Air-rail intermodal robust scheduling and dynamic supply-demand matching method

Abstract

This application provides a robust scheduling and dynamic supply-demand matching method for multi-agent air-rail intermodal transport, comprising: obtaining aircraft timetables, train timetables, aircraft capacity, train capacity, and their respective maintenance window information from a preset database; obtaining an initial set of fixed binding candidate pairs by comparing the time overlap interval with the capacity ratio; incorporating fixed binding pairs and semi-fixed binding pairs into a cross-network resource view; generating a complete binding relationship set based on the automatic generation rules of baggage direct-check tags for fixed binding pairs and the requirements of backup connecting vehicle scheduling channels for semi-fixed binding pairs; extracting the real-time inventory synchronization lock status of aircraft and train ticketing systems and the capacity ratio matching status from the complete binding relationship set; and determining whether service continuity is met by analyzing the inventory linkage release and fine-tuning results under a disturbance simulation scenario.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a robust scheduling and dynamic supply-demand matching method for air-rail intermodal transport. Background Technology

[0002] As an important component of modern integrated transportation, air-rail intermodal transport effectively combines the high-speed and long-distance advantages of aviation with the large capacity and punctuality of rail transit, thereby significantly improving the convenience of passenger travel and overall transportation efficiency.

[0003] This potential for cross-modal collaborative operation makes the rational scheduling of air-rail vehicle resources a core element determining service quality and system efficiency.

[0004] Currently, most air-rail intermodal transport practices are still based on loose connections, with flight and train timetables being compiled independently, and there is a lack of close pre-correspondence between the two networks for train resources.

[0005] While this approach maintains the independence of each network, it often leads to passengers facing long waiting times, severe capacity mismatches, and connection failures during transfers. Especially during peak periods or inclement weather, reliable connecting links between aircraft and trains cannot be established, resulting in a significant decline in passenger experience.

[0006] The most prominent contradiction in air-rail intermodal transport lies in the fact that aircraft and rail vehicles have drastically different operational characteristics and management requirements.

[0007] Aircraft have short turnaround times and are greatly affected by weather and airspace control, while trains need to strictly adhere to fixed routes and frequent departure intervals. In addition, both must carry out their own independent maintenance plans and maintenance cycles.

[0008] These differences make it difficult to establish long-term, stable correspondences for specific flights and trains. Once a flight is delayed or a train malfunctions, the originally envisioned connecting service chain can easily break down, triggering a chain reaction that affects the normal operation of subsequent flights and trains.

[0009] Therefore, how to ensure the service stability and reliable transfer experience brought about by the long-term binding of train car resources, while retaining sufficient flexibility to cope with the sudden disturbances commonly seen in the aviation network, has become a key issue that the cross-network scheduling and binding strategy for air-rail train car resources needs to address. Summary of the Invention

[0010] This invention provides a robust scheduling and dynamic supply-demand matching method for multi-agent air-rail intermodal transport, mainly including: The system retrieves aircraft timetables, train timetables, aircraft capacity, train capacity, and their respective maintenance window information from a pre-set database. An initial set of fixed-binding candidate pairs is obtained by comparing the time overlap interval with the capacity ratio. Strict time alignment is calculated for this initial set. If the reserved time for aircraft and train transfers at shared physical stations is less than the minimum transfer standard, the corresponding candidate pair is removed, resulting in preliminary fixed-binding pairs that meet the strict one-to-one correspondence of fixed-binding times. Based on these preliminary fixed-binding pairs, the system determines whether there is overlap or conflict between aircraft and train maintenance windows. If the overlap duration exceeds a pre-set synchronization and coordination threshold, the pair is transferred to a semi-fixed-binding processing branch, resulting in a classification set distinguishing between fixed-binding and semi-fixed-binding candidate pairs. For the branch transferred to semi-fixed-binding candidate pairs, real-time passenger flow monitoring data and historical delay handling records are obtained. These are then processed using a dynamic seat release mechanism and backup... The allowable time slot adjustment window is calculated using train activation conditions to determine the elastic matching range between aircraft and trains under semi-fixed binding. Fixed binding pairs and semi-fixed binding pairs are uniformly included in the cross-network resource view. A complete binding relationship set is generated based on the automatic generation rules of baggage direct check-in tags for fixed binding pairs and the requirements of backup connecting vehicle scheduling channels for semi-fixed binding pairs. The real-time inventory synchronization lock status of aircraft and train ticketing systems and the proportional matching status of capacity are extracted from the complete binding relationship set. The service continuity is judged by the inventory linkage release and fine-tuning results under the disturbance simulation scenario. If the ticketing inventory linkage release is interrupted or the capacity fine-tuning exceeds the elastic matching range under the disturbance simulation scenario, the process is backtracked to the semi-fixed binding candidate pair processing branch. The latest real-time passenger flow monitoring data and historical delay processing records are incorporated to recalculate the dynamic seat release mechanism and time slot adjustment window, resulting in the final cross-network fixed and semi-fixed integrated collaborative binding scheme.

[0011] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a robust scheduling and dynamic supply-demand matching method for multi-agent air-rail intermodal transport. Addressing core business challenges in air-rail intermodal transport, such as difficulties in strictly aligning aircraft and train schedules, significant differences in capacity scale, frequent maintenance window conflicts, and the susceptibility to interruptions in binding relationships due to passenger flow fluctuations and delays, this invention obtains timetables, capacity scale, and maintenance window information from a pre-set database. First, it compares overlapping intervals with capacity ratios to generate initial fixed binding candidate pairs. Then, through rigorous alignment calculations, pairs with insufficient transfer time are eliminated, resulting in preliminary fixed binding pairs. Subsequently, it detects overlapping conflicts in maintenance windows, transferring pairs exceeding thresholds to semi-fixed branches. Combining real-time passenger flow and historical delay data, it dynamically calculates elastic matching intervals and floating adjustment windows, achieving the classification and fusion of fixed and semi-fixed bindings. Finally, it integrates these into a cross-network resource view, automatically generating baggage transfer and backup connection rules to form a complete binding relationship set. The service continuity of ticket inventory linkage and capacity fine-tuning is verified through disturbance simulation. If the standards are not met, the dynamic release mechanism is backtracked and optimized, iterating to obtain a robust collaborative binding scheme. This method effectively resolves the contradiction between rigid time constraints and dynamic supply-demand mismatch in air-rail intermodal transport scenarios, and improves the efficiency of cross-network resource collaboration and service reliability.

[0012] To enable those skilled in the art to better understand this solution, the key terms used in this document are defined as follows: 1. Fixed Binding: Refers to a strong correlation between aircraft and trains in the spatiotemporal dimension that is deterministic and meets rigid transfer time requirements. 2. Semi-Fixed Binding: Refers to a weak correlation where potential resource conflicts exist, but the transfer goal can be achieved by dynamically adjusting seat allocation or utilizing backup channels. 3. Cross-Network Resource View: Refers to a unified modeling of heterogeneous aviation and railway data, including a digital topology map containing geographic coordinates, timestamps, and capacity status information. 4. Dynamic Seat Release Mechanism: This refers to a dynamic strategy that automatically adjusts the threshold for opening virtual ticket inventory based on real-time passenger flow, transfer channel congestion (≤90% of design capacity), and remaining capacity of subsequent trains (≥10%). The release priority is "trains within the next 1 hour > trains within 1-3 hours > trains more than 3 hours". 5. Timetable Floating Adjustment Window: This refers to the adjustable time range (default ±10 minutes, maximum ±15 minutes) calculated based on the normal distribution of passenger walking speed (mean 1.2 m / s, variance 0.3 m / s), transfer channel length (≤800 meters), and mechanical delay of the baggage transfer system (≤5 minutes). 6. Backup Train Activation Condition: This refers to the criteria for triggering backup train scheduling when real-time passenger flow ≥ a preset threshold (channel design capacity 80%) and historical delay rate ≥30%. 7. Ticket Inventory Synchronous Lock Status: This refers to the simultaneous locking and release of reserved seats for connecting flights and high-speed trains. The linkage status is such that the locking ratio is consistent with the capacity ratio. Attached Figure Description

[0013] Figure 1 This is a flowchart of a robust scheduling and dynamic supply-demand matching method for air-rail intermodal transport according to the present invention.

