A method and system for cooperative scheduling of intelligent auxiliary walking scooters for airports

Abstract

This invention relates to the field of intelligent traffic scheduling technology for airports, and discloses a collaborative scheduling method and system for intelligent auxiliary mobility vehicles at airports. The method includes: extracting the vehicle-level battery charging characteristic parameter set for each mobility vehicle to generate a heterogeneous charging duration function; generating a pulsed passenger arrival time distribution for the satellite terminal using non-negative matrix factorization and a discrete event queuing model; calculating the vehicle-level satellite terminal service state of charge threshold value; screening a set of dual feasible candidate vehicles; using mixed integer linear programming to jointly solve for the optimal target state of charge value and pulse group allocation; and performing aging safety constraint verification and dynamically updating parameters.

Description

Technical Field

[0001] This invention belongs to the field of airport mobility scooter dispatching technology, specifically relating to a collaborative dispatching method and system for intelligent auxiliary mobility scooters at airports. Background Technology

[0002] Large hub airports typically employ a terminal layout with a main building and satellite terminals. Passengers must use automated guided vehicles (AGVs) to reach their boarding gates in the satellite terminals. For passengers with mobility impairments, airports provide shuttle services using intelligent mobility scooters. Because passenger arrival times in the satellite terminals are constrained by train schedules, the demand for these scooters exhibits significant batch-like, pulsed characteristics rather than a continuous, smooth distribution. Existing scheduling methods usually model passenger demand as a continuous flow, failing to accurately capture these pulsed arrival patterns, resulting in a systematic deviation between vehicle readiness time and actual demand time.

[0003] In terms of charging management, the battery health status of each vehicle in the fleet exhibits a heterogeneous distribution due to differences in their service life and historical operating conditions. Differences in battery aging result in variations in the state-of-charge value at the inflection point of charging power degradation, segmented charging power levels, and actual available energy for each vehicle. Existing charging scheduling methods typically employ a uniform linear charging model, applying the same charging time estimation parameters to all vehicles, ignoring individual differences between vehicles. This approach leads to a systematic underestimation of charging time for vehicles with higher aging levels, potentially causing vehicles to be dispatched before reaching the required energy level for service.

[0004] Furthermore, there is a complex coupling relationship between the time occupancy constraints of charging pile resources and the concurrent demands of multiple pulse groups. Given a limited number of charging piles, maximizing the demand coverage of each pulse group while meeting aging safety constraints is a problem that existing scheduling methods have not yet systematically solved. Existing methods lack a complete framework for jointly optimizing heterogeneous charging characteristics, pulse-based demand forecasting, energy accessibility constraints, and charging pile resource allocation, thus limiting the overall service coverage and the reliability of the scheduling scheme. Summary of the Invention

[0005] The purpose of this invention is to provide a collaborative scheduling method and system for intelligent auxiliary mobility vehicles at airports. By acquiring the battery charging characteristic parameters of each mobility vehicle in the fleet, predicting the arrival demand of passengers in the satellite terminal, calculating the energy accessibility of the vehicles, screening dual feasible candidate vehicles, jointly optimizing the vehicle state of charge and pulse group allocation, performing safety constraint verification and dynamically adjusting the execution plan, this invention solves the problem in the prior art that it is impossible to effectively distinguish between real priority flow intrusion events and sensor failures under the super-permeable priority flow condition, and that the fault nodes lack physical consistency data repair methods.

[0006] This invention discloses a collaborative scheduling method for intelligent auxiliary mobility vehicles at airports, comprising the following steps: Obtain the battery management system recorded data of each vehicle in the fleet, extract the average actual charging power and the state of charge value at the inflection point of charging power decay for each vehicle in different state of charge ranges, and generate a set of vehicle-level battery charging characteristic parameters. Based on the set of vehicle-level battery charging characteristic parameters, the relationship between the state of charge and charging time of each vehicle is piecewise linearly fitted according to the inflection point of charging power decay, generating a heterogeneous charging time function and a mapping table between the state of charge and actual energy for each vehicle. Passenger flow data is obtained from multiple data sources and boarding gate area attributes are labeled. The aggregated traffic at the security checkpoint is decomposed into the contribution components of each flight. A discrete event queuing model is established for the destination flow in the satellite terminal to perform event-by-event simulation, generating the pulsed passenger arrival time distribution of each boarding gate in the satellite terminal and the demand for chauffeured vehicles in each pulse group. Based on the state of charge mapping table and airport road network data, calculate the state of charge threshold value for satellite terminal service for each vehicle. The arrival time of each pulse group is deduced backward to the vehicle readiness deadline, and dual feasible candidate vehicles whose charging completion time is earlier than the readiness deadline and meet the energy feasibility conditions are screened. With the goal of maximizing demand coverage, a mixed-integer linear programming approach is used to jointly solve the target state of charge value and pulse group allocation for each vehicle, thereby generating a collaborative scheduling scheme. The aging safety constraint verification is performed on the collaborative scheduling scheme and then the scheme is issued for execution.

[0007] In a preferred embodiment, the vehicle-grade battery charging characteristic parameter set includes battery health status value, cell internal resistance measurement value, cell temperature, charging power decay inflection point state of charge value, and the average actual charging power within each discrete state of charge interval, wherein the charging power decay inflection point state of charge value is defined as the state of charge value corresponding to the transition of charging power from the constant current stage to the constant voltage stage.

[0008] In a preferred embodiment, the heterogeneous charging duration function is a piecewise linear function, dividing the entire state of charge domain into a first segment and a second segment using the state of charge value at the inflection point of charging power decay as the dividing point, wherein: Within the first segment, the charging time is obtained by multiplying the difference between the initial state of charge value and the final state of charge value by the product of the battery nominal capacity and the battery health state value, and then dividing by the average value of the actual charging power within the first segment. In the second segment, the charging time is obtained by multiplying the difference between the initial state of charge value and the final state of charge value by the product of the battery nominal capacity and the battery health state value, and then dividing by the average value of the actual charging power in the second segment. When the target state of charge value exceeds the state of charge value at the inflection point of charging power decay, the total charging time is the sum of the complete charging time of the first segment and the charging time of the second segment.

[0009] In a preferred embodiment, the step of obtaining passenger flow data from multiple data sources includes: obtaining a time series of the number of people passing through security checks every second from the security check system; obtaining the number of passengers who have checked in for each flight and special passenger markers from the check-in system; obtaining boarding gate allocation information and the regional attributes of the main terminal or satellite terminal from the flight database; and obtaining train departure timetables and the maximum passenger capacity per train from the automated people mover system.

