstruction, the lack of real-time updated topological relationships as vehicles move in and out, and the difficulty in balancing safety priority with first-come-first-served fairness—this invention proposes a technical solution that includes real-time acquisition of multi-dimensional vehicle energy consumption and environmental data, construction of a two-layer vehicle-field dynamic graph, online calculation of maximum waiting time and high-risk ranking, local minimum necessary rearrangement based on active coupled units and cross-unit fairness verification, and linkage station mapping and path connectivity. This invention can accurately predict power outage or thermal runaway boundaries, promptly protect high-risk vehicles, maintain overall queuing fairness, and improve station operation efficiency and safety during peak periods.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] The electric vehicle charger safety waiting management method and system based on an improved timing network according to embodiments of the present invention includes the following steps:

[0010] S1. Under the unified time reference of the vehicle's front wheels crossing the station entrance, a unique queue number is generated for the entering vehicle. The initial state of charge of the vehicle, the power of the air conditioner, the power of the external lights and the power of the information terminal are collected and summed to obtain the power consumption of the accessories. At the same time, the vehicle body temperature and the ambient temperature inside the station are acquired in real time and the temperature difference is calculated. The temperature difference is discretized into thermal risk tags according to a preset threshold. Then, the number of available charging stations is read to calculate the service pressure value. Finally, the combination is used to obtain the safe waiting feature vector.

[0011] S2. Based on the queue number, create a vehicle index table and establish a set of topological nodes. Divide active coupling units according to the queue position difference and service pressure difference. Write vehicle-vehicle connections within the unit according to the vehicle distance and occlusion status, and calculate the energy transfer weight and heat dissipation weight respectively. At the same time, establish vehicle-field connections and map the safe waiting feature vector to the node to form a two-layer topology graph that is updated incrementally as the vehicle moves in and out.

[0012] S3. Load the energy transfer graph, heat dissipation coupling graph, and node features into the timing buffer, traverse adjacent edges according to a preset period, deduct the remaining power of the node using energy consumption weight and accessory power consumption, correct the node battery temperature using heat dissipation weight and thermal risk label, and generate an energy-temperature trajectory by combining the energy consumption compensation amount output by the gating sequence unit; obtain the maximum waiting time for each vehicle by integrating the trajectory with the minimum safe power threshold and mark high-risk nodes.

[0013] S4. Based on the maximum waiting time and the path travel time and service pressure consumption time provided by the scheduling module, calculate the safety margin and automatically add temperature difference compensation according to thermal risk to obtain the waiting time field; generate a real-time risk sequence by sorting all vehicles in ascending order according to the field.

[0014] S5. Align the waiting time field with the active coupled unit number, extract the unit urgency label, and perform minimum necessary pre-insertion on vehicles within the unit using a local rearrangement strategy that prioritizes urgency and restricts displacement. Then, obtain the global sorting queue and immediate service label through cross-unit fairness check and path connectivity check.

[0015] S6. Based on the real-time number of available workstations, the global sorting queue is windowed to generate the work sequence for the next charging cycle. A vehicle-workstation pairing table is generated according to the vehicle size, charging gun specifications and workstation location mapping. The reachability and power compatibility of the robotic arm are verified in sequence. After the verification is passed, the charging command is synchronously issued to the workstation allocation module, the station information screen and the driver terminal.

[0016] Optionally, step S1 includes the following sub-steps:

[0017] S11. Set the time benchmark for the front wheels of vehicles entering the station crossing the station entrance as a unified time reference. Assign a unique queue number to the vehicle. And simultaneously record the remaining battery power. Instantaneous power of air conditioner Instantaneous power of external lights Instantaneous power of information terminals ;right , , Summing yields the continuous power consumption of the attachment. triples Write the starting point of the power consumption record table.

[0018] S12, in time base Internal synchronous acquisition of vehicle body temperature With the ambient temperature inside the station Calculate the difference ;when Assign a cold temperature zone label ,when Assign a temperature zone label ,when Assign hot zone labels ;

[0019] By writing the corresponding labels into the thermal risk table, the impact of environmental temperature difference can be discretized, giving clear weight to the accelerated decay of battery usable capacity caused by high temperature or severe cold in the safety assessment.

[0020] S13. Read the number of available charging stations. Combine it with the vehicle queue number Compare and calculate service pressure values. ;Will With the starting point of the power consumption record , and heat risk labels or or Concatenate to generate a safe waiting feature vector ;

[0021] Optionally, step S2 includes the following sub-steps:

[0022] S21. The unique vehicle number obtained in step S1 is... Rather than real-time queuing order Bind and create a vehicle index table. The graph node set is established; any new vehicle entering the queue triggers a node insertion transaction, and any vehicle leaving the queue triggers a node deletion transaction. Both types of transactions are appended with a timestamp. Write back to the index table.

[0023] S22, Index table According to queue order Perform a scan from left to right, first taking the positional difference between two adjacent vehicles. Difference in service pressure Perform parallel judgments;

[0024] If both are not greater than the site activity threshold Then, the two vehicles are assigned to the currently active coupling unit. Otherwise, close immediately. And start the next unit. This continues until the entire train of cars has been divided.

[0025] By using linear grouping, vehicles that are truly under the same crawling rhythm and similar workstation competition pressure are brought together into a unified physical space.

[0026] Subsequently, the real-time vehicle distances within each unit were read. With obstruction of signs ,when Below the slowdown threshold and At the time in the vehicle and Write the connecting edges between them Otherwise, remain disconnected;

[0027] The operation explicitly represents the nearest neighbor relationships most sensitive to the impact on energy consumption and heat dissipation as vehicle-vehicle subgraphs. .

[0028] Then check the first car in the queue With the virtual node at the entrance Write the edges For the last train Virtual node of the nearest available workstation Write the edges This constitutes a vehicle-field subgraph. .

[0029] final, and Vertical stacking forms a two-layer topology It is updated incrementally as the vehicle moves in and out, and any changes to a node or connection are written back with a timestamp.

[0030] S23, Topology Any vehicle connected to the edge Execute energy consumption weight generation: Adjust the energy consumption levels of vehicles on both sides during low-speed driving. Power consumption levels with accessories The energy consumption levels are obtained by superimposing them. The difference between this level and the queuing order Obtain the power transfer weight by weighted combination ;

[0031] The weights are updated in real time according to the creep status and the load level of the surrounding vehicles, directly depicting the intensity of the energy impact caused by the movement and continuous power consumption of adjacent vehicles.

[0032] S24. For the same connected edge Read vehicle thermal risk labels Based on the combination rules, it is mapped to the heat dissipation coupling level. According to vehicle spacing With the wind direction coefficient within the station Graded comparison, Converted into heat dissipation weight And write the edge attributes; when the queue is compact and the environmental temperature difference increases. Automatic lifting, weight reduction when spacing is increased or ventilation is enhanced; at the same time, additional fixed heat dissipation edge coverage is added to vehicles adjacent to buildings or sunshade facilities to cover potential heat sinks or heat sources.

[0033] Based on dynamic heat dissipation weights, the system captures the tendency of sudden capacity degradation in batteries under extreme weather conditions and reserves additional safety margins in the ranking process.

[0034] S25. Vehicle Index Table As the unique primary key, the set of weights for transferring electricity is used. With heat dissipation weight set Node alignment is performed, and a cross pointer structure is used to map the two sets of connected edges;

[0035] When vehicle additions or deletions, or changes in vehicle distance or load are detected, edge additions, removals, or weight overwrites are immediately performed and written to memory in an incremental manner to avoid decision lag caused by full map recalculation.

[0036] S26. Record the starting point of the power generation generated in step S1. Hot risk labels Service pressure value Mapping nodes to their indices to form a four-dimensional feature vector. ;

[0037] Optionally, step S3 includes the following sub-steps:

[0038] S31. The power transfer diagram generated in step S2 is used. Heat dissipation coupling diagram With the four-dimensional feature vector of the node Simultaneously loaded into the timing buffer Treating the buffer zone as a queuing scenario at any given time The complete cross-section;

[0039] Then take the time of the previous cycle. Output of remaining battery power for each vehicle With battery temperature The initial values ​​are written into the same buffer to form a spliced ​​page of historical state and current state.

[0040] S32, Vehicle node Update the remaining battery level:

[0041] The first step is to process all data in the power transfer diagram that are related to... Connected edges Perform a traversal, sequentially reading the edge attribute energy decay labels. The label was calculated during edge construction, and its physical meaning is: during this round of slow creeping, due to... Adjacent vehicles The increased power consumption per unit time due to factors such as reduced vehicle spacing, frequent starts and stops, and changes in inter-wheel airflow resistance, is expressed in units of [missing information]. The larger the label, the tighter the queue and the stronger the energy interference.