[0014] Figure 2 This is a schematic diagram of the multi-agent collaborative scheduling interaction of the present invention.

[0015] Figure 3 This is a schematic diagram illustrating the fixed and semi-fixed fusion binding and inventory linkage backtracking of the present invention.

[0016] Figure 4 This is a heat map of the feasible region where the capacity ratio and time overlap are presented in this invention.

[0017] Figure 5 This is a graph showing the service continuity compliance rate of the present invention under disturbance scenarios.

[0018] Figure 6 This is a convergence curve diagram of inventory linkage release and fine-tuning in this invention.

[0019] Figure 7 This is a hierarchical mapping diagram of air-rail collaborative constraints in this invention.

[0020] Figure 8 This is a schematic diagram of the semi-fixed binding spatiotemporal elastic window of the present invention.

[0021] Figure 9 This is the boundary partitioning diagram for the collaborative scheduling decision of this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0023] In this embodiment, the multi-agent system includes an Air-Agent, a Rail-Agent, and a Coordinator-Agent for cross-network collaborative scheduling. The agents synchronize their states in real time via a distributed message bus. The Air-Agent obtains flight information in real time through the civil aviation data interface; the Rail-Agent collects train availability and on-time / delay information through the railway scheduling API; and the Coordinator-Agent calls the NSGA-III optimization module to generate a binding scheme and sends instructions via a RabbitMQ queue using JSON format messages. The three agents are decoupled using a publish-subscribe pattern, ensuring that a single point of failure does not affect the overall system operation.

[0024] Each agent uses the MQTT 3.1.1 protocol for data transmission, with data format in JSON (fields include: agent ID, state type, data value, and timestamp), and a synchronization frequency of once per minute; the remaining capacity vector of the aviation resource agent is expressed as "number of available seats, baggage compartment capacity (m)". 3 The core dimension is "sold seats, standing passengers, and remaining capacity percentage". The railway resource agent's carriage load status is based on "sold seats, standing passengers, and remaining capacity percentage". The aviation resource agent is responsible for encapsulating and processing the real-time location, maintenance status, and remaining capacity vector of aircraft. The railway resource agent is responsible for encapsulating and processing the train's operating section, physical transfer channel occupancy rate, and carriage load status. The cross-network collaborative scheduling agent is based on a multi-objective optimization algorithm, using the non-dominated sorting genetic algorithm (NSGA-Ⅲ). The objective functions include: min (average passenger transfer time), max (cross-network capacity utilization rate), and min (transfer failure rate). The constraints are time alignment constraint, maintenance window conflict-free constraint, and transfer channel capacity constraint (≤90% of design capacity). The optimal set of binding candidate pairs is solved within the constraint space submitted by each agent, thereby realizing self-organized collaboration between heterogeneous transportation networks. The iteration parameters of this algorithm are set as follows: population size 100, number of iterations 50, crossover probability 0.8, mutation probability 0.1. The constraint space includes: time alignment error ≤5 minutes, transfer time ≥20 minutes. Minutes, maintenance window overlap time ≤ 90 minutes, transfer channel capacity ≤ 90% of design capacity.

[0025] like Figure 2 As shown, the aviation resource agent and the railway resource agent connect to the flight status interface and the train scheduling interface, respectively, to continuously collect operational status, capacity reserves, and time information, and transmit them to the collaborative scheduling side through a standardized message mechanism. After receiving asynchronous status on the message bus, the collaborative scheduling agent calls the multi-objective optimization module to generate cross-network binding decisions, and then sends the scheduling instructions back to the aviation and railway agents for execution. This architecture achieves coordinated control of the aviation and railway networks through decoupled data collection, unified decision-making, and bidirectional transmission, ensuring the stability of fixed binding relationships while supporting rapid replanning in semi-fixed scenarios, thus improving cross-network resource scheduling efficiency and system fault tolerance.

[0026] like Figure 1As shown, the process of this invention begins with acquiring basic data. First, it matches the capacity ratio of aircraft and trains during overlapping time periods, and then performs strict time alignment and minimum transfer constraint verification. For candidate pairs whose maintenance window conflicts exceed a threshold, the system automatically switches to a semi-fixed binding processing branch, calculating the elastic window based on real-time passenger flow monitoring and historical delay information; for candidate pairs that meet the rigid conditions, it enters the fixed binding path. Subsequently, both types of binding are uniformly incorporated into the cross-network resource view, performing inventory synchronization locking, disturbance simulation, and fine-tuning correction. Finally, based on the service continuity judgment result, a collaborative binding scheme is output, and backtracking optimization is triggered for scenarios that do not meet the standards. This process forms a closed-loop robust scheduling mechanism of screening, fusion, evaluation, and backtracking.

[0027] like Figure 7 As shown, this diagram aligns the spatial layer, time window layer, and capacity layer into separate columns. Solid lines are used to advance within each layer, while dashed lines are used between layers to avoid overlapping. 801-804 correspond to spatial entities, 811-814 to window constraints, and 821-824 to resource determination. Gray window bands and capacity bars are used to display status intervals, thereby reducing text overlap and improving information hierarchy.

[0028] Specifically, this embodiment of a robust scheduling and dynamic supply-demand matching method for multi-agent air-rail intermodal transport may include: Step S101: Obtain aircraft timetables, train timetables, aircraft capacity scale, train capacity scale, and their respective maintenance window information from the preset database, and obtain an initial fixed binding candidate pair set by comparing the time overlap interval with the capacity ratio.

[0029] Aircraft and train timetables are retrieved from a pre-defined database. The overlapping time slots of aircraft and trains on the timeline are compared using the timetable data to obtain a set of overlapping time slots. For each overlapping time slot, the aircraft and train capacity for that time period are obtained. The capacity ratio is calculated by determining the ratio of aircraft capacity to train capacity within the overlapping time slot. If the capacity ratio is within a pre-defined reasonable range, the corresponding aircraft and train combination for that overlapping time slot is retained. After confirming that the capacity ratio meets the pre-defined reasonable range, maintenance window information is used to determine if there are maintenance conflicts between aircraft and trains within the overlapping time slots. If a maintenance conflict exists, the combination is removed. The retained aircraft and train combinations form the initial set of fixed-binding candidate pairs.

[0030] like Figure 4As shown, the horizontal axis represents the capacity ratio of aircraft to trains, and the vertical axis represents the length of the time overlap interval. The heat intensity reflects the feasibility score of candidate pairings under constraints. The figure shows that when the capacity ratio is within a reasonable range and the overlap time meets the minimum transfer requirement, the feasible area expands significantly; when the ratio deviates from the reasonable range or the overlap time is insufficient, the feasibility decreases significantly. This figure is used to guide the pre-screening strategy for candidate pairs, reducing invalid combinations from entering subsequent optimization calculations.

[0031] In one possible implementation, the process of retrieving aircraft and train timetables from a pre-defined database can be accomplished by querying structured data within the system.