[0010] In a preferred embodiment, the discrete event queuing model uses the automated people mover system platform as a finite-capacity queue, and executes passenger arrival events and train arrival service events alternately in chronological order, wherein: In a passenger arrival event, the total number of special passenger arrivals for all satellite terminal flights within each minute time step is cumulatively added to the queue; In train arrival service events, the service consumption for each event is the smaller of the current number of people in the queue and the maximum passenger capacity of a single train. The served passengers are marked as boarding passengers for the corresponding train, while unserved passengers are held up until the next train. For each group of boarding passengers, the arrival time of that group of passengers at the boarding gate area is obtained by adding the Automated People Mover system operating time and the walking time in the satellite terminal.

[0011] In a preferred embodiment, the calculation of the state of charge threshold value for satellite terminal services includes: obtaining the round-trip path distance from the charging pile location to the boarding gate of the satellite terminal based on airport road network data; multiplying the round-trip path distance by the energy consumption rate per unit distance and adding a safety energy margin to obtain the minimum actual energy value; and dividing the minimum actual energy value by the product of the battery nominal capacity and the battery health status value to obtain the state of charge threshold value for satellite terminal services.

[0012] In a preferred embodiment, the constraints of the mixed-integer linear programming problem include: Single allocation constraint: Each vehicle can be assigned to at most one pulse group; Demand ceiling constraint: The number of vehicles allocated to each pulse group shall not exceed its demand quantity; Candidate set constraint: Vehicles are only assigned to the pulse group corresponding to their respective dual feasible candidate sets; Charging station time occupancy constraint: The piecewise linear expression of the heterogeneous charging duration function ensures that the charging time periods of different vehicles on the same charging station do not overlap; Time alignment constraint: The charging completion time of the assigned vehicle shall not be later than the ready cutoff time of the corresponding pulse group.

[0013] In a preferred embodiment, the aging safety constraint verification includes: comparing the target state of charge value of each vehicle with the upper limit of the derating charging cutoff state of charge corresponding to its battery health state; for vehicles whose target state of charge value exceeds the upper limit of the derating charging cutoff state of charge, reverting the target state of charge to the upper limit of the derating charging cutoff state of charge, and redistributing the released charging pile time to the vehicle with the lowest actual available energy in the fleet.

[0014] In a preferred embodiment, the execution includes the following steps: The system receives real-time updates on the queuing status of the automated passenger rapid transit system. When the arrival time of a pulse deviates, the system recalculates the ready cutoff time for the affected pulse group. For time-infeasible allocation relationships, the system selects a substitute vehicle from the set of dual feasible candidate vehicles to perform a vehicle replacement. The actual charging time of each vehicle is compared with the predicted value calculated based on the heterogeneous charging time function. When the deviation rate between the actual charging time and the predicted value exceeds the set threshold, the piecewise fitting parameters in the battery charging characteristic parameter set of that vehicle are updated.

[0015] A collaborative scheduling system for intelligent auxiliary mobility vehicles at airports includes the following modules: The battery characteristic extraction module is used to acquire the battery management system recorded data of each vehicle in the fleet, extract the average actual charging power and the state of charge value at the inflection point of charging power decay, and generate a set of vehicle-level battery charging characteristic parameters, a heterogeneous charging time function, and a mapping table between state of charge and actual energy. The multi-source data acquisition module is used to acquire data from the security check system, check-in system, flight database and automated passenger rapid transit system to generate a multi-source passenger flow dataset containing regional attribute labels; The pulse demand forecasting module is used to decompose the aggregated traffic in the security checkpoint into the contribution components of each flight, establish a discrete event queuing model for the destination flow in the satellite terminal to perform event-by-event extrapolation, and output the pulsed passenger arrival time distribution and the demand for chauffeured vehicles in each pulse group. The energy accessibility calculation module is used to calculate the satellite terminal service charge state threshold value for each vehicle based on airport road network data and the state of charge mapping table with actual energy. The dual feasible candidate screening module is used to back-calculate the vehicle readiness deadline from the arrival time of each pulse group, and use the heterogeneous charging duration function to screen vehicles that simultaneously meet the energy feasibility and time feasibility conditions, and output the dual feasible candidate vehicle set for each pulse group. The joint optimization solution module is used to solve the target state of charge value and pulse group allocation for each vehicle by using mixed integer linear programming with the goal of maximizing demand coverage. The safety verification and adjustment module is used to perform derating charging cutoff charge state upper limit verification on the target charge state value, reverse overloaded vehicles, and reallocate the released charging pile time. The dynamic execution and feedback module is designed to issue the final collaborative scheduling scheme, receive queuing status update data in real time to execute vehicle replacement, and update the segmented fitting parameters when the charging time deviation exceeds the set threshold.

[0016] The beneficial effects of this invention are: 1. This invention generates a vehicle-level heterogeneous charging time function by performing piecewise linear fitting on the charging behavior of each vehicle according to the inflection point of charging power decay. The accuracy of the charging time estimation for each vehicle is improved compared with the unified linear model, and the time extension effect of aging vehicles in the constant voltage charging stage is accurately quantified.

[0017] 2. By combining nonnegative matrix decomposition with a discrete event queuing model, this invention accurately characterizes the pulsed batch arrival pattern of passengers in the satellite terminal. The demand forecast for mobility scooters is modified from the continuous smooth assumption to a pulsed distribution that conforms to the train departure constraints of the automated people mover system, thereby reducing the systematic deviation between vehicle readiness time and actual demand time.

[0018] 3. By calculating the satellite terminal service state of charge threshold value for each vehicle individually, the higher state of charge requirement for vehicles with higher aging is included in the candidate screening conditions, thus suppressing the situation where vehicles with insufficient energy are mistakenly dispatched.

[0019] 4. This invention uses mixed integer linear programming to jointly solve for the optimal target state of charge value and pulse group allocation, and supplements it with aging safety constraint verification and charging pile time release redistribution. Under the condition of limited charging pile resources, the demand coverage of each pulse group is improved, while the risk of battery overcharging is controlled.