[0042] The second step is to read the node's own power consumption. The value is derived from the sum of the real-time power consumption of the vehicle's air conditioning, lights, and information terminal, and the unit is also [unit missing]. .

[0043] The third step is to combine all adjacent edges Perform summation and combine with Add them together to get the total amount of electricity that the vehicle must deduct in this cycle.

[0044]

[0045] in, Indicates the time interval The electricity amount deducted internally. Then use directly... Update the node's remaining battery power field and write it back to the buffer.

[0046] By unifying the additional power consumption caused by driving with the continuous power consumption of vehicle accessories into a one-time deduction, the actual power consumption experience of drivers when waiting in slow queues is reflected. The operations team only needs to check... This allows you to determine how quickly the vehicle's battery is depleting and how much battery remains available.

[0047] S33, Vehicle node Update the battery temperature:

[0048] The first step is to perform a thermal coupling analysis on all components in the thermal coupling diagram. Connected edges Perform a traversal, reading the heat dissipation weight of each edge attribute. .

[0049] The weights are generated by a combination of vehicle distance, wind direction, and thermal risk rules between the two vehicles, reflecting the influence of neighboring vehicles on each other. The intensity of the effect of heat dissipation or heat accumulation, in units of The higher the weight, the more obstructed the heat dissipation or the more obvious the heat accumulation.

[0050] The second step is to read the node's own hot risk label. The label values ​​are categorized based on the temperature difference between the vehicle and the environment. A positive value indicates a high-temperature zone, and a negative value indicates a low-temperature zone. The unit is also 1. It is used to describe the natural upward or downward trend of battery temperature caused by the environment.

[0051] The third step is to combine all adjacent edges Perform summation and combine with Add them together to get the change in battery temperature during this cycle.

[0052]

[0053] in, Indicates the time interval The expected rise or fall in internal battery temperature. Update the node temperature field.

[0054] Based on the above treatment, the heat accumulation caused by close vehicle spacing and unfavorable wind direction is combined with the natural temperature rise and fall caused by high temperature or severe cold environment, and directly reflected in the The dispatch system can then detect vehicles with excessively high or low temperatures in real time, which may lead to drastic changes in the available battery capacity.

[0055] S34. The power deduction sequence generated in step S32 is processed. Compared with the temperature change sequence generated in step S33 The data is concatenated according to the vehicle number and written into the gate control input area. and keep up with the times Consistent timestamps;

[0056] S35, Select the nearest gated input area Wheel , Vehicle queuing order Combined into a multidimensional time slice sequence, and sent to the entry control sequence unit. The unit compares the current sequence with the usual pattern through internal switches. Once it detects abnormal fluctuations caused by a sudden increase in air conditioning speed, short-term acceleration, or changes in airflow within the station, it calculates an additional compensation amount. And output it to the energy consumption compensation stack.

[0057] S36, Output the compensation stack With explicit deduction Add them together to get the complete energy consumption in the short term. ;according to and Generate a power change curve, along with the latest temperature. Write to a circular cache of node history and use rolling over to retain the most recent ones. The system generates a record; by integrating rule-based calculations and temporary corrections, a traceable energy-temperature dual trajectory is formed, providing continuous input for safety window calculations and avoiding decision distortion due to trajectory breakpoints.

[0058] S37. Integrate the power change curve over the time axis and sum it with the current remaining power at the node. Comparison based on the minimum safety threshold of the battery Determine the maximum waiting time The data is then written to the node safety window field; the field presents the vehicle's remaining buffer under the non-shutdown constraint in minutes, avoiding the need for dispatchers to manually calculate between multi-dimensional data.

[0059] S38, For all nodes A threshold scan of the field will reveal fields that fall below the warning threshold. Vehicles marked as high-risk nodes Then Output.

[0060] Optionally, step S4 includes the following sub-steps:

[0061] S41. Store the power change curve in the node's historical records in minute sequence. Current remaining power of the node Perform initial overlay; if the overlay value exceeds the station's minimum safety threshold... Then the time pointer Proceed to the next minute, then take the next segment of the curve and continue accumulating;

[0062] When it first appears The vehicle is locked immediately. The ultimate waiting point under uninterrupted power supply conditions The segment-by-segment comparison directly aligns the vehicle's actual power consumption rhythm with a unified safety baseline.

[0063] S42. Combine the maximum possible time to wait with the estimated actual turn time into a simple, on-site readable statement of "how many more minutes can I wait?" and complete it in four sequential actions:

[0064] First, read the limit waiting endpoint obtained in step S41. (Meaning: vehicle) If power consumption continues, (Always touch the minimum safe battery level), and simultaneously read the time when the vehicle enters the queue. The difference between the two This refers to the total number of minutes the vehicle can remain in the station, excluding queuing time.

[0065] Second, perform a one-way traversal of the topology generated by S22 in the direction of exit: along the vehicle... The path from the vehicle ahead to the available workstation is tracked node by node, collecting real-time distance between vehicles, the active unit number, and the workstation release rate returned by the scheduling module. The scheduler adds the approximate time required to move one meter to the inertia of the active unit, linearly accumulating the total travel time required for the vehicle to travel from its current position to the nearest available workstation. .

[0066] Third, read the service pressure value within the node. (Definition: The average additional waiting time for each additional vehicle in a queue), compare it with... Adding them together gives a more accurate prediction of queuing time based on the actual situation. Then, the queuing time is subtracted from the maximum allowable dwell time to obtain the original safety margin; if this margin is lower than the station's preset thermal threshold... This indicates that vehicles are more likely to lose range in high or low temperatures, according to the thermal risk label. Automatic addition of temperature difference compensation The compensation range increases with the severity of the environment.

[0067] Fourth, incorporate safety margin and temperature difference compensation into the formula.

[0068]

[0069] in, This is the number of minutes you can wait, rounded down. Waiting for the end at the limit; This refers to the vehicle's arrival time. For estimated travel time; Service pressure takes time; For temperature difference compensation; To ensure the result is an integer number of minutes, round down.

[0070] Finally Write nodes can wait for the duration field and overwrite the old value.

[0071] Through this process, information on vehicle battery limit, actual queuing pace, and ambient temperature difference is compressed into a single intuitive indicator. Based on this, dispatchers can quickly identify high-risk vehicles that urgently need to cut in line or enter the station immediately, and on-site personnel can also understand at a glance how much longer they can wait.

[0072] S43. Number the vehicle With waiting time The paired data is written to the queuing node table, overwriting the old value. Immediately after this, a full ascending scan of the table is triggered, performing a scan on all vehicles. Sorting data from smallest to largest temporarily generates a real-time risk sequence. The sequence naturally pushes the vehicles most prone to power outages to the front, ensuring that any subsequent scheduling decisions are based on the latest risk profile and avoiding time lag between the field and the algorithm.

[0073] S44, Real-time risk sequence The data is pushed to the dispatch center, which reads the number of vehicles at the front end of the sequence and compares it with the number of currently available workstations. The system matches and marks successfully matched vehicles as those to be served immediately. Then, work station instructions are simultaneously issued on the information screens in the station, the dispatch terminal, and the driver's mobile terminal. On-site guides and drivers drive according to the instructions. Through this closed loop, the abstract waiting time is directly transformed into an executable action, ensuring that no vehicle will cause traffic congestion or safety hazards due to power depletion or temperature imbalance during the queuing period, while maintaining an overall sense of fairness in the first-come, first-served system.

[0074] Optionally, step S5 includes the following sub-steps:

[0075] S51. Write step S4 back to the node's waitable duration field. Active coupling unit numbering as specified by S22 Perform alignment and write to the node; group by unit number, and execute once for each vehicle within each unit. The algorithm performs an ascending scan, records the minimum value within each cell in real time, and stores this minimum value as a cell urgency label. ;based on In subsequent sorting, it can quickly identify which queue is at the earliest risk of power outage without having to compare the entire table for each vehicle, significantly shortening the response time during peak periods.

[0076] S52, will have the same unit number Vehicles are placed in the same work set, and a local rearrangement is carried out using a priority-based, displacement-constrained strategy: vehicles within the set are arranged according to... Reorder in ascending order, then scan from front to back along the original inbound sequence, only for... Vehicles below the station warning threshold are subject to minimum necessary pre-insertion, and the insertion distance does not cross the cell boundary. Based on the natural constraint of the physical continuity of the coupled cells, high-risk vehicles are given priority protection in this cell, while avoiding rearrangement that could affect adjacent cells and cause chain displacement of the entire train.