[0032] For example, assuming the database stores aircraft flight information, including departure time, landing time, and route code, as well as train departure time, arrival time, and line number, the system will first extract this timetable data from the database based on the user's input query conditions, such as a specific date or region, ensuring that the data is updated in real time to reflect the latest scheduling adjustments. This acquired data provides the foundation for subsequent comparisons. Next, by comparing the timetable data with the overlapping intervals of aircraft and trains on the timeline, a set of overlapping time intervals is obtained.

[0033] Specifically, aircraft and train timetables can be mapped onto a unified time axis. For example, if aircraft A operates from 9:00 AM to 11:00 AM and train B operates from 10:00 AM to 12:00 PM, the overlapping period is calculated as 10:00 AM to 11:00 AM. All such overlapping periods are then collected for further analysis. This comparison method helps identify potential collaboration opportunities and avoids wasting resources. For each overlapping period, the corresponding aircraft and train capacity is obtained.

[0034] In one possible implementation, capacity can be defined as the number of available seats or payload. For example, for the overlapping time slot of 10:00 to 11:00, the system queries that aircraft A has a capacity of 200 seats and train B has a capacity of 500 seats. This data comes from historical records in the database or real-time monitoring to ensure accurate reflection of actual capacity. The capacity ratio is obtained by calculating the ratio of aircraft capacity to train capacity within the overlapping time slot.

[0035] For example, dividing 200 seats on an aircraft by 500 seats on a train yields a ratio of 0.4. If a reasonable range is pre-defined, between 0.3 and 0.7, it is considered appropriate. This ratio calculation helps assess resource matching and promotes efficient intermodal transport planning. If the capacity ratio is within a pre-defined reasonable range, the aircraft and train combinations corresponding to the overlapping time slots are retained.

[0036] In one possible implementation, the system determines whether there is a maintenance conflict between the aircraft and the train within the time overlap interval based on the maintenance window information. If a maintenance conflict exists, the combination is removed.

[0037] For example, maintenance window information might include a 30-minute routine inspection for aircraft A at 10:30 and a scheduled maintenance for train B at 10:45. If these windows overlap with other intervals, the system will automatically detect and eliminate the combination to avoid safety hazards. This judgment process is achieved through time window comparison, ensuring the feasibility of the combination and providing a reliable basis for subsequent binding, thereby improving overall scheduling efficiency and safety. The retained aircraft and train combinations form the initial set of candidate fixed binding pairs.

[0038] In one possible implementation, combinations such as aircraft A and train C with no conflicting overlap and appropriate proportions are included in the candidate pair set for further optimization decisions. For example, in a multimodal transport system, this can lead to a more stable resource allocation effect.

[0039] Example: Aircraft A (Flight CA123, Beijing Capital Airport - Shanghai Hongqiao Airport, departure 9:00 / landing 11:00, capacity 200 seats), Train B (G456, Shanghai Hongqiao Station - Nanjing South Station, departure 10:00 / arrival 12:00, capacity 500 seats); After comparing the timetables, the time overlap is 10:00-11:00; the capacity ratio = 200 / 500 = 0.4 (within the reasonable range of 0.3-0.7); the maintenance window for aircraft A is 14:00-15:00, and the maintenance window for train B is 16:00-17:00, with no maintenance conflict, therefore they are included in the initial fixed binding candidate pair set.

[0040] Step S102: For the initial set of fixed binding candidate pairs, strict time alignment calculation is performed. If the reserved time between the aircraft and the train in the transfer channel of the shared physical station is less than the minimum transfer standard, the corresponding candidate pair is eliminated to obtain the initial fixed binding pairs that meet the strict one-to-one correspondence of fixed binding time.

[0041] Step 1: Obtain initial fixed-binding candidate pair data containing aircraft and train schedules from the pre-established candidate pair database. For each candidate pair, extract its corresponding physical station information and transfer channel data to obtain a complete candidate pair time and space mapping dataset. Step 2: Based on the candidate pair time and space mapping dataset, use a time alignment calculation method to compare the arrival and departure time differences of aircraft and train schedules at physical stations one by one. If the time difference is less than a preset minimum standard threshold, mark the candidate pair as not meeting the conditions, obtaining a marked candidate pair filtering result set. Step 3: Filter the marked candidate pair filtering result set, removing all candidate pairs marked as not meeting the conditions, retaining candidate pairs that meet the time alignment and reserved time requirements, and determining the preliminary fixed-binding pair set. Step 4: For the initial set of fixed binding pairs, obtain the usage status of the transfer channel corresponding to each candidate pair at the physical station. Determine if the channel supports a one-to-one binding relationship. If the channel is conflicting (e.g., occupied by more than two binding pairs simultaneously) or unavailable (e.g., due to maintenance or congestion), remove the relevant candidate pairs to obtain the optimized set of fixed binding pairs. Step 5: Based on the optimized set of fixed binding pairs, check the matching accuracy of aircraft and train times one by one. Use a preset time window verification rule. If the matching accuracy exceeds the allowable range, further adjust the priority of candidate pairs to determine the final set of strictly aligned time binding pairs. Step 6: Integrate the data from the final set of strictly aligned time binding pairs to generate a complete binding relationship record containing physical station, transfer channel, and reserved time information. Determine if the record meets all preset constraints. If any non-compliance exists, backtrack to the optimized set for re-filtering to obtain a fixed binding result set that meets all standards such as time alignment, reserved time, and available transfer channels.

[0042] In one possible implementation, when obtaining initial fixed-binding candidate pair data from a pre-established candidate pair set database, one can first query the database for aircraft schedules, such as flight departure times at Beijing Capital International Airport and high-speed rail departure times at Beijing South Railway Station. Then, for each candidate pair, one can extract the corresponding physical station information, such as Terminal 3 of Beijing Capital International Airport and the underground transfer level of Beijing South Railway Station, as well as transfer channel data, such as the length and capacity of dedicated channels, thereby forming a complete time and space mapping dataset. This process is implemented through a database query language to ensure that the data is accurately mapped to specific station coordinates and channel identifiers.

[0043] For example, based on this mapping dataset, when using the time alignment calculation method, aircraft times (e.g., flight arrival time of 10:00 AM) and train times (e.g., high-speed rail departure time of 10:30 AM) can be compared one by one. The arrival and departure time difference at physical stations (e.g., airport-high-speed rail integrated stations) is calculated to be 30 minutes. If this time difference is less than the preset minimum standard threshold (e.g., 20 minutes), the candidate pair is marked as not meeting the conditions. The result set will record all such marks for easy subsequent processing.

[0044] In one possible implementation, by filtering the set of candidate pairs after labeling, a conditional filtering algorithm can be used to remove candidate pairs with too small a time difference. For example, combinations with a time difference of more than 25 minutes, such as the combination of flights and high-speed rail at Shanghai Hongqiao Airport, can be retained to ensure that the time alignment and reserved time requirements are met. This will determine the initial set of fixed binding pairs, which will give priority to the convenience of passenger transfers.

[0045] For example, when obtaining the usage status of the transfer channel corresponding to each candidate pair at the physical station for the initial set of fixed binding pairs, it is possible to check whether the channel supports a one-to-one binding relationship. For example, if the channel status shows a peak period conflict such as the channel capacity being full or under maintenance, the relevant candidate pair is removed. For example, a certain pair of Guangzhou Baiyun Airport and the high-speed rail station is removed, resulting in an optimized set of fixed binding pairs. This helps to avoid congestion in actual operation.

[0046] In one possible implementation, when checking the matching accuracy of aircraft and train times one by one based on the optimized set of fixed binding pairs, a preset time window verification rule is adopted, for example, the window is set to plus or minus 5 minutes. If the actual flight delay causes the matching accuracy to exceed the allowable range, such as more than 10 minutes, the priority of the candidate pair is adjusted to low level, and the final time strictly aligned binding pair set is determined. This adjustment process is implemented through a priority sorting algorithm to ensure the reliability of the binding pairs.