[0020] 5. This invention monitors the deviation between the actual charging time and the predicted value through a dynamic feedback module, and updates the segmented fitting parameters of the corresponding vehicle when the deviation exceeds a set threshold. The estimation accuracy of the heterogeneous charging time function is continuously corrected as the operating cycle accumulates. Attached Figure Description

[0021] Figure 1 This is a flowchart of the overall process of the airport intelligent auxiliary mobility vehicle collaborative scheduling method provided in the embodiments of the present invention; Detailed Implementation

[0022] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example Reference Appendix Figure 1 This invention provides a collaborative scheduling method for intelligent mobility scooters at airports. This method is applied to large hub airports comprising a main terminal and satellite terminals. The battery health status of each scooter in the fleet is distributed within a preset range, and passengers must reach the satellite terminal boarding gates via an automated people mover system. The collaborative scheduling method includes the following steps: Step S100: Obtain the battery management system recorded data of each vehicle in the fleet, extract the average actual charging power and the state of charge value at the inflection point of charging power decay for each vehicle in different state of charge ranges, and generate a set of vehicle-level battery charging characteristic parameters.

[0024] In this step, data recorded by the battery management system of each vehicle in the fleet is collected, including the current battery health status value. Measured state of charge and charging power curves for the most recent charging cycles, and current measured cell internal resistance. and cell temperature ,in Indicates the first in the team The identification number of the vehicle.

[0025] In one embodiment of the present invention, the measured curves of multiple charging cycles of each personal mobility vehicle are statistically processed to extract the average actual charging power of the personal mobility vehicle in several discrete state-of-charge intervals. and the state of charge value at the inflection point of charging power decay ,in Indicates the first The numbering of each state-of-charge interval. The state-of-charge value at the inflection point of charging power decay. It is defined as the state of charge (SOC) value at which the charging power transitions from a high power level in the constant current stage to a low power level in the constant voltage stage as the charging power decay value increases. The SOC value at the inflection point of charging power decay exhibits a heterogeneous distribution due to different battery aging degrees; vehicles with higher aging degrees have relatively lower SOC values ​​at the inflection point of charging power decay.

[0026] Therefore, the vehicle-grade battery charging characteristic parameter set It is generated, and its expression is:

[0027] in, For the first The battery health status value of a mobility scooter. For the first The measured internal resistance of the battery cells in a mobility scooter. For the first The battery cell temperature of a mobility scooter For the first The state of charge value at the inflection point of charging power degradation for a mobility scooter. For the first The first commuter car The average actual charging power within each state-of-charge interval This represents the number of divisions within the charge state interval.

[0028] Step S200: Based on the vehicle-level battery charging characteristic parameter set, the relationship between the state of charge and charging time of each vehicle is segmented according to the inflection point of charging power decay. Within each segment, a linear approximation is used to fit the charging time required for the vehicle to charge from the current state of charge to any target state of charge. At the same time, the target state of charge is converted into the actual usable energy value of the vehicle using the battery health status value, generating a heterogeneous charging time function and a state of charge and actual energy mapping table for each vehicle.

[0029] It should be noted that in this step, for each mobility scooter, the inflection point of its charging power degradation is used. Using this as a dividing point, the entire state of charge domain is divided into two segments: the first segment is from the current state of charge... to The interval, the second segment is from To the target state of charge The range.

[0030] During the first segment, the charging power remains approximately constant at a high level. From the state of charge Charge to (in Required charging time Calculated as:

[0031] in, This represents the initial state of charge value within the first segment. This represents the termination state of charge value within the first segment. This refers to the battery's nominal capacity (in watt-hours). For the first The vehicle's battery health status value. This represents the average actual charging power (in watts) of the vehicle within the first segment. The unit is watt-hour. The unit is watt, and the result of dividing the two is... The unit is hour, and the dimensions are consistent.

[0032] In the second segment, the charging power decreases to a lower level due to the increased internal resistance during the constant voltage charging phase. ,from Charge to (in Required charging time Calculated as:

[0033] in, This represents the average actual charging power of the vehicle within the second segment (in watts).

[0034] Therefore, the heterogeneous charging time function for each vehicle It is generated as a piecewise linear function:

[0035] in, For the first The current state of charge of the vehicle. For the first The target state of charge of a mobility scooter From Charge to The required full first-segment charging time is calculated in the same way as... Same, that is, let , Substituting into the first piecewise formula, we get:

[0036] Meanwhile, the state of charge mapping table to actual energy It is generated, which maps any state of charge value to the actual usable energy of the mobility scooter:

[0037] in, For any state of charge value, The unit is watt-hour.

[0038] Step S300: Obtain the number of passengers passing through security checks every second in real time from the security check system, obtain the number of passengers who have checked in for each flight and special passenger tags from the check-in system, obtain the boarding gate allocation information from the flight database and mark the area attribute of the boarding gate as the main building or satellite terminal, obtain the train departure timetable and the maximum passenger capacity per train from the automated passenger transport system, and generate a multi-source passenger flow dataset containing area attribute labels.

[0039] It should be noted that in step S300, information from multiple data sources is acquired in parallel. The number of people passing through security checks every second, obtained from the security check system, is recorded as a time series. ,in For timestamps, For timestamps The corresponding number of passengers passing through security per second. Data obtained from the check-in system includes information for each flight. Total number of passengers who have checked in And the number of special passengers marked as requiring mobility scooter services. ,in The flight number. Gate allocation information retrieved from the flight database includes the flight number. Corresponding gate number and its regional attributes ,in Indicates the main building. The satellite terminal is indicated. Data obtained from the Automated People Mover system includes train departure schedules. and maximum passenger capacity per train ,in For shift number, This represents the total number of shifts within the scheduling cycle.

[0040] Therefore, the multi-source passenger flow dataset includes regional attribute labels. Generated:

[0041] Step S400: The non-negative matrix factorization method is used to decompose the aggregated traffic of the security checkpoint into the minute-by-minute contribution components of each flight. The decomposition results are divided into the main terminal destination flow and the satellite terminal destination flow according to the boarding gate area attributes. A discrete event queuing model is established for the satellite terminal destination flow. The actual boarding flight number and corresponding departure time of each batch of passengers are calculated event by event. The pulsed passenger arrival time distribution of each boarding gate in the satellite terminal and the demand for shuttle vehicles in each pulse group are generated.

[0042] It should be noted that in step S400, the number of people passing the inspection every second is first aggregated into a traffic vector passing the inspection every minute. ,in This is a minute-level time index. A non-negative matrix factorization method is performed, with the aggregated flow matrix as input. ( (Constraints include the number of minutes within the scheduling cycle and the total number of passengers on each flight) Its output is the minute-by-minute contribution component matrix of each flight. ,in This represents the total number of flights involved in the current scheduling cycle. The The column represents the flight The minute-by-minute contribution to the aggregated traffic of security checks.