[0077] S53. Reconnect the local lists after the partial rearrangement of each unit into a global candidate queue according to the original queuing order, and calculate the displacement difference of each vehicle from its original position to its current position. If the cumulative value of the displacement difference of the entire queue exceeds the station's fairness threshold, then roll back the vehicle with the smallest displacement to its original order according to the principle of priority rollback across units, until the cumulative value falls back to within the threshold. Through the cross-unit fairness check, the system simultaneously satisfies the immediate protection of high-risk vehicles and the overall consideration of the first-come-first-served principle.

[0078] S54. Write the globally sorted queue, which has been verified through fair testing, into the scheduling center, and prioritize queues located at the front of the queue and... Vehicles still below the safety buffer threshold are given an immediate service tag; the virtual connection in S22 is invoked to perform connectivity verification on the driving path from the current first vehicle to the idle workstation. If the scheduling is found to cause path blockage, a micro-switch is triggered until full connectivity is achieved; finally, the queue order, service tag and connectivity results are synchronously published to the workstation allocation module, the station information screen and the driver's mobile terminal.

[0079] Optionally, step S6 includes the following sub-steps:

[0080] S61. The global sorted queue output in step S5. With real-time idle workstations To align, a window truncation method is used to generate the job sequence for the next charging cycle: first, the window width is calculated. ,in This represents the maximum number of parallel workstations that can be operated in a single cycle at the station; then from... Head extraction sequence before Vehicles, forming an execution sequence And write it to the task area;

[0081] S62, Execution sequence The vehicle-workstation pairing table is generated by mapping the vehicle size, charging gun specifications, and physical location of the workstation. The pairing table is then checked in two rounds: the first round checks the reachability of the robotic arm, and the second round checks the power compatibility. If either check fails, the next vehicle is taken from the sorting queue and remapped until the pairing table passes all checks.

[0082] An electric vehicle charger safety waiting management system based on an improved time-series network includes:

[0083] The vehicle entry data acquisition terminal integrates a geomagnetic coil, a millimeter-wave ranging node, and a vehicle-to-pile communication interface. It is used to obtain a unified time reference the instant the front wheels of the vehicle pass the station entrance, and to collect raw data such as the vehicle's remaining battery power, accessory power consumption, vehicle body temperature, and ambient temperature in real time.

[0084] The safety feature generation module sums the vehicle's remaining battery power with the power consumption of the accessories to generate a battery power starting point record, generates a thermal risk label based on the temperature difference between the vehicle body and the environment, and calculates the service pressure value in combination with the number of available charging stations. Finally, the modules are spliced ​​together to form a safety waiting feature vector.

[0085] The two-layer topology construction module divides active coupling units based on vehicle queuing order and service pressure difference, writes vehicle-vehicle connections and vehicle-field connections respectively, adds energy transfer weights and heat dissipation coupling weights to the connections and carries timestamp incremental updates to construct a vehicle-field two-layer topology graph that is dynamically refreshed as vehicles move in and out.

[0086] The timing simulation and compensation module loads the energy transfer diagram, heat dissipation coupling diagram and node characteristics into the timing buffer, traverses adjacent edges to deduct node power, corrects node temperature, and calls the gated sequence unit to output energy consumption compensation for short-term anomalies, generating a continuous energy-temperature trajectory.

[0087] The safety margin assessment module integrates the energy-temperature trajectory to obtain the maximum wait time, calculates the safety margin by combining the path movement time and service pressure time, adds temperature difference compensation according to the thermal risk label, outputs the wait time field and generates a real-time risk sequence.

[0088] The queue rearrangement and workstation allocation module performs minimum necessary pre-insertion based on the urgency label within the active coupling unit, and obtains a globally sorted queue after performing fairness checks and path connectivity checks across units. It then connects to the real-time idle workstation window to capture the work sequence, completes vehicle-workstation pairing, and sequentially passes the robotic arm accessibility and power compatibility checks. After passing the checks, the charging command is synchronously sent to the workstation allocation module, the station information screen, and the driver terminal.

[0089] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0090] 1. By simultaneously collecting multi-dimensional data such as state of charge, air conditioning power, exterior light power, information terminal power, vehicle body temperature, and station ambient temperature at the moment the vehicle enters, and superimposing creep energy consumption and accessory power consumption in real time, a four-dimensional feature vector of power-thermal risk and a two-layer vehicle-field dynamic map are constructed. This enables accurate assessment of the coupling effect of battery available capacity decay and heat dissipation obstruction, overcoming the shortcomings of existing technologies that cannot reflect the real power consumption rhythm based solely on static SOC thresholds, and significantly improving the accuracy of risk identification.

[0091] 2. Incremental edge weight refresh and cross pointer mapping are adopted to avoid full graph recalculation. The time consumption of a single topology update is small. The maximum waiting time can be output in real time when vehicles move forward or backward, the peak vehicle distance changes, or the load of the surrounding area changes suddenly, and the risk of power failure or thermal runaway can be warned in advance, thus solving the problem of warning lag in the existing solution.

[0092] 3. By introducing active coupling unit partitioning and a priority strategy of urgency + displacement-constrained local minimum necessity, coupled with cross-unit fairness verification and path connectivity verification, we can ensure that high-risk vehicles can enter first while maintaining overall fairness of first-come-first-served, thus avoiding driver disputes and channel blockages caused by traditional forced queue jumping. Attached Figure Description

[0093] Figure 1 The flowchart shows the electric vehicle charger safety waiting management method and system based on improved time-series networks proposed in this invention.

[0094] Figure 2 This is a flowchart illustrating the construction topology of the electric vehicle charger safety waiting management method and system based on improved temporal networks proposed in this invention.

[0095] Figure 3 This is a flowchart of the timing buffer of the electric vehicle charger safety waiting management method and system based on improved timing network proposed in this invention.

[0096] Figure 4 This is a schematic diagram of the safety window of the electric vehicle charger safety waiting management method and system based on improved timing network proposed in this invention. Detailed Implementation

[0097] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0098] refer to Figure 1-4 A method and system for managing the safety waiting time of electric vehicle chargers based on improved time-series networks, comprising the following steps:

[0099] S1. Under the unified time reference of the vehicle's front wheels crossing the station entrance, a unique queue number is generated for the entering vehicle. The initial state of charge of the vehicle, the power of the air conditioner, the power of the external lights and the power of the information terminal are collected and summed to obtain the power consumption of the accessories. At the same time, the vehicle body temperature and the ambient temperature inside the station are acquired in real time and the temperature difference is calculated. The temperature difference is discretized into thermal risk tags according to a preset threshold. Then, the number of available charging stations is read to calculate the service pressure value. Finally, the combination is used to obtain the safe waiting feature vector.

[0100] S2. Based on the queue number, create a vehicle index table and establish a set of topological nodes. Divide active coupling units according to the queue position difference and service pressure difference. Write vehicle-vehicle connections within the unit according to the vehicle distance and occlusion status, and calculate the energy transfer weight and heat dissipation weight respectively. At the same time, establish vehicle-field connections and map the safe waiting feature vector to the node to form a two-layer topology graph that is updated incrementally as the vehicle moves in and out.

[0101] S3. Load the energy transfer graph, heat dissipation coupling graph, and node features into the timing buffer, traverse adjacent edges according to a preset period, deduct the remaining power of the node using energy consumption weight and accessory power consumption, correct the node battery temperature using heat dissipation weight and thermal risk label, and generate an energy-temperature trajectory by combining the energy consumption compensation amount output by the gating sequence unit; obtain the maximum waiting time for each vehicle by integrating the trajectory with the minimum safe power threshold and mark high-risk nodes.

[0102] S4. Based on the maximum waiting time and the path travel time and service pressure consumption time provided by the scheduling module, calculate the safety margin and automatically add temperature difference compensation according to thermal risk to obtain the waiting time field; generate a real-time risk sequence by sorting all vehicles in ascending order according to the field.

[0103] S5. Align the waiting time field with the active coupled unit number, extract the unit urgency label, and perform minimum necessary pre-insertion on vehicles within the unit using a local rearrangement strategy that prioritizes urgency and restricts displacement. Then, obtain the global sorting queue and immediate service label through cross-unit fairness check and path connectivity check.