[0047] For example, by strictly aligning the final time binding sets for data integration, a complete binding relationship record can be generated, which includes physical sites such as Chengdu Shuangliu Airport, transfer channels such as dedicated elevators, and reserved time information such as a 15-minute buffer. Then, it is determined whether the record meets all preset constraints. For example, if the channel is unavailable, it is backtracked to the optimized set for re-filtering, and finally a fixed binding result set that meets all standards is obtained. This integration can improve the overall efficiency of air-rail intermodal transport.

[0048] Step S103: Based on the initial fixed binding pair, determine whether there is an overlap or conflict between the aircraft maintenance window and the train maintenance window. If the overlap duration of the maintenance windows exceeds the preset synchronization and coordination threshold, the binding pair is transferred to the semi-fixed binding processing branch to obtain a classification set that distinguishes between fixed binding pairs and semi-fixed binding candidate pairs.

[0049] Obtain a preliminary set of fixed binding pairs. For each preliminary fixed binding pair, obtain the start and end times of the aircraft maintenance window and the train maintenance window. Determine if there is an overlap between the start and end times of the aircraft maintenance window and the train maintenance window, obtaining the overlapping interval. If an overlap interval exists, calculate the corresponding duration of the overlapping interval, obtaining the overlapping duration. Determine if the overlapping duration exceeds a preset synchronization coordination threshold. If it exceeds the synchronization coordination threshold, classify the preliminary fixed binding pair into the semi-fixed binding candidate pair set. If the overlapping duration does not exceed the synchronization coordination threshold, retain the preliminary fixed binding pair in the fixed binding pair set. The retained fixed binding pairs must meet the triple conditions of "no maintenance window conflict + strict time alignment + available transfer channel". Through the above division operation, obtain the classification results of the fixed binding pair set and the semi-fixed binding candidate pair set.

[0050] The overlapping conflict judgment of the maintenance windows is essentially an exclusive analysis of the time that aircraft and trains occupy physical resources at shared sites. The aircraft maintenance window not only includes the scheduled maintenance time, but also calculates a 'dynamic unavailable time interval' with safety redundancy by fusing real-time maintenance support progress obtained from sensor data; the train maintenance window combines track occupancy status and maintenance instructions for the power contact network. When the intersection of the 'dynamic unavailable time intervals' of the two in the time domain exceeds the synchronization coordination threshold, the system adjusts the task priorities among multiple agents, shifting the binding pair from a hard time constraint to a flexible adjustment branch with probability distribution characteristics.

[0051] For example, in the shared station business scenario of air and high-speed rail transfer, after obtaining the initial set of fixed binding pairs, the start and end times of the aircraft maintenance window and the start and end times of the train maintenance window are obtained for each initial fixed binding pair.

[0052] Specifically, an aircraft maintenance window refers to the predetermined time range during which an aircraft is stopped at a station for necessary maintenance, including a series of operations such as airframe structural inspection, engine performance testing, refueling, and avionics calibration. In contrast, a train maintenance window is an unavailable period for trains to perform routine maintenance such as brake system testing, wheelset inspection, seat cleaning, and electrical system verification. The establishment of these windows is intended to ensure the safety and reliability of the trains.

[0053] In one embodiment, if the aircraft maintenance window for a pair is from 8:00 AM to 11:00 AM and the train maintenance window is from 9:00 AM to 12:00 PM, then there is an overlapping interval from 9:00 AM to 11:00 AM, lasting two hours. If the preset synchronization coordination threshold is 1.5 hours, the pair is assigned to a semi-fixed binding candidate pair set because it exceeds the threshold.

[0054] Specifically, when the overlap duration does not exceed a threshold, for example, if another pair of aircraft maintenance windows runs from 2 PM to 3 PM and a train window runs from 2:30 PM to 4 PM, with an overlap duration of only half an hour, they are retained in the fixed-binding pair set. By classifying based on whether the overlap duration exceeds the threshold, it is possible to effectively distinguish between strictly bound fixed sets and semi-fixed candidate sets that allow for adjustment.

[0055] For example, this classification process is based on the intersection analysis of maintenance windows, ensuring that fixed bindings avoid timing mismatch issues caused by maintenance conflicts.

[0056] In one embodiment, the classification results directly support subsequent binding optimization decisions, improving transfer efficiency.

[0057] Example: Aircraft C (Flight MU789, Shanghai Hongqiao Airport - Guangzhou Baiyun Airport, departure 13:00 / landing 15:00, maintenance window 15:30-17:30), Train D (G1234, Guangzhou South Station - Shenzhen North Station, departure 14:30 / arrival 16:30, maintenance window 15:00-16:30); The overlapping maintenance window period is 15:30-16:30, lasting 60 minutes; the preset synchronization and coordination threshold is 90 minutes, which is not exceeded, so it is retained as a fixed binding pair; if the overlapping duration is 120 minutes (e.g., if the aircraft maintenance window is adjusted to 15:00-17:00), it will be transferred to a semi-fixed binding candidate pair.

[0058] Step S104: For the branch that is transferred to the semi-fixed binding candidate pair, obtain real-time passenger flow monitoring data and historical delay processing records, calculate the allowable time fluctuation adjustment window through the dynamic seat release mechanism and the backup train activation conditions, and determine the flexible matching range between aircraft and trains under semi-fixed binding.

[0059] Acquire real-time passenger flow monitoring data and historical delay handling records. Analyze the delay distribution and average delay duration for each pair of train pairings using historical delay handling records. Calculate the trigger threshold for activating backup trains based on the delay distribution and average delay duration. If the real-time passenger flow monitoring data exceeds the preset passenger flow threshold, activate the backup train. Employ a dynamic seat release mechanism, releasing seat resources for the corresponding time slots based on the activated backup trains and current real-time passenger flow monitoring data. Calculate the adjustable time range for aircraft and trains under semi-fixed pairing by using the released seat resources and the time slot adjustment window. Determine the flexible matching interval between aircraft and trains within the adjustable time range.

[0060] When calculating the elastic matching interval, the system uses a Kalman filter-based passenger flow prediction model to process real-time monitoring data and estimate the physical throughput capacity of the transfer channel at time T in the future (unit: people / minute). The state equation of this model is X(k) = A·X(k-1) + B·u(k-1) (A is the state transition matrix, with values ​​[[0.98, 0.02], [0.01, 0.99]]; B is the control matrix, with values ​​[[0.1], [0.05]]), and the observation equation is Z(k) = H·X(k) (H is the observation matrix, with values ​​[[1, 0], [0, 1]]). Both process noise and observation noise follow a normal distribution of N(0, 0.01). The dynamic seat release mechanism is not a simple ticketing operation, but rather automatically adjusts the opening threshold of the virtual inventory based on the physical congestion level of the transfer channel and the remaining dynamic load of subsequent trains. The calculation of the time-shift adjustment window takes into account the normal distribution characteristics of passenger walking speed and the mechanical delay of baggage transfer system, thereby determining the maximum elastic time difference interval between aircraft landing and train departure to ensure successful physical transfer.

[0061] Acquire real-time passenger flow monitoring data and historical delay handling records. This data is updated in real time through the integrated system to provide a basis for business decisions.

[0062] For example, the distribution of delays and the average delay duration of each binding candidate pair are statistically analyzed by recording historical delays.