[0043] Based on boarding gate area attributes The decomposition result is divided into two parts: the main building destination flow. Includes all Flights, Satellite Terminal destination Includes all The flight schedule.

[0044] Extract the proportion of special passengers in the destination flow of the satellite terminal. (This refers to flight information.) ( (This refers to the sequence of special passenger numbers arriving at the Automated People Mover system platform every minute.) Calculated as:

[0045] in, For flights In minutes The contribution weight, For flights Special passenger numbers, For flights The total number of passengers who have checked in. The arrival of special passengers for all satellite terminal flights is superimposed onto a minute-by-minute sequence of the cumulative queue number of passengers on the platform. .

[0046] The discrete event queuing model uses the platforms of an automated people mover system as finite-capacity queues, and its operating logic is as follows: (a) Queue state initialization: At the start of the scheduling cycle, the number of people in the queue is... .

[0047] (b) Passenger arrival event: every minute time step New arriving passengers Added to the queue, the number of people in the queue is updated. ,in This represents the number of people in the queue at the end of the previous minute. For minutes Total number of special passengers arriving on all satellite terminal flights.

[0048] (c) Train arrival service event: when time Compared with a certain train in the departure timetable Departure time During alignment, a service event is triggered. The cost of each service event is [not specified]. The minimum of the current number of people in the queue and the remaining capacity of the train:

[0049] in, Departure time The number of people queuing at that time This represents the maximum passenger capacity for a single train. The train being served... Passenger number 1 was marked as the first Passengers boarding on this train / bus will have their corresponding departure times as follows: The number of people in the queue has been updated. ,in This indicates the queue status after service completion. Unserved passengers will be held until the next scheduled departure.

[0050] (d) Event-by-event deduction: The above arrival events and service events are executed alternately in chronological order until the end of the scheduling cycle.

[0051] In one embodiment of the present invention, the collaborative scheduling method further includes the following additional step: for each batch of boarding passengers, the running time of the automated people mover system is added. Walking time within the satellite hall ( (Using the destination gate number), calculate the time when this group of passengers arrives at the gate area. :

[0052] in, For the first The departure time of the bus / train. The travel time for Automated People Mover (APM) trains from the main platform to the satellite platform. From the satellite terminal platform to the boarding gate Walking time.

[0053] Thus, the pulsed passenger arrival time distribution for each boarding gate in the satellite terminal is generated, with each pulse group... By shift number Unique identifier, whose attributes include arrival time And the number of passengers in this pulse group who require mobility scooter services . That is, the first The number of special passengers boarding the train on a given flight.

[0054] The above method provides a general format for accurately characterizing the batch arrival patterns of passengers in satellite terminals. This allows the forecasting of vehicle demand to be corrected from a continuous, smooth distribution to a pulsed distribution that reflects actual physical processes.

[0055] Step S500: Obtain the round-trip path distance from each charging pile location to each boarding gate in the satellite terminal from the airport road network data. Calculate the minimum actual energy value required for each vehicle to complete a single round-trip mission in the satellite terminal based on the state of charge and actual energy mapping table. Use the minimum actual energy value to back-calculate the satellite terminal service state of charge threshold value for each vehicle through the state of charge and actual energy mapping table, and generate a vehicle-level satellite terminal energy accessibility parameter table.

[0056] It should be noted that in step S500, the first [item] is obtained from the airport road network data. Location of the charging station for the scooter To the boarding gate in the satellite terminal round trip distance Energy consumption per unit distance The minimum actual energy required to complete a single round trip to the satellite terminal is determined based on the vehicle's load and road resistance parameters. Calculated as:

[0057] in, The distance of the round trip (in meters). For the first Energy consumption rate per unit distance of a vehicle (in watt-hours per meter). This is a preset safety energy margin (in watt-hours). All three quantities are dimensionless. The unit is watt-hour.

[0058] Subsequently, this minimum actual energy value is mapped to the actual energy using a state of charge to energy mapping table. Recalculated as the first Satellite terminal service charge status threshold for vehicles :

[0059] in, This refers to the battery's nominal capacity (in watt-hours). For the first The vehicle's battery health status value (dimensionless). The unit is watt-hour, and the numerator and denominator have the same dimensions. This is a dimensionless ratio of states of charge.

[0060] This verifies that vehicles with a higher degree of aging ( (Lower) The satellite terminal's state of charge threshold is correspondingly higher, meaning it needs to be charged to a higher state of charge to meet the energy requirements for round-trip travel to the satellite terminal. Table of Energy Accessibility Parameters for Vehicle-Grade Satellite Terminals It was generated.

[0061] Step S600: Calculate the arrival time of each pulse group in the pulsed passenger arrival time distribution back into the vehicle departure time, and generate the vehicle readiness cutoff time sequence for satellite terminal service; for each vehicle in the fleet, calculate the charging time required to charge from the current state of charge to the state of charge threshold value for satellite terminal service based on its heterogeneous charging time function, compare the charging completion time with each time point in the readiness cutoff time sequence, and select vehicles whose charging completion time is earlier than a certain readiness cutoff time point as energy-time dual feasible candidate vehicles for that pulse group, and generate a dual feasible candidate vehicle set for each pulse group.

[0062] It should be noted that in step S600, for each pulse group Its arrival time The departure time of the vehicle is calculated backwards. :

[0063] in, For pulse group Corresponding boarding gate The arrival time of passengers, The vehicle was driven from the charging station area to the boarding gate in the satellite terminal. The required travel time to the standby area. Therefore, the vehicle readiness cutoff time sequence for satellite terminal services. It was generated.

[0064] The first in the team A mobility scooter, utilizing the heterogeneous charging duration function generated in step S200. Calculate from the current state of charge Charge to Satellite Terminal Service State of Charge Threshold Required charging time:

[0065] The charging completion time of the mobility scooter Calculated as:

[0066] in, The current time for scheduling calculation execution. For the first The charging time required for a vehicle to reach the satellite terminal's state of charge threshold.

[0067] For each pulse group Vehicles that meet both of the following conditions Vehicles selected as energy-time dual-feasibility candidate vehicles for this pulse group: Condition 1 (Energy Feasibility): ,in The upper limit of the derating charge cutoff state of charge for this mobility scooter; Condition 2 (Time Feasible): .

[0068] Therefore, the set of dual feasible candidate vehicles for each pulse group It was generated.