[0104] S6. Based on the real-time number of available workstations, the global sorting queue is windowed to generate the work sequence for the next charging cycle. A vehicle-workstation pairing table is generated according to the vehicle size, charging gun specifications and workstation location mapping. The reachability and power compatibility of the robotic arm are verified in sequence. After the verification is passed, the charging command is synchronously issued to the workstation allocation module, the station information screen and the driver terminal.

[0105] Step S1 includes the following sub-steps:

[0106] S11. Set the time benchmark for the front wheels of vehicles entering the station crossing the station entrance as a unified time reference. Assign a unique queue number to the vehicle. And simultaneously record the remaining battery power. Instantaneous power of air conditioner Instantaneous power of external lights Instantaneous power of information terminals ;right , , Summing yields the continuous power consumption of the attachment. triples Write the starting point of the power consumption record table.

[0107] S12, in time base Internal synchronous acquisition of vehicle body temperature With the ambient temperature inside the station Calculate the difference ;when Assign a cold temperature zone label ,when Assign a temperature zone label ,when Assign hot zone labels ;

[0108] By writing the corresponding labels into the thermal risk table, the impact of environmental temperature difference can be discretized, giving clear weight to the accelerated decay of battery usable capacity caused by high temperature or severe cold in the safety assessment.

[0109] S13. Read the number of available charging stations. Combine it with the vehicle queue number Compare and calculate service pressure values. ;Will With the starting point of the power consumption record , and heat risk labels or or Concatenate to generate a safe waiting feature vector ;

[0110] Step S2 includes the following sub-steps:

[0111] S21. The unique vehicle number obtained in step S1 is... Rather than real-time queuing order Bind and create a vehicle index table. The graph node set is established; any new vehicle entering the queue triggers a node insertion transaction, and any vehicle leaving the queue triggers a node deletion transaction. Both types of transactions are appended with a timestamp. Write back to the index table.

[0112] S22, Index table According to queue order Perform a scan from left to right, first taking the positional difference between two adjacent vehicles. Difference in service pressure Perform parallel judgments;

[0113] If both are not greater than the site activity threshold Then, the two vehicles are assigned to the currently active coupling unit. Otherwise, close immediately. And start the next unit. This continues until the entire train of cars has been divided.

[0114] By using linear grouping, vehicles that are truly under the same crawling rhythm and similar workstation competition pressure are brought together into a unified physical space.

[0115] Subsequently, the real-time vehicle distances within each unit were read. With obstruction of signs ,when Below the slowdown threshold and At the time in the vehicle and Write the connecting edges between them Otherwise, remain disconnected;

[0116] The operation explicitly represents the nearest neighbor relationships most sensitive to the impact on energy consumption and heat dissipation as vehicle-vehicle subgraphs. .

[0117] Then check the first car in the queue With the virtual node at the entrance Write the edges For the last train Virtual node of the nearest available workstation Write the edges This constitutes a vehicle-field subgraph. .

[0118] final, and Vertical stacking forms a two-layer topology It is updated incrementally as the vehicle moves in and out, and any changes to a node or connection are written back with a timestamp.

[0119] S23, Topology Any vehicle connected to the edge Execute energy consumption weight generation: Adjust the energy consumption levels of vehicles on both sides during low-speed driving. Power consumption levels with accessories The energy consumption levels are obtained by superimposing them. The difference between this level and the queuing order Obtain the power transfer weight by weighted combination ;

[0120] The weights are updated in real time according to the creep status and the load level of the surrounding vehicles, directly depicting the intensity of the energy impact caused by the movement and continuous power consumption of adjacent vehicles.

[0121] S24. For the same connected edge Read vehicle thermal risk labels Based on the combination rules, it is mapped to the heat dissipation coupling level. According to vehicle spacing With the wind direction coefficient within the station Graded comparison, Converted into heat dissipation weight And write the edge attributes; when the queue is compact and the environmental temperature difference increases. Automatic lifting, weight reduction when spacing is increased or ventilation is enhanced; at the same time, additional fixed heat dissipation edge coverage is added to vehicles adjacent to buildings or sunshade facilities to cover potential heat sinks or heat sources.

[0122] Based on dynamic heat dissipation weights, the system captures the tendency of sudden capacity degradation in batteries under extreme weather conditions and reserves additional safety margins in the ranking process.

[0123] S25. Vehicle Index Table As the unique primary key, the set of weights for transferring electricity is used. With heat dissipation weight set Node alignment is performed, and a cross pointer structure is used to map the two sets of connected edges;

[0124] When vehicle additions or deletions, or changes in vehicle distance or load are detected, edge additions, removals, or weight overwrites are immediately performed and written to memory in an incremental manner to avoid decision lag caused by full map recalculation.

[0125] S26. Record the starting point of the power generation generated in step S1. Hot risk labels Service pressure value Mapping nodes to their indices to form a four-dimensional feature vector. ;

[0126] Step S3 includes the following sub-steps:

[0127] S31. The power transfer diagram generated in step S2 is used. Heat dissipation coupling diagram With the four-dimensional feature vector of the node Simultaneously loaded into the timing buffer Treating the buffer zone as a queuing scenario at any given time The complete cross-section;

[0128] Then take the time of the previous cycle. Output of remaining battery power for each vehicle With battery temperature The initial values ​​are written into the same buffer to form a spliced ​​page of historical state and current state.

[0129] S32, Vehicle node Update the remaining battery level:

[0130] The first step is to process all data in the power transfer diagram that are related to... Connected edges Perform a traversal, sequentially reading the edge attribute energy decay labels. The label was calculated during edge construction, and its physical meaning is: during this round of slow creeping, due to... Adjacent vehicles The increased power consumption per unit time due to factors such as reduced vehicle spacing, frequent starts and stops, and changes in inter-wheel airflow resistance, is expressed in units of [missing information]. The larger the label, the tighter the queue and the stronger the energy interference.

[0131] The second step is to read the node's own power consumption. The value is derived from the sum of the real-time power consumption of the vehicle's air conditioning, lights, and information terminal, and the unit is also [unit missing]. .

[0132] The third step is to combine all adjacent edges Perform summation and combine with Add them together to get the total amount of electricity that the vehicle must deduct in this cycle.

[0133]

[0134] in, Indicates the time interval The electricity amount deducted internally. Then use directly... Update the node's remaining battery power field and write it back to the buffer.

[0135] By unifying the additional power consumption caused by driving with the continuous power consumption of vehicle accessories into a one-time deduction, the actual power consumption experience of drivers when waiting in slow queues is reflected. The operations team only needs to check... This allows you to determine how quickly the vehicle's battery is depleting and how much battery remains available.

[0136] S33, Vehicle node Update the battery temperature:

[0137] The first step is to perform a thermal coupling analysis on all components in the thermal coupling diagram. Connected edges Perform a traversal, reading the heat dissipation weight of each edge attribute. .

[0138] The weights are generated by a combination of vehicle distance, wind direction, and thermal risk rules between the two vehicles, reflecting the influence of neighboring vehicles on each other. The intensity of the effect of heat dissipation or heat accumulation, in units of The higher the weight, the more obstructed the heat dissipation or the more obvious the heat accumulation.

[0139] The second step is to read the node's own hot risk label. The label values ​​are categorized based on the temperature difference between the vehicle and the environment. A positive value indicates a high-temperature zone, and a negative value indicates a low-temperature zone. The unit is also 1. It is used to describe the natural upward or downward trend of battery temperature caused by the environment.

[0140] The third step is to combine all adjacent edges Perform summation and combine with Add them together to get the change in battery temperature during this cycle.

[0141]

[0142] in, Indicates the time interval The expected rise or fall in internal battery temperature. Update the node temperature field.

[0143] Based on the above treatment, the heat accumulation caused by close vehicle spacing and unfavorable wind direction is combined with the natural temperature rise and fall caused by high temperature or severe cold environment, and directly reflected in the The dispatch system can then detect vehicles with excessively high or low temperatures in real time, which may lead to drastic changes in the available battery capacity.

[0144] S34. The power deduction sequence generated in step S32 is processed. Compared with the temperature change sequence generated in step S33 The data is concatenated according to the vehicle number and written into the gate control input area. and keep up with the times Consistent timestamps;

[0145] S35, Select the nearest gated input area Wheel , Vehicle queuing order Combined into a multidimensional time slice sequence, and sent to the entry control sequence unit. The unit compares the current sequence with the usual pattern through internal switches. Once it detects abnormal fluctuations caused by a sudden increase in air conditioning speed, short-term acceleration, or changes in airflow within the station, it calculates an additional compensation amount. And output it to the energy consumption compensation stack.