[0063] Specifically, this statistical process first aggregates all historical delay events, then calculates the frequency distribution of delays by pairing them together, such as the proportion during peak and off-peak periods, and simultaneously calculates the average duration of each delay. For the candidate pairing of Chengdu Shuangliu Airport and Chengdu East Railway Station, historical records show that delays occur in the afternoon, accounting for 40%, with an average delay duration of 16 minutes. Based on the delay distribution and average delay duration, the trigger threshold for activating backup trains is calculated. This calculation considers the probability and impact of delays to determine the activation conditions. If real-time passenger flow monitoring data exceeds a preset passenger flow threshold, a backup train is activated. For example, when monitoring data shows a passenger flow of 520 people, exceeding the preset threshold of 400 people, the system automatically activates a backup train. A dynamic seat release mechanism is adopted, releasing seat resources at the corresponding time based on the activated backup train and the current real-time passenger flow monitoring data.

[0064] Specifically, the dynamic mechanism will release additional seats at corresponding train times based on passenger flow overload, such as releasing thirty seats to accommodate more connecting passengers. The adjustable time range between the aircraft and the train under semi-fixed binding is calculated by using the released seat resources and the time adjustment window.

[0065] like Figure 8As shown, each semi-fixed candidate pair corresponds to an adjustable time window interval, where the dark kernel represents the priority matching area and the outer interval represents the tolerable fluctuation range. The center reference line corresponds to the ideal alignment time. This figure reflects how, under conditions of delays and passenger flow fluctuations, this invention achieves flexible adjustment of binding relationships through spatiotemporal window management, balancing service continuity and transfer reliability.

[0066] In one possible implementation, combining the released seat capacity with a set ten-minute floating adjustment window yields an adjustable time range of eight minutes before and after. A flexible matching interval between the aircraft and train within this adjustable time range is determined. This interval allows for flexible adjustments by both parties under a semi-fixed binding to accommodate actual delays.

[0067] Example: Semi-fixed binding candidate pair (flight MU789 + train G1234), historical delay handling records show: the delay rate of this line is 35%, with an average delay time of 16 minutes; the preset passenger flow threshold is 480 people / hour, and the real-time monitored passenger flow is 520 people (exceeding the threshold), so the backup train G1235 is activated (departure at 15:00 / arrival at 17:00); the dynamic seat release mechanism releases 30 seats for G1235, with a time fluctuation adjustment window of ±10 minutes, and the final flexible matching range is determined to be 14:50-15:10.

[0068] Step S105: Integrate fixed binding pairs and semi-fixed binding pairs into the cross-network resource view, and generate a complete binding relationship set based on the automatic generation rules of baggage direct-check tags for fixed binding pairs and the requirements of backup shuttle vehicle dispatch channels for semi-fixed binding pairs.

[0069] Retrieve all fixed and semi-fixed binding pairs from the cross-network resource view. For fixed binding pairs, read the baggage check-through tag content. Automatically generate binding rules for the fixed binding pairs based on the baggage check-through tag content. For semi-fixed binding pairs, read the backup shuttle vehicle dispatch channel requirements. Generate binding constraints for the semi-fixed binding pairs based on the backup shuttle vehicle dispatch channel requirements. Merge the binding rules for fixed and semi-fixed binding pairs into a unified binding rule set. Perform matching checks on all binding pairs in the cross-network resource view using this unified binding rule set to determine the complete set of binding relationships.

[0070] like Figure 3As shown, fixed binding rules and semi-fixed elastic constraints converge into a complete set of binding relationships, driving subsequent inventory synchronization locking and disturbance simulation evaluation. The system makes continuity judgments based on the matching status of inventory release volume and capacity ratio. When the judgment meets the criteria, the current binding relationship is maintained and the result is output; when the judgment does not meet the criteria, it automatically enters the backtracking optimization branch to recalculate the semi-fixed constraints and dynamic release parameters, and participates in the fusion evaluation again. This structure unifies static rules and dynamic feedback in the same closed loop, achieving robust scheduling for delays and passenger flow fluctuations.

[0071] In one possible implementation, the cross-network resource view can be understood as a digital platform that integrates air and rail transportation systems, aggregating information on various binding pairs, such as the connection between air flights and high-speed trains.

[0072] Specifically, when the system needs to retrieve all fixed binding pairs, it extracts pre-defined permanent matches from the database, such as the fixed links between specific flights at Beijing Capital International Airport and high-speed rail services at Beijing South Railway Station. These binding pairs are unaffected by temporary changes. Semi-fixed binding pairs involve matches that allow for some flexibility, such as the connection between flights at Shanghai Hongqiao International Airport and trains at Hongqiao Railway Station, but which may require a backup plan due to delays. By querying the resource library of the view, the system can retrieve data for these binding pairs in real time, ensuring coverage of all network nodes.

[0073] For example, in actual business operations, the process of reading the contents of baggage transfer tags for fixed binding pairs can extract key information from the passenger's baggage tag, such as flight number, train number, and transfer code. Suppose a passenger flies from Guangzhou Baiyun Airport to Chengdu and then transfers to a high-speed train at Chengdu East Station. The system scans the QR code on the tag and reads the embedded binding identifier, including departure time, baggage number, and destination code, thereby confirming the fixed binding pair.

[0074] In one possible implementation, binding rules for fixed binding pairs are automatically generated based on the content of these baggage direct-attachment tags, which involves analyzing tag data to formulate a set of rules.

[0075] For example, the system will parse the timestamp and routing information in the tag. If the time interval between the flight's estimated arrival time and the train's departure time is less than 30 minutes, a rule will be generated to prioritize baggage transfer to avoid delays. At the same time, the rule may include baggage weight limits and security check standards to ensure that the binding is executed efficiently in a fixed mode.

[0076] For example, for semi-fixed binding pairs, when reading the backup shuttle vehicle dispatch channel requirements, the channel data of the dispatch system is accessed. These channels define the availability and conditions of backup vehicles, such as activating backup buses or chauffeured cars from the airport to the train station when the flight is delayed for more than 1 hour.

[0077] Specifically, in the scenarios of Beijing Daxing Airport and high-speed rail stations, the system reads channel requirements, including vehicle capacity, response time, and path optimization, to ensure that semi-fixed binding can cope with emergencies.

[0078] In one possible implementation, binding constraints for semi-fixed binding pairs are generated based on the requirements of these backup shuttle vehicle scheduling channels. The system creates constraints based on channel data, such as setting an activation threshold of a delay duration greater than 45 minutes, and constraining backup vehicles to meet environmental standards and passenger capacity limits, thereby forming a flexible but bounded binding framework.

[0079] For example, when merging the binding rules of fixed binding pairs with the binding constraints of semi-fixed binding pairs, the system uses a merging algorithm to integrate the two and obtain a unified set of binding rules.

[0080] Specifically, the rules and constraints are first classified and compared. For example, the time-fixed nature of fixed rules is combined with the flexible threshold of semi-fixed constraints to form a set, which includes shared verification standards, such as all binding pairs needing to verify the consistency of passenger identity.

[0081] In one possible implementation, this unified set of binding rules is used to perform matching checks on all binding pairs in the cross-network resource view. The system will check each binding pair one by one to see if it conforms to the rules. For example, for a binding pair from Shenzhen Airport to Guangzhou South Railway Station, it will check whether its time window, baggage handling and backup channel match, and finally determine the complete set of binding relationships to ensure the continuity and reliability of the entire transportation network.

[0082] Step S106: Extract the real-time inventory synchronization lock status of aircraft and train ticketing systems and the proportional matching status of capacity from the complete binding relationship set. Determine whether the service continuity meets the standard by the inventory linkage release and fine-tuning results under the disturbance simulation scenario.