[0069] Step S700: With the goal of maximizing the demand coverage of all pulse groups, construct a constraint satisfaction problem for each pulse group and its set of dual feasible candidate vehicles. Use a mixed integer linear programming solver to jointly solve the optimal target state of charge value of each vehicle and the pulse group allocation, and generate a pulse-aligned heterogeneous charging collaborative scheduling scheme.

[0070] It should be noted that in step S700, the decision variables are defined as follows: binary variables Indicates the first Was the vehicle assigned to a pulse group? Continuous variables Indicates the first The optimal target state of charge value for the vehicle.

[0071] The objective function aims to maximize the required coverage of all pulse groups.

[0072] in, This represents the total number of pulse groups. For pulse group A set of two feasible candidate vehicles The required number of mobility scooters for this pulse group.

[0073] The constraints include: Constraint 1 (Single Assignment Constraint): Each vehicle can be assigned to at most one pulse group or left unassigned for main building service.

[0074] Constraint 2 (Demand Cap Constraint): The number of vehicles allocated to each pulse group shall not exceed its demand quantity.

[0075] Constraint 3 (Candidate Set Constraint): A vehicle can only be assigned to the pulse group corresponding to its bifeasible candidate set.

[0076] Constraint 4 (Charging Pile Time Occupancy Constraint): A piecewise linear expression of the heterogeneous charging duration function for each vehicle is adopted. For vehicles assigned to pulse groups... vehicles Its target state of charge is at least as high as the satellite terminal service state of charge threshold:

[0077] in, The standard state of charge (SBC) value for the main building service. The duration of charging station occupancy is determined by the heterogeneous charging duration function. The piecewise linear expression embedding constraint ensures that the charging time periods of different vehicles on the same charging pile do not overlap.

[0078] Constraint 5 (Time Alignment Constraint): Assigned to pulse group vehicles The charging completion time must be no later than the ready cutoff time of the pulse group:

[0079] in, The current time for scheduling calculation execution. For pulse group Corresponding boarding gate The ready deadline.

[0080] For vehicles not assigned to any pulse group ( Its target state of charge is set to the main building service standard value. .

[0081] The mixed-integer linear programming solver solves the constraint satisfaction problem described above and outputs the optimal target state of charge value for each vehicle. pulse group assignment This leads to the generation of a pulse-aligned heterogeneous charging collaborative scheduling scheme.

[0082] Furthermore, in the objective function The operation is a non-linear term, and when embedding a mixed-integer linear programming solver, auxiliary variables are introduced. Perform linearization: Let and Replace the objective function with ,in The variables are non-negative integers, thus ensuring that the objective function satisfies the linearity requirement of mixed-integer linear programming.

[0083] Based on this definition, the allocation relationship and target state of charge value can be combined very flexibly to maximize service coverage under the constraint of limited charging pile resources.

[0084] Step S800: Perform aging safety constraint verification on the pulse-aligned heterogeneous charging collaborative scheduling scheme, compare the optimal target state of charge value of each vehicle with the upper limit of the derating charging cutoff state of charge corresponding to its battery health state, roll back the target state of charge of vehicles exceeding the limit to the safe upper limit value, and reallocate the released charging pile time to the vehicle with the lowest actual available energy in the fleet, update the target state of charge value and pulse group allocation relationship of the affected vehicles, and generate the verified final collaborative scheduling scheme.

[0085] It should be noted that in step S800, for each mobility scooter... Its derating charge cutoff state of charge upper limit Based on battery health status value Confirmed. The upper limit of the derating charge cutoff state of charge is the upper limit of the safe charging state of charge for the battery under its current aging condition. Charging operations exceeding this upper limit will accelerate battery life degradation.

[0086] Perform the following checks on each vehicle: like If so, the vehicle is marked as an oversized vehicle, and its target state of charge is reverted to:

[0087] The time released by the rollback operation for the charging station Calculated as:

[0088] The release time is the difference in charging time between the original target state of charge and the safe upper limit. The heterogeneous charging duration function generated in step S200, For the first The maximum state of charge limit for derating charging of a personal mobility vehicle. This represents the optimal target state of charge value before rollback.

[0089] In one embodiment of the present invention, after the overloaded vehicle returns to its target state of charge, it is necessary to check whether the actual available energy after the return still meets the energy requirements of the allocated pulse group. If If this happens, the pulse group allocation relationship for that vehicle is cancelled, and the vehicle will switch to serving the main building.

[0090] The released charging time is redistributed to the vehicles in the fleet with the lowest actual available energy. For example, vehicles in the fleet that are not currently charging and have the lowest actual available energy. The vehicle with the lowest charge is selected, and its target state of charge is appropriately increased from the original value. The increase is constrained by the release time and the heterogeneous charging duration function of the vehicle.

[0091] Based on the above method, a verified final cooperative scheduling scheme is generated, including the final target state of charge value for each vehicle. And the relationship between pulse group allocation.

[0092] Step S900: The target state of charge value and pulse group assignment of each vehicle in the final collaborative scheduling scheme are issued for execution. Vehicles assigned to pulse groups will exit the charging pile and head to the satellite hall waiting area according to the ready deadline after reaching the satellite hall service state of charge threshold value. During the charging process, the queuing status update data of the automated passenger rapid transit system will be received in real time. When the pulse arrival time deviates, the ready deadline of the affected pulse group will be recalculated and the time feasibility of the assigned vehicles will be checked. For the time infeasible allocation relationship, vehicle replacement will be performed. At the same time, the actual charging time of each vehicle will be compared with the predicted value. When the deviation exceeds the set threshold, the piecewise fitting parameters in the battery charging characteristic parameter set of the vehicle will be updated.

[0093] It should be noted that in step S900, the final coordinated scheduling plan is issued to the fleet management system for execution. This applies to those assigned to pulse groups. vehicles When its state of charge reaches during the charging process At that time, the vehicle entered the ready state and reached the ready deadline. Upon reaching the exit marker, proceed to the waiting area for the corresponding boarding gate in the satellite terminal.

[0094] In one embodiment of the invention, dynamic feedback processing is performed. The real-time queuing number at the automated people mover system platform is continuously monitored and compared with the derivation results of the discrete event queuing model in step S400. When the actual queuing state deviates from the predicted state (e.g., train delays causing departure time shifts), the arrival time of the affected pulse group is... Updated, its corresponding readiness deadline It was recalculated.

[0095] For vehicles already assigned to the affected pulse group Re-examine the feasibility conditions based on time:

[0096] in, For vehicles The moment when charging is complete. This is the updated readiness deadline.