[0146] S36, Output the compensation stack With explicit deduction Add them together to get the complete energy consumption in the short term. ;according to and Generate a power change curve, along with the latest temperature. Write to a circular cache of node history and use rolling over to retain the most recent ones. The system generates a record; by integrating rule-based calculations and temporary corrections, a traceable energy-temperature dual trajectory is formed, providing continuous input for safety window calculations and avoiding decision distortion due to trajectory breakpoints.

[0147] S37. Integrate the power change curve over the time axis and sum it with the current remaining power at the node. Comparison based on the minimum safety threshold of the battery Determine the maximum waiting time The data is then written to the node safety window field; the field presents the vehicle's remaining buffer under the non-shutdown constraint in minutes, avoiding the need for dispatchers to manually calculate between multi-dimensional data.

[0148] S38, For all nodes A threshold scan of the field will reveal fields that fall below the warning threshold. Vehicles marked as high-risk nodes Then Output.

[0149] Step S4 includes the following sub-steps:

[0150] S41. Store the power change curve in the node's historical records in minute sequence. Current remaining power of the node Perform initial overlay; if the overlay value exceeds the station's minimum safety threshold... Then the time pointer Proceed to the next minute, then take the next segment of the curve and continue accumulating;

[0151] When it first appears The vehicle is locked immediately. The ultimate waiting point under uninterrupted power supply conditions The segment-by-segment comparison directly aligns the vehicle's actual power consumption rhythm with a unified safety baseline.

[0152] S42. Combine the maximum possible time to wait with the estimated actual turn time into a simple, on-site readable statement of "how many more minutes can I wait?" and complete it in four sequential actions:

[0153] First, read the limit waiting endpoint obtained in step S41. (Meaning: vehicle) If power consumption continues, (Always touch the minimum safe battery level), and simultaneously read the time when the vehicle enters the queue. The difference between the two This refers to the total number of minutes the vehicle can remain in the station, excluding queuing time.

[0154] Second, perform a one-way traversal of the topology generated by S22 in the direction of exit: along the vehicle... The path from the vehicle ahead to the available workstation is tracked node by node, collecting real-time distance between vehicles, the active unit number, and the workstation release rate returned by the scheduling module. The scheduler adds the approximate time required to move one meter to the inertia of the active unit, linearly accumulating the total travel time required for the vehicle to travel from its current position to the nearest available workstation. .

[0155] Third, read the service pressure value within the node. (Definition: The average additional waiting time for each additional vehicle in a queue), compare it with... Adding them together gives a more accurate prediction of queuing time based on the actual situation. Then, the queuing time is subtracted from the maximum allowable dwell time to obtain the original safety margin; if this margin is lower than the station's preset thermal threshold... This indicates that vehicles are more likely to lose range in high or low temperatures, according to the thermal risk label. Automatic addition of temperature difference compensation The compensation range increases with the severity of the environment.

[0156] Fourth, incorporate safety margin and temperature difference compensation into the formula.

[0157]

[0158] in, This is the number of minutes you can wait, rounded down. Waiting for the end at the limit; This refers to the vehicle's arrival time. For estimated travel time; Service pressure takes time; For temperature difference compensation; To ensure the result is an integer number of minutes, round down.

[0159] Finally Write nodes can wait for the duration field and overwrite the old value.

[0160] Through this process, information on vehicle battery limit, actual queuing pace, and ambient temperature difference is compressed into a single intuitive indicator. Based on this, dispatchers can quickly identify high-risk vehicles that urgently need to cut in line or enter the station immediately, and on-site personnel can also understand at a glance how much longer they can wait.

[0161] S43. Number the vehicle With waiting time The paired data is written to the queuing node table, overwriting the old value. Immediately after this, a full ascending scan of the table is triggered, performing a scan on all vehicles. Sorting data from smallest to largest temporarily generates a real-time risk sequence. The sequence naturally pushes the vehicles most prone to power outages to the front, ensuring that any subsequent scheduling decisions are based on the latest risk profile and avoiding time lag between the field and the algorithm.

[0162] S44, Real-time risk sequence The data is pushed to the dispatch center, which reads the number of vehicles at the front end of the sequence and compares it with the number of currently available workstations. The system matches and marks successfully matched vehicles as those to be served immediately. Then, work station instructions are simultaneously issued on the information screens in the station, the dispatch terminal, and the driver's mobile terminal. On-site guides and drivers drive according to the instructions. Through this closed loop, the abstract waiting time is directly transformed into an executable action, ensuring that no vehicle will cause traffic congestion or safety hazards due to power depletion or temperature imbalance during the queuing period, while maintaining an overall sense of fairness in the first-come, first-served system.

[0163] Step S5 includes the following sub-steps:

[0164] S51. Write step S4 back to the node's waitable duration field. Active coupling unit numbering as specified by S22 Perform alignment and write to the node; group by unit number, and execute once for each vehicle within each unit. The algorithm performs an ascending scan, records the minimum value within each cell in real time, and stores this minimum value as a cell urgency label. ;based on In subsequent sorting, it can quickly identify which queue is at the earliest risk of power outage without having to compare the entire table for each vehicle, significantly shortening the response time during peak periods.

[0165] S52, will have the same unit number Vehicles are placed in the same work set, and a local rearrangement is carried out using a priority-based, displacement-constrained strategy: vehicles within the set are arranged according to... Reorder in ascending order, then scan from front to back along the original inbound sequence, only for... Vehicles below the station warning threshold are subject to minimum necessary pre-insertion, and the insertion distance does not cross the cell boundary. Based on the natural constraint of the physical continuity of the coupled cells, high-risk vehicles are given priority protection in this cell, while avoiding rearrangement that could affect adjacent cells and cause chain displacement of the entire train.

[0166] S53. Reconnect the local lists after the partial rearrangement of each unit into a global candidate queue according to the original queuing order, and calculate the displacement difference of each vehicle from its original position to its current position. If the cumulative value of the displacement difference of the entire queue exceeds the station's fairness threshold, then roll back the vehicle with the smallest displacement to its original order according to the principle of priority rollback across units, until the cumulative value falls back to within the threshold. Through the cross-unit fairness check, the system simultaneously satisfies the immediate protection of high-risk vehicles and the overall consideration of the first-come-first-served principle.

[0167] S54. Write the globally sorted queue, which has been verified through fair testing, into the scheduling center, and prioritize queues located at the front of the queue and... Vehicles still below the safety buffer threshold are given an immediate service tag; the virtual connection in S22 is invoked to perform connectivity verification on the driving path from the current first vehicle to the idle workstation. If the scheduling is found to cause path blockage, a micro-switch is triggered until full connectivity is achieved; finally, the queue order, service tag and connectivity results are synchronously published to the workstation allocation module, the station information screen and the driver's mobile terminal.

[0168] Step S6 includes the following sub-steps:

[0169] S61. The global sorted queue output in step S5. With real-time idle workstations To align, a window truncation method is used to generate the job sequence for the next charging cycle: first, the window width is calculated. ,in This represents the maximum number of parallel workstations that can be operated in a single cycle at the station; then from... Head extraction sequence before Vehicles, forming an execution sequence And write it to the task area;

[0170] S62, Execution sequence The vehicle-workstation pairing table is generated by mapping the vehicle size, charging gun specifications, and physical location of the workstation. The pairing table is then checked in two rounds: the first round checks the reachability of the robotic arm, and the second round checks the power compatibility. If either check fails, the next vehicle is taken from the sorting queue and remapped until the pairing table passes all checks.

[0171] In this implementation, the electric vehicle charger safety waiting management system based on the improved time-series network includes:

[0172] The vehicle entry data acquisition terminal integrates a geomagnetic coil, a millimeter-wave ranging node, and a vehicle-to-pile communication interface. It is used to obtain a unified time reference the instant the front wheels of the vehicle pass the station entrance, and to collect raw data such as the vehicle's remaining battery power, accessory power consumption, vehicle body temperature, and ambient temperature in real time.

[0173] The safety feature generation module sums the vehicle's remaining battery power with the power consumption of the accessories to generate a battery power starting point record, generates a thermal risk label based on the temperature difference between the vehicle body and the environment, and calculates the service pressure value in combination with the number of available charging stations. Finally, the modules are spliced ​​together to form a safety waiting feature vector.