[0083] Obtain the real-time inventory synchronization and locking status of the aircraft and train ticketing systems within the complete binding relationship set, as well as the proportional matching status of capacity. Input inventory change data through a disturbance simulation scenario to obtain the inventory release values ​​linked between aircraft and trains. The core parameter of this disturbance simulation scenario is: delay duration Δt follows a pattern of N(15, 5). 2The data is distributed normally over 60 minutes (range 0-60 minutes), and the capacity loss Δc follows a uniform distribution of U(0, 0.1) (i.e., seat unavailability ≤ 10%). The perturbation tensor is defined as [Δt, Δc]. The service continuity compliance criteria are: transfer failure rate ≤ 5%, and inventory release interruption duration ≤ 3 minutes. Based on the proportional matching status of inventory release values ​​and capacity capacity, a rule-based judgment is used. If the release exceeds a preset threshold, a fine-tuning adjustment is initiated. For the fine-tuning adjustment, the current lockout status data is obtained, and the adjusted inventory synchronization status is calculated through the proportional matching relationship. The difference between the adjusted inventory synchronization status and the original lockout status is determined to obtain the fine-tuning result. Based on the fine-tuning result and the inventory release values ​​under the perturbation simulation scenario, it is determined whether the service continuity meets the criteria. If the service continuity does not meet the criteria, the perturbation simulation scenario is returned, and the adjustment parameters are re-entered to obtain a new inventory release value.

[0084] The disturbance simulation scenario involves injecting delay durations into the system. and capacity loss The perturbation tensor is used to test the convergence of the binding relationship under non-ideal conditions. The criterion for judging whether service continuity is met is defined as follows: under the action of a given perturbation tensor, the system can keep the overall transfer failure rate below a set threshold (e.g., 5%) by adjusting the inventory lock status within a preset rescheduling time. This fine-tuning mechanism based on feedback loop ensures that when facing uncertain and sudden events such as flight delays, the system can sacrifice some local optima (e.g., fine-tuning the lock status of some seats) to exchange for the stability (i.e., robustness) of the global scheduling scheme.

[0085] For example, when retrieving the real-time inventory synchronization and lock status of aircraft and train ticketing systems within a complete set of binding relationships, a specific business scenario can be considered, such as a multimodal transport service in cooperation between an airline and a high-speed rail operator. The system first extracts real-time data from the binding relationship set. Assuming the aircraft is a flight from Beijing to Shanghai and the train is a high-speed rail line from Shanghai to Nanjing, the inventory synchronization and lock status refers to whether the reserved number of flight seats and high-speed rail tickets has been locked in a linked state. For example, if the flight has 100 seats reserved for connecting passengers, the corresponding high-speed rail seats should be matched and locked. The capacity matching status involves the ratio of the flight's passenger capacity to the high-speed rail's seat capacity, for example, ensuring a match at a ratio of 1:0.8. If the flight capacity is 200 people, then the high-speed rail needs to match a capacity of 160 people. This status is obtained through database queries to ensure real-time data updates and avoid overbooking.

[0086] like Figure 5As shown, the system compliance rate decreases with increasing delay disturbance magnitude, but the proposed solution maintains higher service continuity across all disturbance levels, with a particularly significant advantage in medium-to-high disturbance ranges. This result demonstrates that the combination of fixed and semi-fixed mechanisms with inventory-linked fine-tuning effectively suppresses disturbance propagation and reduces the probability of transfer link interruptions. This performance graph serves as quantitative evidence of the robustness advantage of this invention.

[0087] In one possible implementation, inventory change data is input into a perturbation simulation scenario to obtain the inventory release value between aircraft and trains, which can simulate inventory fluctuations caused by sudden events such as flight delays.

[0088] For example, input data includes a 10% seat release disturbance caused by a 2-hour flight delay. The system uses a simulation engine to calculate the linkage effect. The specific process is to first establish a disturbance model and use inventory changes as input parameters. For example, if there are 100 seats originally locked and 15 seats are released after the disturbance, the linkage release value is calculated, that is, the high-speed rail releases 12 seats accordingly (proportional to 0.8). This value reflects the dynamic response of cross-system inventory and helps predict potential resource waste.

[0089] For example, based on the proportional matching status between inventory release values ​​and capacity, a rule-based system determines whether to initiate a fine-tuning phase if the release exceeds a preset threshold. In actual operation, the preset threshold might be set at 20% of the release value; for instance, if the release exceeds 20 seats, an adjustment is triggered. The rule-based judgment process involves comparing the release value with the matching status. For example, if releasing 25 seats exceeds the threshold of 20, the system automatically switches to the fine-tuning phase. This ensures the stability of resource utilization and avoids service interruptions caused by over-release.

[0090] In one possible implementation, the current lockout status data is obtained during the fine-tuning process. The adjusted inventory synchronization status is calculated using a proportional matching relationship. For example, the system can read current data such as 80 remaining locked seats for flights, and then apply a proportional matching formula to adjust to 64 locked seats for high-speed rail. Specifically, the original lockout status is multiplied by an adjustment coefficient. For instance, a coefficient of 0.9 is used to fine-tune and reduce the number of locked seats, resulting in a new status of 72 flight seats and 57.6 high-speed rail seats (rounded down to 58). This process is implemented using an iterative algorithm to ensure that the adjusted status better reflects actual needs.

[0091] For example, determining the difference between the adjusted inventory synchronization status and the original locked status yields the result of fine-tuning. In a business example, if the original status was 100 locked flight seats and the adjusted status is 90, the difference is a reduction of 10 seats. This result quantifies the impact of the adjustment and is used for subsequent evaluation. For instance, in intermodal services, this difference helps operators identify additional resources that need to be allocated, such as adding spare flight seats to compensate.

[0092] In one possible implementation, service continuity is determined by comparing the results of fine-tuning with the inventory release values ​​under simulated disturbance scenarios. A continuity standard could be set, such as the release value not exceeding 15% of the adjustment result. For example, if the adjustment results in a reduction of 10 units, a release value of 12 units is within the acceptable range; otherwise, it is not. This judgment process involves comparing two values ​​to ensure the continuity of the overall service chain. From a business perspective, this can improve passenger satisfaction by reducing travel disruptions caused by inventory fluctuations.

[0093] For example, if service continuity is not met, the system returns to the disturbance simulation scenario and re-enters adjustment parameters to obtain new inventory release values. In actual operation, if the system determines that the target is not met, it will loop back to the simulation scenario, input new parameters (such as reducing the disturbance amplitude by 5%), and recalculate the release values. For example, if the original release was 15 units, the new parameters would reduce it to 10 units, thus iterating and optimizing until the target is met. This looping mechanism enhances the robustness of the system.

[0094] Step S107: If the ticket inventory release is interrupted or the capacity adjustment exceeds the elastic matching range in the disturbance simulation scenario, the process will backtrack to the semi-fixed binding candidate pair processing branch, incorporate the latest real-time passenger flow monitoring data and historical delay processing records to recalculate the dynamic seat release mechanism and time floating adjustment window, and obtain the final cross-network fixed and semi-fixed fusion collaborative binding scheme.

[0095] Obtain the ticket inventory release status under the disturbance simulation scenario. Determine whether the ticket inventory release is interrupted or whether the capacity adjustment exceeds the elastic matching range. If the ticket inventory release is interrupted or the capacity adjustment exceeds the elastic matching range, obtain the candidate pair set from the semi-fixed binding candidate pair processing branch. Obtain the current passenger flow distribution data from the real-time passenger flow monitoring data interface and the corresponding route delay duration sequence from the historical delay processing record database. Calculate the dynamic seat release quantity based on the passenger flow distribution data and the delay duration sequence to obtain the current set of releaseable seats. For the set of releaseable seats and the semi-fixed binding candidate pair set, determine the binding time range of each candidate pair through the time floating adjustment window to obtain the adjusted binding time set. Based on the adjusted binding time set and the cross-network fixed binding rules, determine the final collaborative binding scheme.