[0097] If the above conditions are no longer met, the allocation relationship is marked as time-infeasible. Perform a vehicle replacement operation: from the set of doubly feasible candidate vehicles. Select the backup vehicle whose current charging completion time is closest to and no later than the updated ready deadline, and transfer the pulse group allocation relationship to the backup vehicle.

[0098] At the same time, for each vehicle that is charging Its actual charging time After charging is complete, it is recorded and compared with the predicted value. Comparison. Deviation rate Calculated as:

[0099] in, For the first The actual charging time of the vehicle This is a prediction of charging time calculated based on a heterogeneous charging duration function.

[0100] when Exceeding the set threshold At that time, update the battery charging characteristic parameter set of the vehicle. The piecewise fitting parameters are used. For example, if the actual time takes longer than the predicted value, the average charging power of the corresponding segment is adjusted downwards. If the actual charging time is shorter than the predicted value, the average charging power will be adjusted upwards. The updated parameters will be used in the collaborative optimization of the next scheduling cycle.

[0101] It should be further noted that the non-negative matrix factorization method used in step S400 is a conventional algorithm, and its input is the aggregated flow matrix. and the total number of passengers on each flight The output is the minute-by-minute contribution component matrix for each flight. ,in Each element is nonnegative and the sum of each column is subject to Constraints. The mixed-integer linear programming used in step S700 is a conventional algorithm, and its inputs are decision variables. and Given the objective function and all constraints, the output is the optimal solution that satisfies the constraints. and .

[0102] In one embodiment of the present invention, steps S100 to S900 are executed periodically, and the length of each scheduling cycle is set according to the flight schedule density and the departure interval of the automated people mover system. At the beginning of each scheduling cycle, steps S100 to S800 are executed sequentially to generate the final coordinated scheduling scheme for the current cycle, and the dynamic feedback processing of step S900 is continuously executed within the current cycle until the end of the cycle.

[0103] For example, when multiple flights have gates located in the satellite terminal and their departure times are close together within a certain time period, the number of people queuing at the automated people mover (APM) platform will accumulate. Full capacity throttling will cause some passengers to be delayed until later flights, resulting in multiple closely spaced pulse groups. In this case, the mixed-integer linear programming in step S700 will face more stringent charging pile time resource constraints. However, through the precise expression of the heterogeneous charging time function and the pre-screening of the dual feasible candidate set, the solution space is reduced, allowing the solver to output the optimal allocation scheme within an acceptable computation time.

[0104] The present invention also provides a collaborative scheduling system for intelligent auxiliary mobility vehicles at airports, the collaborative scheduling system comprising the following modules: The battery characteristic extraction module is used to acquire the battery management system recorded data of each vehicle in the fleet, extract the average actual charging power and the state-of-charge value at the inflection point of charging power decay for each vehicle in different state-of-charge ranges, and generate a set of vehicle-level battery charging characteristic parameters. Based on the vehicle-level battery charging characteristic parameter set, the charging time relationship of each vehicle is piecewise linearly fitted according to the inflection point of charging power decay, generating a heterogeneous charging time function. The battery health status value is used to convert the state of charge (SOC) into actual usable energy, generating a mapping table between SOC and actual energy. .

[0105] The multi-source data acquisition module is used to obtain the number of passengers passing through security checks every second from the security check system, the number of passengers who have checked in for each flight and special passenger tags from the check-in system, the boarding gate allocation information and regional attribute labels from the flight database, and the train departure timetable and maximum passenger capacity per train from the automated people mover system, generating a multi-source passenger flow dataset containing regional attribute labels. .

[0106] The pulse demand prediction module is used to predict the received multi-source passenger flow dataset. The nonnegative matrix factorization method is used to decompose the aggregated traffic flow at the security checkpoint into minute-by-minute contribution components of each flight and classify them according to regional attributes. A discrete event queuing model is established for the destination flow in the satellite terminal to perform event-by-event extrapolation, outputting the pulsed passenger arrival time distribution of each boarding gate in the satellite terminal and the demand for chauffeured vehicles in each pulse group. .

[0107] The energy accessibility calculation module is used to obtain the round-trip path distance from the airport road network data and combine it with the state of charge and actual energy mapping table. Calculate the minimum actual energy required for each mobility scooter to complete a single round trip to the satellite terminal. And calculate it back into the satellite hall service state of charge threshold value for each vehicle. Output a table of energy accessibility parameters for vehicle-grade satellite terminals.

[0108] The dual-feasibility candidate selection module is used to calculate the vehicle readiness deadline by working backward from the arrival time of each pulse group output by the pulse demand prediction module and the state of charge threshold value output by the energy accessibility calculation module. It then uses a heterogeneous charging duration function to calculate the charging completion time of each vehicle, and selects vehicles that simultaneously meet both energy and time feasibility conditions for each pulse group, outputting a dual-feasibility candidate vehicle set for each pulse group. .

[0109] The joint optimization solution module is used to receive the set of two feasible candidate vehicles and the demand quantity of each pulse group. With the objective function of maximizing demand coverage, and with constraints such as single allocation constraint, demand upper limit constraint, candidate set constraint, charging pile time occupancy constraint and time alignment constraint, a mixed integer linear programming solver is used to jointly solve the optimal target state of charge value of each vehicle and the pulse group allocation, and output a pulse-aligned heterogeneous charging collaborative scheduling scheme.

[0110] The safety verification and adjustment module is used to receive the output scheme of the joint optimization solution module, compare the optimal target state of charge value of each vehicle with its upper limit of the derated charging cutoff state of charge, perform target state of charge rollback and pulse group allocation relationship verification for vehicles exceeding the limit, reallocate the released charging pile time to the vehicle waiting to be charged with the lowest actual available energy, and output the verified final collaborative scheduling scheme.

[0111] The dynamic execution and feedback module is used to send the final collaborative scheduling plan to the fleet management system for execution, receive real-time updates of queuing status data from the automated passenger rapid transit system, recalculate the ready deadline and execute vehicle replacement when the pulse arrival time deviates, and monitor the actual charging time of each vehicle and update the piecewise fitting parameters in the battery charging characteristic parameter set when the deviation exceeds the set threshold.

[0112] It should be noted that the data transfer relationships between the above modules are as follows: the output of the battery characteristic extraction module is transferred to the dual feasible candidate selection module and the joint optimization solution module; the output of the multi-source data acquisition module is transferred to the pulse demand prediction module; the output of the pulse demand prediction module is transferred to the dual feasible candidate selection module and the joint optimization solution module; the output of the energy reachability calculation module is transferred to the dual feasible candidate selection module; the output of the dual feasible candidate selection module is transferred to the joint optimization solution module; the output of the joint optimization solution module is transferred to the safety verification and adjustment module; the output of the safety verification and adjustment module is transferred to the dynamic execution and feedback module; and the parameter update results of the dynamic execution and feedback module are fed back to the battery characteristic extraction module for the next scheduling cycle.