[0174] The two-layer topology construction module divides active coupling units based on vehicle queuing order and service pressure difference, writes vehicle-vehicle connections and vehicle-field connections respectively, adds energy transfer weights and heat dissipation coupling weights to the connections and carries timestamp incremental updates to construct a vehicle-field two-layer topology graph that is dynamically refreshed as vehicles move in and out.

[0175] The timing simulation and compensation module loads the energy transfer diagram, heat dissipation coupling diagram and node characteristics into the timing buffer, traverses adjacent edges to deduct node power, corrects node temperature, and calls the gated sequence unit to output energy consumption compensation for short-term anomalies, generating a continuous energy-temperature trajectory.

[0176] The safety margin assessment module integrates the energy-temperature trajectory to obtain the maximum wait time, calculates the safety margin by combining the path movement time and service pressure time, adds temperature difference compensation according to the thermal risk label, outputs the wait time field and generates a real-time risk sequence.

[0177] The queue rearrangement and workstation allocation module performs minimum necessary pre-insertion based on the urgency label within the active coupling unit, and obtains a globally sorted queue after performing fairness checks and path connectivity checks across units. It then connects to the real-time idle workstation window to capture the work sequence, completes vehicle-workstation pairing, and sequentially passes the robotic arm accessibility and power compatibility checks. After passing the checks, the charging command is synchronously sent to the workstation allocation module, the station information screen, and the driver terminal.

[0178] Example 1:

[0179] A certain operations and maintenance center plans to upgrade its largest daily charging station with intelligent features in 2024. This station has 42 DC fast charging bays and 8 parallel queuing lanes, with a peak hourly traffic flow of up to 160 vehicles. Over the past two years, the station has experienced several typical safety issues, including vehicles being forced to push their batteries due to prolonged queuing causing their State of Charge (SOC) to drop below 3%, battery temperatures soaring on extreme heat days triggering BMS protection, and lane congestion caused by drivers vying for charging bays. Meanwhile, the Human Resources Service Department has systematically reviewed charger fault types and causes in its charging station training simulation method research project, obtaining a list of fault factors including 27 indicators such as ambient temperature, operating method, single charging duration, and cumulative service life. This list is planned to be embedded into the operations and maintenance training simulation platform. Based on these pain points and fundamental research findings, the company ultimately decided to introduce the electric vehicle charger safety waiting management method and system based on improved time-series networks described in this invention, simultaneously implementing it in two major scenarios: intelligent scheduling and simulation teaching.

[0180] From March to April 2024, the technical team installed geomagnetic coils and millimeter-wave ranging nodes at all entrances, upgraded the OCPP1.6J protocol stack on the communication layer of each charger, and deployed six sets of temperature and humidity IoT probes in the waiting area. Simultaneously, real-time fields such as the number of times the charging gun was plugged in and out, the overall running time of the charger, and the internal IGBT temperature rise were added according to the fault factor list. These 68 data sources were aggregated by the edge server and pushed to the timing buffer every 5 seconds via MQTT.

[0181] Based on the timestamp of a vehicle's front wheels crossing the access control line, a safe waiting feature vector is generated in real time, and a vehicle-parking two-layer topology is constructed. Unlike conventional implementations, this embodiment adds a charger health degradation coefficient to the edge attributes of the thermal coupling graph. This coefficient is derived from the key dimensions (ambient temperature, service life, IGBT temperature rise) in the fault factor list through an exponential decay function. It is used to dynamically amplify the negative impact of aging or high-temperature workstations on queuing safety margins, thereby automatically reducing the probability of high-fault-probability workstations being assigned to high-risk vehicles at the decision-making level.

[0182] Historical queuing logs from May 2023 to February 2024 (a total of 1,087,236 vehicle records) were selected as pre-training samples. Gated Sequence Units (GRUs) were used to learn the energy-temperature trajectory, and three types of events—"peak SOC < 5%", "battery temperature > 55℃", and "queue interruption leading to idle workstations"—were used as supervision labels. The model achieved a recall of 93.8% and an F1 score of 0.914 on the validation set, representing a 14.6% improvement in F1 compared to methods that only estimate based on SOC thresholds and average power consumption.

[0183] The system officially switched to online mode on May 1, 2024, and ran continuously for 90 days. As of July 30, 2024, a total of 412,587 vehicles had been processed, with 94% of the queuing status adjusted in real time through second-level reordering logic. Compared with the previous system (which used a first-come, first-served approach with simple SOC warnings), the key performance indicators are as follows:

[0184]

[0185] On June 23, 2024, with an outside temperature of 36℃ and 53 vehicles queuing at the station, the system detected a cold chain delivery vehicle with a State of Charge (SOC) of only 7% and a vehicle-environment temperature difference of +22℃, placing it in the high-risk sequence. Because the charger at the nearest workstation was deemed to have a low health coefficient due to its operating life exceeding 25,000 hours and IGBT temperature rise, the scheduling module guided the vehicle to workstation number 7, which had a better health coefficient. Simultaneously, it performed minimum necessary pre-insertion on the active coupling unit where the vehicle was located, resulting in an overall displacement difference of only 1.8m and a total queue fairness threshold occupancy of 12%. The vehicle successfully connected to the charger 4 minutes later, avoiding potential power outages and shutdowns. Subsequent inspection revealed that the maintenance log for workstation number 7 consistently showed a temperature rise below 50℃, verifying the effectiveness of the fault weight injection mechanism.

[0186] In a charging facility operation and maintenance training session organized by the Human Resources Service Office in July 2024, 60 days of system operation logs were imported into a simulation platform. A three-pronged teaching script—real queuing video recording, dynamic topology playback, and fault weight visualization—helped trainees visually observe the impact of station aging and environmental temperature changes on queuing safety margins. Post-training questionnaires showed that trainees' understanding of the linkage between the fault factor list and scheduling strategies increased from 41% before the training to 88%, achieving the expected teaching objectives.

[0187] In summary, this embodiment verifies that the present invention can output reliable safe waiting time in real time, effectively protect high-risk vehicles, and simultaneously improve the overall efficiency and fairness of the station. At the same time, the injection of fault factor weights enables the scheduling strategy to be deeply coupled with the health status of the charger, providing rich and reliable practical data for subsequent operation and maintenance simulation training.

[0188] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any modifications or equivalent changes made based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A method and system for managing the safe waiting time of electric vehicle chargers based on improved time-series networks, characterized in that, Includes the following steps: S1. Under the unified time reference of the vehicle's front wheels crossing the station entrance, a unique queue number is generated for the entering vehicle. The initial state of charge of the vehicle, the power of the air conditioner, the power of the external lights and the power of the information terminal are collected and summed to obtain the power consumption of the accessories. At the same time, the vehicle body temperature and the ambient temperature inside the station are acquired in real time and the temperature difference is calculated. The temperature difference is discretized into thermal risk tags according to a preset threshold. Then, the number of available charging stations is read to calculate the service pressure value. Finally, the combination is used to obtain the safe waiting feature vector. S2. Based on the queue number, create a vehicle index table and establish a set of topological nodes. Divide active coupling units according to the queue position difference and service pressure difference. Write vehicle-vehicle connections within the unit according to the vehicle distance and occlusion status, and calculate the energy transfer weight and heat dissipation weight respectively. At the same time, establish vehicle-field connections and map the safe waiting feature vector to the node to form a two-layer topology graph that is updated incrementally as the vehicle moves in and out. S3. Load the energy transfer graph, heat dissipation coupling graph and node features into the time buffer, traverse adjacent edges according to a preset period, deduct the remaining power of the node using energy consumption weight and accessory power consumption, correct the node battery temperature using heat dissipation weight and thermal risk label, and generate an energy-temperature trajectory by combining the energy consumption compensation amount output by the gated sequence unit. The maximum waiting time for each vehicle is obtained by integrating the trajectory with the minimum safe battery threshold, and high-risk nodes are marked. S4. Based on the maximum waiting time and the path travel time and service pressure consumption time provided by the scheduling module, calculate the safety margin and automatically add temperature difference compensation according to thermal risk to obtain the waiting time field; generate a real-time risk sequence by sorting all vehicles in ascending order according to the field. S5. Align the waiting time field with the active coupled unit number, extract the unit urgency label, and perform minimum necessary pre-insertion on vehicles within the unit using a local rearrangement strategy that prioritizes urgency and restricts displacement. Then, obtain the global sorting queue and immediate service label through cross-unit fairness check and path connectivity check. S6. Based on the real-time number of available workstations, the global sorting queue is windowed to generate the work sequence for the next charging cycle. A vehicle-workstation pairing table is generated according to the vehicle size, charging gun specifications and workstation location mapping. The reachability and power compatibility of the robotic arm are verified in sequence. After the verification is passed, the charging command is synchronously issued to the workstation allocation module, the station information screen and the driver terminal.