[0096] The collaborative binding scheme is a structured data table with the following field definitions: ① Flight number (e.g., CA123); ② Train number (e.g., G456); ③ Binding type (fixed / semi-fixed); ④ Flexible time window (for semi-fixed binding only, format: "[start time, end time]"); ⑤ Number of seats that can be released (dynamically adjusted value); ⑥ Transfer channel identifier (e.g., T3 - High-speed rail channel 1); ⑦ Baggage direct check-in identifier (yes / no, for fixed binding only); ⑧ Backup connection activation conditions (for semi-fixed binding only, e.g., "delay ≥ 30 minutes"); ⑨ Effective time period (e.g., "2024-01-01 to 2024-12-31").

[0097] like Figure 6 As shown, with the increase of backtracking iterations, both the inventory release amount and the locking deviation after fine-tuning gradually decrease and tend to stabilize, indicating that the system can complete disturbance absorption and parameter convergence within a limited number of iterations. This figure intuitively demonstrates the stabilization capability of the feedback loop under abnormal scenarios, providing graphical support for the backtrackable, convergent, and sustainable dynamic scheduling characteristics of this invention.

[0098] like Figure 9 As shown, based on the delay disturbance intensity and passenger flow load index, the system can be divided into a fixed binding priority zone, a fixed semi-fixed coordination zone, and a semi-fixed backtracking optimization zone. Different zones correspond to different scheduling strategy switching logics, guiding the system to quickly select the optimal control action when the operating state changes. This figure illustrates that the invention has clear state partitioning and strategy boundary definitions, which can improve the stability and interpretability of online decision-making.

[0099] For example, in the inventory management business of aircraft and train ticketing systems, the ticket inventory linkage release status under the disturbance simulation scenario is first obtained.

[0100] Specifically, this state describes the real-time progress of inventory release after a simulated disturbance input. For example, when flight delay data is input, it observes whether train seats are released accordingly. It determines whether the coordinated release of ticket inventory is interrupted or whether capacity adjustments exceed the elastic matching range. If the coordinated release of ticket inventory is interrupted or capacity adjustments exceed the elastic matching range, a candidate pair set is obtained from the semi-fixed binding candidate pair processing branch.

[0101] Specifically, semi-fixed binding candidate pairs are pre-screened potential collaborative pairings, such as combinations of specific flights and adjacent trains. Current passenger flow distribution data is obtained from the real-time passenger flow monitoring data interface, and the corresponding line delay duration sequence is obtained from the historical delay processing record database. Based on the passenger flow distribution data and the delay duration sequence, the dynamic number of seats to be released is calculated, resulting in the current set of seats that can be released.

[0102] For example, passenger flow distribution data shows that passenger volume at a certain hub station increases by 30% during peak hours, while delay sequence data shows an average delay of 18 minutes. By comprehensively analyzing these factors, the number of seats to be dynamically released is calculated to be 40, forming a set of releaseable seats to support subsequent binding adjustments. For the set of releaseable seats and the set of semi-fixed binding candidate pairs, the binding time range of each candidate pair is determined through a time-floating adjustment window, resulting in the adjusted binding time set.

[0103] Specifically, the time-floating adjustment window is a flexible time mechanism that allows candidate binding times to float within a certain window to optimize matching. The final collaborative binding scheme is determined by matching the adjusted set of binding times with the cross-network fixed binding rules.

[0104] Based on actual testing (test scenario: Beijing-Shanghai-Nanjing intermodal route, sample size 1000 flights + corresponding high-speed rail services, covering weekday peak hours, holidays, inclement weather, etc.), after the implementation of this solution: ① Transfer success rate increased from 75% to 92%; ② Service continuity compliance rate under delay scenarios (Δt≥30 minutes) increased from 68% to 89%; ③ Cross-network capacity utilization rate increased by 35%; ④ Average passenger transfer waiting time decreased from 45 minutes to 27 minutes; ⑤ Ticket overbooking / empty-running rate decreased by 28%.

[0105] Supplementary Explanation To ensure that the technical solution of this invention can be fully understood and implemented by those skilled in the art, detailed supplementary explanations are provided for the key algorithms, models, thresholds, etc. involved in the patent application documents, as follows: 1. Dynamic seat release mechanism The dynamic seat release mechanism is an intelligent ticketing management strategy based on real-time passenger flow data and capacity status. The core of this mechanism lies in automatically adjusting the threshold for opening virtual ticket inventory based on passenger load, transfer channel congestion, and remaining capacity of subsequent trains, thereby achieving dynamic optimization of seat resource allocation. The specific operation process is as follows: When the system detects that real-time passenger flow exceeds a preset threshold (usually 80% of the channel's design capacity), it automatically initiates a dynamic seat release process. First, the system assesses the congestion level of the current transfer channel. If the channel occupancy rate exceeds 90% of its design capacity, priority is given to releasing seats for trains departing within the next hour. If trains departing within the next hour are already full, seats for trains departing 1-3 hours later are released sequentially. Finally, if all of the above trains are full, seats for trains departing more than 3 hours later are considered for release. This mechanism ensures timely release of seat resources during peak passenger flow periods, preventing passengers from being unable to complete their connecting journeys due to insufficient seats, while maximizing the utilization of transport capacity.

[0106] 2. Adjust the window based on time fluctuations The timetable adjustment window is used to determine the acceptable range of timetable adjustments for aircraft and trains under semi-fixed coupling. This window is calculated based on a comprehensive consideration of passenger walking speed, transfer corridor length, and mechanical delays in the baggage transfer system. Specifically: Passenger walking speed is estimated using a normal distribution model, with a mean of 1.2 m / s and a variance of 0.3 m / s. This is based on on-site observations and statistical analysis of a large number of passengers transferring between airports and high-speed rail stations. Transfer corridor length is typically no more than 800 meters, designed based on the actual distances between major domestic hub airports and high-speed rail stations. The mechanical delay of the baggage transfer system is no more than 5 minutes, based on the actual operating efficiency of the airport baggage handling system.

[0107] Based on the above factors, the system's default timetable adjustment window is ±10 minutes, with a maximum of ±15 minutes. The adjustment logic for this window is as follows: when the congestion level of the transfer passage is low, the system can appropriately expand the floating window; when the passage is severely congested, the system will shrink the floating window to ensure that passengers can complete the transfer smoothly.

[0108] 3. Conditions for Activating Backup Train Services The activation criteria for backup train services are the standards by which the system determines to activate backup capacity under the dual pressure of passenger flow and delays. These criteria include two key elements: First, real-time passenger flow must reach a preset threshold, which is more than 80% of the channel's designed capacity. This threshold is based on historical passenger flow data to ensure that backup capacity can be activated in a timely manner during peak periods, preventing passengers from being unable to complete intermodal transport due to insufficient seating.

[0109] Secondly, the historical delay rate must reach over 30%. This threshold is based on statistical analysis of delays on the route over the past year, ensuring timely response to sudden passenger flow pressures during periods of high delay rates.

[0110] The system will only activate the backup train when both conditions are met. Once activated, the system will automatically release the corresponding seat resources of the backup train based on real-time passenger flow data to ensure the continuity of intermodal services.

[0111] 4. Kalman Filter Passenger Flow Forecasting Model The Kalman filter-based passenger flow prediction model is a predictive algorithm used to estimate the future physical throughput capacity of transfer corridors. This model accurately predicts passenger flow in transfer corridors by integrating historical passenger flow data, real-time monitoring data, and external factors such as weather.