[0113] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement steps S100 to S900 of the above-described collaborative scheduling method for intelligent auxiliary mobility vehicles at airports.

[0114] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements steps S100 to S900 of the above-described collaborative scheduling method for intelligent auxiliary mobility vehicles at airports.

[0115] The following provides a further explanation of this embodiment using specific parameter scenarios: This embodiment is applied to a large international hub airport, which includes a main terminal and a satellite terminal. The fleet is equipped with 50 intelligent auxiliary mobility vehicles. The battery health status values ​​of each vehicle are distributed within a preset range (e.g., between low and high values). The departure interval of the automated people mover system is a fixed value (e.g., several minutes). The maximum passenger capacity of a single train is a preset capacity value. The average round-trip path distance between the boarding gate of the satellite terminal and the charging pile area of ​​the main terminal is a preset distance value.

[0116] In step S100, for a certain commuter vehicle with a low battery health status value, its battery management system data is obtained, and the extracted charging power decay inflection point state of charge value is located in the medium to high state of charge range. In step S200, for the commuter vehicle, a linear fitting with a larger slope is used in the state of charge interval below the charging power decay inflection point (corresponding to the higher charging power in the constant current charging stage), and a linear fitting with a smaller slope is used in the state of charge interval above the charging power decay inflection point (corresponding to the decrease in charging power due to the increase in internal resistance in the constant voltage charging stage). This generates a heterogeneous charging time function for the commuter vehicle. Compared with vehicles with higher battery health status values, the charging time slope of the aging vehicle is significantly increased in the interval above the charging power decay inflection point. In step S300, the check-in passenger data for a certain flight is obtained, wherein special passengers marked as needing shuttle service and whose destination is the satellite terminal constitute the input data for the flight's demand for shuttle service in the satellite terminal; In step S400, for the aforementioned special passenger flow, a minute-by-minute cumulative sequence of passenger arrivals at the Automated People Mover (APM) platform is generated. A discrete event queuing model is executed: when the first train arrives, the service consumption is the minimum of the current queue size and the train's maximum passenger capacity. If the queue size does not exceed the capacity, all passengers board without delay. After superimposing the train travel time and walking time within the satellite concourse, the pulsed arrival time of this group of passengers and the corresponding demand for mobility scooters are determined. In step S500, based on the round-trip path distance between the satellite terminal boarding gate and the charging pile area, the minimum actual energy value required for the aging vehicle to complete a single round trip is calculated. By using the state of charge and actual energy mapping table, the threshold value of the state of charge for the aging vehicle in satellite terminal service is obtained, which is higher than the corresponding threshold value for vehicles with higher battery health status. In step S600, the vehicle readiness deadline is calculated by working backward from the arrival time of the pulse group. For the aging vehicle, the time required to charge from the current state of charge to the satellite terminal service state of charge threshold is calculated based on its heterogeneous charging duration function. The calculated charging completion time is compared with the readiness deadline. If the charging completion time is earlier than the readiness deadline, the aging vehicle is selected into the dual feasible candidate vehicle set of the pulse group. In step S700, the mixed integer linear programming solver performs joint optimization on the heterogeneous charging time of 50 vehicles and the demand for all pulse groups, and outputs the optimal target state of charge value and pulse group allocation for each vehicle. In step S800, during the verification process, a vehicle with a high degree of aging is detected. The optimal target state of charge value obtained by solving the mixed integer linear programming exceeds the upper limit of the derating charging cutoff state of charge for the vehicle with a high degree of aging. The system rolls back its target state of charge to the safe upper limit value, and the released charging time is allocated to another vehicle in the fleet with the lowest actual available energy. In step S900, a charging time deviation threshold is set to a preset ratio. If the deviation rate of the actual charging time of a vehicle from the predicted value exceeds the charging time deviation threshold, the system automatically updates the segmented fitting parameters of the vehicle (increasing the charging time slope of the corresponding segment for cases where the actual charging time is longer), and the updated segmented fitting parameters are used in the next scheduling cycle.

[0117] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A collaborative scheduling method for intelligent auxiliary mobility vehicles at airports, characterized in that, Includes the following steps: Obtain the battery management system recorded data of each vehicle in the fleet, extract the average actual charging power and the state of charge value at the inflection point of charging power decay for each vehicle in different state of charge ranges, and generate a set of vehicle-level battery charging characteristic parameters. Based on the set of vehicle-level battery charging characteristic parameters, the relationship between the state of charge and charging time of each vehicle is piecewise linearly fitted according to the inflection point of charging power decay, generating a heterogeneous charging time function and a mapping table between the state of charge and actual energy for each vehicle. Passenger flow data is obtained from multiple data sources and boarding gate area attributes are labeled. The aggregated traffic at the security checkpoint is decomposed into the contribution components of each flight. A discrete event queuing model is established for the destination flow in the satellite terminal to perform event-by-event simulation, generating the pulsed passenger arrival time distribution of each boarding gate in the satellite terminal and the demand for chauffeured vehicles in each pulse group. Based on the state of charge mapping table and airport road network data, calculate the state of charge threshold value for satellite terminal service for each vehicle. The arrival time of each pulse group is deduced backward to the vehicle readiness deadline, and dual feasible candidate vehicles whose charging completion time is earlier than the readiness deadline and meet the energy feasibility conditions are screened. With the goal of maximizing demand coverage, a mixed-integer linear programming approach is used to jointly solve the target state of charge value and pulse group allocation for each vehicle, thereby generating a collaborative scheduling scheme. The aging safety constraint verification is performed on the collaborative scheduling scheme and then the scheme is issued for execution.

2. The collaborative scheduling method for intelligent auxiliary mobility vehicles at airports according to claim 1, characterized in that, The vehicle-grade battery charging characteristic parameter set includes battery health status value, cell internal resistance measurement value, cell temperature, charging power decay inflection point state of charge value, and the average actual charging power within each discrete state of charge interval. The charging power decay inflection point state of charge value is defined as the state of charge value corresponding to the transition of charging power from the constant current stage to the constant voltage stage.