2. The electric vehicle charger safety waiting management method and system based on improved time-series networks according to claim 1, characterized in that, Step S1 includes the following sub-steps: S11. Set the time benchmark for the front wheels of vehicles entering the station crossing the station entrance as a unified time reference. Assign a unique queue number to the vehicle. And simultaneously record the remaining battery power. Instantaneous power of air conditioner Instantaneous power of external lights Instantaneous power of information terminals ;right , , Summing yields the continuous power consumption of the attachment. triples Write the starting point of the power consumption record table; S12, in time base Internal synchronous acquisition of vehicle body temperature With the ambient temperature inside the station Calculate the difference ;when Assign a cold temperature zone label ,when Assign a temperature zone label ,when Assign hot zone labels ; S13. Read the number of available charging stations. Combine it with the vehicle queue number Compare and calculate service pressure values. ;Will With the starting point of the power consumption record , and heat risk labels or or Concatenate to generate a safe waiting feature vector .

3. The electric vehicle charger safety waiting management method and system based on improved time-series networks according to claim 1, characterized in that, Step S2 includes the following sub-steps: S21. The unique vehicle number obtained in step S1 is... Rather than real-time queuing order Bind and create a vehicle index table. The graph node set is established; any new vehicle entering the queue triggers a node insertion transaction, and any vehicle leaving the queue triggers a node deletion transaction. Both types of transactions are appended with a timestamp. Write back to the index table; S22, Index table According to queue order Perform a scan from left to right, first taking the positional difference between two adjacent vehicles. Difference in service pressure Perform parallel judgments; If both are not greater than the site activity threshold Then, the two vehicles are assigned to the currently active coupling unit. Otherwise, close immediately. And start the next unit. This continues until the entire train of cars has been divided. By using linear grouping, vehicles that are truly under the same crawling rhythm and similar workstation competition pressure are brought together into a unified physical space; Subsequently, the real-time vehicle distances within each unit were read. With obstruction of signs ,when Below the slowdown threshold and At the time in the vehicle and Write the connecting edges between them Otherwise, remain disconnected; The operation explicitly represents the nearest neighbor relationships most sensitive to the impact on energy consumption and heat dissipation as vehicle-vehicle subgraphs. ; Then check the first car in the queue With the virtual node at the entrance Write the connection edge For the last train Virtual node of the nearest available workstation Write the connection edge This constitutes a vehicle-field subgraph. ; final, and Vertical stacking forms a two-layer topology It is updated incrementally as the vehicle moves in and out, and any changes to a node or connection are written back with a timestamp. S23, Topology Any vehicle connected to the edge Execute energy consumption weight generation: Adjust the energy consumption levels of vehicles on both sides during low-speed driving. Power consumption levels with accessories The energy consumption levels are obtained by superimposing them. The difference between this level and the queuing order Obtain the power transfer weight by weighted combination ; S24. For the same connected edge Read vehicle thermal risk labels Based on the combination rules, it is mapped to the heat dissipation coupling level. According to vehicle spacing With the wind direction coefficient within the station Graded comparison, Converted into heat dissipation weight And write the edge attributes; when the queue is compact and the environmental temperature difference increases. Automatic lifting, weight reduction when spacing is increased or ventilation is enhanced; at the same time, additional fixed heat dissipation edge coverage is added to vehicles adjacent to buildings or sunshade facilities to cover potential heat sinks or heat sources. S25. Vehicle Index Table As the unique primary key, the set of weights for transferring electricity is used. With heat dissipation weight set Node alignment is performed, and a cross pointer structure is used to map the two sets of connected edges; S26. Record the starting point of the power generation generated in step S1. Hot risk labels Service pressure value Mapping nodes to their indices to form a four-dimensional feature vector. .

4. The electric vehicle charger safety waiting management method and system based on improved time-series networks according to claim 1, characterized in that, Step S3 includes the following sub-steps: S31. The power transfer diagram generated in step S2 is used. Heat dissipation coupling diagram With the four-dimensional feature vector of the node Simultaneously loaded into the timing buffer Treating the buffer zone as a queuing scenario at any given time The complete cross-section; Then take the time of the previous cycle. Output of remaining battery power for each vehicle With battery temperature The initial values ​​are written into the same buffer to form a spliced ​​page of historical state and current state. S32, Vehicle node Update the remaining battery level: The first step is to process all data in the power transfer diagram that are related to... Connected edges Perform a traversal, sequentially reading the edge attribute energy decay labels. The label was calculated during edge construction, and its physical meaning is: during this round of slow creeping, due to... Adjacent vehicles The increased power consumption per unit time due to factors such as reduced vehicle spacing, frequent starts and stops, and changes in inter-wheel airflow resistance, is expressed in units of [missing information]. ; The larger the label, the tighter the queue and the stronger the energy interference. The second step is to read the node's own power consumption. The value is derived from the sum of the real-time power consumption of the vehicle's air conditioning, lights, and information terminal, and the unit is also [unit missing]. ; The third step is to combine all adjacent edges Perform summation and combine with Add them together to get the total battery charge that the vehicle must deduct in this cycle; ； in, Indicates the time interval The electricity deducted internally was then used directly. Update the node's remaining battery power field and write it back to the buffer; By unifying the additional power consumption caused by driving with the continuous power consumption of vehicle accessories into a one-time deduction, the actual power consumption experience of drivers when waiting in slow queues is reflected. The operations team only needs to check... This allows you to determine how quickly the vehicle's battery is depleting and how much battery remains available. S33, Vehicle node Update the battery temperature: The first step is to perform a thermal coupling analysis on all components in the thermal coupling diagram. Connected edges Perform a traversal, reading the heat dissipation weight of each edge attribute. ; The weights are generated by a combination of vehicle distance, wind direction, and thermal risk rules between the two vehicles, reflecting the influence of neighboring vehicles on each other. The intensity of the effect of heat dissipation or heat accumulation, in units of A higher weight indicates more obstructed heat dissipation or more significant heat accumulation. The second step is to read the node's own hot risk label. The label values ​​are categorized based on the temperature difference between the vehicle and the environment. A positive value indicates a high-temperature zone, and a negative value indicates a low-temperature zone. The unit is also 1. This is used to describe the natural upward or downward trend of battery temperature caused by the environment; The third step is to combine all adjacent edges Perform summation and combine with Add them together to get the change in battery temperature during this cycle; ； in, Indicates the time interval The expected rise or fall in internal battery temperature, using Update the node temperature field; Based on the above treatment, the heat accumulation caused by close vehicle spacing and unfavorable wind direction is combined with the natural temperature rise and fall caused by high temperature or severe cold environment, and directly reflected in the The dispatch system can detect vehicles with excessively high or low temperatures in real time, which may cause a sharp change in the available battery capacity. S34. The power deduction sequence generated in step S32 is processed. Compared with the temperature change sequence generated in step S33 The data is concatenated according to the vehicle number and written into the gate control input area. and keep up with the times Consistent timestamps; S35, Select the nearest gated input area Wheel , Vehicle queuing order Combined into a multidimensional time slice sequence, and sent to the entry control sequence unit. The unit compares the current sequence with the usual pattern through internal switches. Once it detects abnormal fluctuations caused by a sudden increase in air conditioning speed, short-term acceleration, or changes in airflow within the station, it calculates an additional compensation amount. And output to the energy compensation stack; S36, Output the compensation stack With explicit deduction Add them together to get the complete energy consumption in the short term. ;according to and Generate a power change curve, along with the latest temperature. Write to a circular cache of node history and use rolling over to retain the most recent ones. The system generates a record; by integrating rule-based calculations and temporary corrections, a traceable energy-temperature dual trajectory is formed, providing continuous input for safety window calculations and avoiding decision distortion due to trajectory breakpoints. S37. Integrate the power change curve over the time axis and sum it with the current remaining power at the node. Comparison based on the minimum safety threshold of the battery Determine the maximum waiting time The data is written to the node safety window field; the field presents the vehicle's remaining buffer under the non-shutdown constraint in minutes, avoiding the need for dispatchers to manually calculate between multi-dimensional data. S38, For all nodes A threshold scan of the field will reveal fields that fall below the warning threshold. Vehicles marked as high-risk nodes Then Output.