3. The collaborative scheduling method for airport intelligent auxiliary mobility vehicles according to claim 2, characterized in that, The heterogeneous charging duration function is a piecewise linear function. The entire state of charge domain is divided into a first segment and a second segment, with the state of charge value at the inflection point of charging power decay as the dividing point. Where: Within the first segment, the charging time is obtained by multiplying the difference between the initial state of charge value and the final state of charge value by the product of the battery nominal capacity and the battery health state value, and then dividing by the average value of the actual charging power within the first segment. In the second segment, the charging time is obtained by multiplying the difference between the initial state of charge value and the final state of charge value by the product of the battery nominal capacity and the battery health state value, and then dividing by the average value of the actual charging power in the second segment. When the target state of charge value exceeds the state of charge value at the inflection point of charging power decay, the total charging time is the sum of the complete charging time of the first segment and the charging time of the second segment.

4. The collaborative scheduling method for airport intelligent auxiliary mobility vehicles according to claim 3, characterized in that, The process of obtaining passenger flow data from multiple data sources includes: obtaining a time series of the number of people passing through security checks every second from the security check system; obtaining the number of passengers who have checked in for each flight and special passenger markers from the check-in system; obtaining boarding gate allocation information and the regional attributes of the main terminal or satellite terminal from the flight database; and obtaining train departure timetables and the maximum passenger capacity per train from the automated people mover system.

5. The collaborative scheduling method for airport intelligent auxiliary mobility vehicles according to claim 4, characterized in that, The discrete event queuing model uses the automated people mover system platform as a finite-capacity queue, and executes passenger arrival events and train arrival service events alternately in time sequence, wherein: In a passenger arrival event, the total number of special passenger arrivals for all satellite terminal flights within each minute time step is cumulatively added to the queue; In train arrival service events, the service consumption for each event is the smaller of the current number of people in the queue and the maximum passenger capacity of a single train. The served passengers are marked as boarding passengers for the corresponding train, while unserved passengers are held up until the next train. For each group of boarding passengers, the arrival time of that group of passengers at the boarding gate area is obtained by adding the Automated People Mover system operating time and the walking time in the satellite terminal.

6. The collaborative scheduling method for airport intelligent auxiliary mobility vehicles according to claim 5, characterized in that, The calculation of the state of charge threshold value for satellite terminal services includes: obtaining the round-trip path distance from the charging pile location to the boarding gate of the satellite terminal based on airport road network data; multiplying the round-trip path distance by the energy consumption rate per unit distance and adding the safety energy margin to obtain the minimum actual energy value; and dividing the minimum actual energy value by the product of the battery nominal capacity and the battery health status value to obtain the state of charge threshold value for satellite terminal services.

7. The collaborative scheduling method for airport intelligent auxiliary mobility vehicles according to claim 6, characterized in that, The constraints of the mixed-integer linear programming problem include: Single allocation constraint: Each vehicle can be assigned to at most one pulse group; Demand ceiling constraint: The number of vehicles allocated to each pulse group shall not exceed its demand quantity; Candidate set constraint: Vehicles are only assigned to the pulse group corresponding to their respective dual feasible candidate sets; Charging station time occupancy constraint: The piecewise linear expression of the heterogeneous charging duration function ensures that the charging time periods of different vehicles on the same charging station do not overlap; Time alignment constraint: The charging completion time of the assigned vehicle shall not be later than the ready cutoff time of the corresponding pulse group.

8. The collaborative scheduling method for airport intelligent auxiliary mobility vehicles according to claim 7, characterized in that, The aging safety constraint verification includes: comparing the target state of charge value of each vehicle with the upper limit of the derating charging cutoff state of charge corresponding to its battery health state; for vehicles whose target state of charge value exceeds the upper limit of the derating charging cutoff state of charge, the target state of charge is rolled back to the upper limit of the derating charging cutoff state of charge, and the released charging pile time is redistributed to the vehicle with the lowest actual available energy in the fleet.

9. The collaborative scheduling method for intelligent auxiliary mobility vehicles at airports according to claim 8, characterized in that, The execution process includes the following steps: The system receives real-time updates on the queuing status of the automated passenger rapid transit system. When the arrival time of a pulse deviates, the system recalculates the ready cutoff time for the affected pulse group. For time-infeasible allocation relationships, the system selects a substitute vehicle from the set of dual feasible candidate vehicles to perform a vehicle replacement. The actual charging time of each vehicle is compared with the predicted value calculated based on the heterogeneous charging time function. When the deviation rate between the actual charging time and the predicted value exceeds the set threshold, the piecewise fitting parameters in the battery charging characteristic parameter set of that vehicle are updated.

10. A collaborative scheduling system for airport intelligent auxiliary mobility scooters, used to execute the collaborative scheduling method for airport intelligent auxiliary mobility scooters as described in any one of claims 1 to 9, characterized in that, Includes the following modules: The battery characteristic extraction module is used to acquire the battery management system recorded data of each vehicle in the fleet, extract the average actual charging power and the state of charge value at the inflection point of charging power decay, and generate a set of vehicle-level battery charging characteristic parameters, a heterogeneous charging time function, and a mapping table of state of charge and actual energy. The multi-source data acquisition module is used to acquire data from the security check system, check-in system, flight database and automated passenger rapid transit system to generate a multi-source passenger flow dataset containing regional attribute labels; The pulse demand forecasting module is used to decompose the aggregated traffic in the security checkpoint into the contribution components of each flight, establish a discrete event queuing model for the destination flow in the satellite terminal to perform event-by-event extrapolation, and output the pulsed passenger arrival time distribution and the demand for chauffeured vehicles in each pulse group. The energy accessibility calculation module is used to calculate the satellite terminal service charge state threshold value for each vehicle based on airport road network data and the state of charge mapping table with actual energy. The dual feasible candidate screening module is used to back-calculate the arrival time of each pulse group to the vehicle readiness deadline, and use the heterogeneous charging duration function to screen vehicles that simultaneously meet the energy feasibility and time feasibility conditions, and output the dual feasible candidate vehicle set for each pulse group. The joint optimization solution module is used to solve the target state of charge value and pulse group allocation for each vehicle by using mixed integer linear programming with the goal of maximizing demand coverage. The safety verification and adjustment module is used to perform derating charging cutoff charge state upper limit verification on the target charge state value, and to perform backtracking and reallocation of the released charging pile time for over-limit vehicles. The dynamic execution and feedback module is used to issue the final collaborative scheduling plan, receive queuing status update data in real time to execute vehicle replacement, and update the segmented fitting parameters when the charging time deviation exceeds the set threshold.

Images