5. The electric vehicle charger safety waiting management method and system based on improved time-series networks according to claim 1, characterized in that, Step S4 includes the following sub-steps: S41. Store the power change curve in the node's historical records in minute sequence. Current remaining power of the node Perform initial overlay; if the overlay value exceeds the station's minimum safety threshold... Then the time pointer Proceed to the next minute, then take the next segment of the curve and continue accumulating; When it first appears The vehicle is locked immediately. The ultimate waiting point under uninterrupted power supply conditions The segment-by-segment comparison directly aligns the vehicle's actual power consumption rhythm with a unified safety baseline. S42. Combine the maximum possible time to wait with the estimated actual turn time into a simple, on-site readable statement of "how many more minutes can I wait?" and complete it in four sequential actions: First, read the limit waiting endpoint obtained in step S41. Meaning: vehicle If power consumption continues, Constantly monitor the minimum safe battery level and simultaneously read the time when vehicles enter the queue. The difference between the two This refers to the total number of minutes the train can remain in the station, excluding queuing time. Second, perform a one-way traversal of the topology generated by S22 in the direction of exit: along the vehicle... The path from the vehicle ahead to the available workstation is tracked node by node, collecting real-time distance between vehicles, the active unit number, and the workstation release rate returned by the scheduling module. The scheduler adds the approximate time required to move one meter to the inertia of the active unit, linearly accumulating the total travel time required for the vehicle to travel from its current position to the nearest available workstation. ; Third, read the service pressure value within the node. Definition: In a queue, the average additional waiting time for each additional vehicle is calculated by multiplying it by the average waiting time for each additional vehicle. Adding them together gives a more accurate prediction of queuing time based on the actual situation. Then, the queuing time is subtracted from the maximum allowable dwell time to obtain the original safety margin; if this margin is lower than the station's preset thermal threshold... This indicates that vehicles are more likely to lose range in high or low temperatures, according to the thermal risk label. Automatic addition of temperature difference compensation The compensation range increases with the severity of the environment. Fourth, incorporate safety margin and temperature difference compensation into the formula; ； in, This is the number of minutes you can wait, rounded down. Waiting for the end at the limit; This refers to the vehicle's arrival time. For estimated travel time; Service pressure takes time; For temperature difference compensation; To ensure the result is an integer number of minutes, round down. Finally Write to the node and wait for the duration field, then overwrite the old value; S43. Number the vehicle With waiting time The paired data is written to the queuing node table, overwriting the old value. Immediately after this, a full ascending scan of the table is triggered, performing a scan on all vehicles. Sorting data from smallest to largest temporarily generates a real-time risk sequence. The sequence naturally pushes the vehicles most prone to power outages to the front, ensuring that any subsequent scheduling decisions are based on the latest risk profile and avoiding time lag between the field and the algorithm; S44, Real-time risk sequence The data is pushed to the dispatch center, which reads the number of vehicles at the front end of the sequence and compares it with the number of currently available workstations. The system matches and marks successfully matched vehicles as those to be served immediately. Then, work station instructions are simultaneously issued on the information screens in the station, the dispatch terminal, and the driver's mobile terminal. On-site guides and drivers drive according to the instructions. Through this closed loop, the abstract waiting time is directly transformed into an executable action, ensuring that no vehicle will cause traffic congestion or safety hazards due to power depletion or temperature imbalance during the queuing period, while maintaining an overall sense of fairness in the first-come, first-served system.

6. The electric vehicle charger safety waiting management method and system based on improved time-series networks according to claim 1, characterized in that, Step S5 includes the following sub-steps: S51. Write step S4 back to the node's waitable duration field. Active coupling unit numbering as specified by S22 Perform alignment and write to the node; group by unit number, and execute once for each vehicle within each unit. The algorithm performs an ascending scan, records the minimum value within each cell in real time, and stores this minimum value as a cell urgency label. ;based on In subsequent sorting, it can easily identify which queue is at the earliest risk of power outage, without the need for a full table comparison of each vehicle, significantly shortening the response time during peak periods; S52, will have the same unit number Vehicles are placed in the same work set, and a local rearrangement is carried out using a priority-based, displacement-constrained strategy: vehicles within the set are arranged according to... Reorder in ascending order, then scan from front to back along the original inbound sequence, only for... Vehicles below the station warning threshold are subject to minimum necessary pre-insertion, and the insertion distance does not cross the cell boundary; based on the natural constraint of the physical continuity of the coupled cells, high-risk vehicles are given priority protection in this cell, while avoiding rearrangement from affecting adjacent cells and causing chain displacement of the entire train. S53. Reconnect the local lists after the partial rearrangement of each unit into a global candidate queue according to the original queuing order, and calculate the displacement difference of each vehicle from its original position to its current position. If the cumulative displacement difference of the entire queue exceeds the station's fairness threshold, then roll back the vehicle with the smallest displacement to its original order according to the principle of priority rollback across units, until the cumulative value falls back to within the threshold. Through the cross-unit fairness check, the system simultaneously satisfies the immediate protection of high-risk vehicles and the overall consideration of the first-come-first-served principle. S54. Write the globally sorted queue, which has been verified through fair testing, into the scheduling center, and prioritize queues located at the front of the queue and... Vehicles still below the safety buffer threshold will be given an immediate service tag; The virtual connection in S22 is invoked to perform connectivity verification on the driving path from the current first vehicle to the idle workstation. If the scheduling is found to cause path blockage, a micro-switch is triggered until the connection is fully established. Finally, the queue order, service tag and connectivity results are synchronously published to the workstation allocation module, the station information screen and the driver's mobile terminal.

7. The electric vehicle charger safety waiting management method and system based on improved time-series networks according to claim 1, characterized in that, Step S6 includes the following sub-steps: S61. The global sorted queue output in step S5. With real-time idle workstations To align, a window truncation method is used to generate the job sequence for the next charging cycle: first, the window width is calculated. ,in This represents the maximum number of parallel workstations that can be operated in a single cycle at the station; then from... Head extraction sequence before Vehicles, forming an execution sequence And write it to the task area; S62, Execution sequence A vehicle-workstation pairing table is generated by mapping vehicle size, charging gun specifications, and workstation physical location. The pairing table is then validated in two rounds: the first round verifies the reachability of the robotic arm, and the second round verifies power compatibility. If any verification fails, the next vehicle is taken from the sorting queue and remapped until the pairing table passes all verifications.

8. An electric vehicle charger safety waiting management system based on an improved time-series network, comprising the electric vehicle charger safety waiting management method based on an improved time-series network as described in any one of claims 1 to 7, characterized in that, include: The vehicle entry data acquisition terminal integrates a geomagnetic coil, a millimeter-wave ranging node, and a vehicle-to-pile communication interface. It is used to obtain a unified time reference the instant the front wheels of the vehicle pass the station entrance, and to collect raw data such as the vehicle's remaining battery power, accessory power consumption, vehicle body temperature, and ambient temperature in real time. The safety feature generation module sums the vehicle's remaining battery power with the power consumption of the accessories to generate a battery power starting point record, generates a thermal risk label based on the temperature difference between the vehicle body and the environment, and calculates the service pressure value in combination with the number of available charging stations. Finally, the modules are spliced ​​together to form a safety waiting feature vector. The two-layer topology construction module divides active coupling units based on vehicle queuing order and service pressure difference, writes vehicle-vehicle connections and vehicle-field connections respectively, adds energy transfer weights and heat dissipation coupling weights to the connections and carries timestamp incremental updates to construct a vehicle-field two-layer topology graph that is dynamically refreshed as vehicles move in and out. The timing simulation and compensation module loads the energy transfer diagram, heat dissipation coupling diagram and node characteristics into the timing buffer, traverses adjacent edges to deduct node power, corrects node temperature, and calls the gated sequence unit to output energy consumption compensation for short-term anomalies, generating a continuous energy-temperature trajectory. The safety margin assessment module integrates the energy-temperature trajectory to obtain the maximum wait time, calculates the safety margin by combining the path movement time and service pressure time, adds temperature difference compensation according to the thermal risk label, outputs the wait time field and generates a real-time risk sequence. The queue rearrangement and workstation allocation module performs minimum necessary pre-insertion based on the urgency label within the active coupling unit, and obtains a globally sorted queue after performing fairness checks and path connectivity checks across units. Then, the system connects to the real-time available workstation window to capture the work sequence, completes vehicle-workstation pairing, and sequentially passes the robotic arm accessibility and power compatibility checks. After the checks are passed, the charging command is synchronously sent to the workstation allocation module, the station information screen, and the driver terminal.

Images