[0112] The model's state equation is based on historical passenger flow data, taking into account the continuity and changing trends of passenger flow. The control matrix reflects the degree of influence of external factors (such as weather and holidays) on passenger flow. The observation equation is used to compare the prediction results with actual monitoring data to continuously optimize the prediction accuracy.

[0113] The parameter settings of this model are based on the analysis and verification of a large amount of actual operating data. The state transition matrix is ​​set to [[0.98, 0.02], [0.01, 0.99]], representing the smoothness and inertia of passenger flow changes; the control matrix is ​​set to [[0.1], [0.05]], representing the degree of influence of external factors on passenger flow; process noise and observation noise are both set to 0.01 to ensure the stability and reliability of the prediction results.

[0114] 5. Non-dominated sorting genetic algorithm (NSGA-III) The Non-Dominated Sorting Genetic Algorithm (NSGA-III) is a multi-objective optimization algorithm used to solve multi-constraint optimization problems in air-rail intermodal transport scheduling. In this invention, this algorithm is used to determine the optimal set of binding candidate pairs, and its objective function includes three key indicators: 1. Minimize average passenger transfer time: Reduce passenger waiting time during transfers by optimizing time alignment and transfer lane usage.

[0115] 2. Maximize cross-network capacity utilization: Improve overall resource utilization efficiency by rationally matching the capacity of aircraft and trains.

[0116] 3. Minimize transfer failure rate: Reduce the probability of intermodal transport failure by strictly controlling maintenance window conflicts and transfer channel capacity.

[0117] The constraints of this algorithm include: (1) Time alignment error not exceeding 5 minutes: Ensure the time matching accuracy between aircraft and train.

[0118] (2) Transfer time shall be no less than 20 minutes: Ensure that passengers have enough time to complete the transfer.

[0119] (3) The maintenance window overlap time shall not exceed 90 minutes: to avoid conflicts between aircraft and trains during maintenance windows.

[0120] (4) The capacity of the transfer channel shall not exceed 90% of the design capacity: ensure that the transfer channel will not affect passenger passage due to overloading.

[0121] The algorithm parameters were verified through extensive experiments. The population size was set to 100, the number of iterations was set to 50, the crossover probability was set to 0.8, and the mutation probability was set to 0.1. These parameter settings ensured that the algorithm could find a high-quality optimized solution within a reasonable time.

[0122] 6. Parameters of the disturbance simulation scenario The disturbance simulation scenario is a crucial step in testing the robustness of the system, and its core parameter settings are as follows: (1) Delay duration Δt: follows a normal distribution with a mean of 15 minutes and a standard deviation of 5 minutes (range 0-60 minutes). This parameter setting is based on statistical analysis of historical flight delay data to ensure that the disturbance simulation can truly reflect the delay situation in actual operation.

[0123] (2) Capacity loss Δc: follows a uniform distribution in [0,0.1], indicating that the seat unavailability rate does not exceed 10%. This parameter reflects the seat unavailability that may occur in actual operation, such as temporary ticket refunds by passengers, system failures, etc.

[0124] The perturbation tensor [Δt, Δc] is used to simulate various possible perturbation scenarios, helping the system test its performance under different perturbation conditions. The service continuity compliance criteria are: a transfer failure rate not exceeding 5%, and an interruption duration of inventory release not exceeding 3 minutes.

[0125] 7. Maintain the synchronization and coordination threshold for window overlap duration. The synchronization and coordination threshold for overlapping maintenance windows is 90 minutes. This threshold is based on statistical analysis of historical maintenance conflict data. Analysis of maintenance window conflicts between aircraft and trains at shared stations over the past year revealed that when the overlap exceeds 90 minutes, the impact of maintenance conflicts on intermodal services increases significantly, potentially leading to service disruptions or severe delays. Therefore, 90 minutes was determined to be a reasonable synchronization and coordination threshold.

[0126] 8. Minimum transfer standard The minimum transfer time is set at 20 minutes, a standard determined based on on-site observations of passenger transfer processes and passenger satisfaction surveys. Observations at several major hub airport-high-speed rail stations revealed that the average passenger walking speed is 1.2 meters per second, and the average baggage transfer time is 5 minutes. Including necessary security checks and waiting time, a 20-minute transfer time ensures that most passengers can complete their transfers smoothly while avoiding excessively long waiting times that could negatively impact the passenger experience.

[0127] 9. The reasonable range of transport capacity ratio The pre-set reasonable range for the capacity ratio is 0.3 to 0.7, a range derived from statistical analysis of historical intermodal transport data. Analysis of the capacity matching of various air-rail intermodal transport combinations over the past year revealed that when the ratio of aircraft capacity to train capacity is within the range of 0.3-0.7, the efficiency of intermodal services and passenger satisfaction reach an optimal balance. A ratio that is too small leads to wasted train capacity, while a ratio that is too large may result in underutilization of aircraft seats, impacting the economic benefits of intermodal services.

[0128] The above supplementary explanation details the specific implementation methods, determination basis, and reasonable scope of the key algorithms, models, and thresholds involved in this invention, ensuring that those skilled in the art can implement the invention based on the description in the specification, thus satisfying the requirements of Article 26, Paragraph 3 of the Patent Law.

[0129] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A robust scheduling and dynamic supply-demand matching method for air-rail intermodal transport, characterized in that, include: The aircraft timetable, train timetable, aircraft capacity, train capacity and their respective maintenance window information are obtained from the preset database. An initial set of fixed binding candidate pairs is obtained by comparing the time overlap interval with the capacity ratio. For the initial set of fixed binding candidate pairs, strict time alignment calculation is performed to eliminate candidate pairs where the reserved time between the aircraft and the train in the shared physical station transfer channel is less than the minimum transfer standard, so as to obtain the initial fixed binding pairs that meet the strict one-to-one correspondence of fixed binding time. Based on the initial fixed binding pair, it is determined whether there is an overlap or conflict between the aircraft maintenance window and the train maintenance window. If the overlap duration of the maintenance windows exceeds the preset synchronization and coordination threshold, the binding pair is transferred to the semi-fixed binding processing branch, resulting in a classification set of fixed binding pair set and semi-fixed binding candidate pair set. For branches that are transferred to semi-fixed binding candidate pairs, real-time passenger flow monitoring data and historical delay handling records are obtained. The allowable time fluctuation adjustment window is calculated through dynamic seat release mechanism and standby train activation conditions to determine the flexible matching range between aircraft and trains under semi-fixed binding. Fixed binding pairs and semi-fixed binding pairs are uniformly incorporated into the cross-network resource view, and a complete binding relationship set is generated based on the automatic generation rules of baggage direct-check tags for fixed binding pairs and the requirements of the backup shuttle vehicle dispatch channel for semi-fixed binding pairs. Extract the real-time inventory synchronization and locking status of aircraft and train ticketing systems and the proportional matching status of capacity from the complete set of binding relationships. Determine whether the service continuity meets the standard by analyzing the inventory linkage release and fine-tuning results under the disturbance simulation scenario.

2. The method according to claim 1, characterized in that, Further includes: If the ticket inventory release is interrupted or the capacity adjustment exceeds the elastic matching range in the disturbance simulation scenario, the process will backtrack to the semi-fixed binding candidate pair processing branch. The latest real-time passenger flow monitoring data and historical delay processing records will be incorporated to recalculate the dynamic seat release mechanism and time slot adjustment window, resulting in the final cross-network fixed and semi-fixed collaborative binding scheme.

3. The method according to claim 2, characterized in that, The method is executed by a multi-agent system, which includes: an aviation resource agent agent for collecting aircraft operational status data and executing aviation-side resource adjustment commands; a railway resource agent agent for collecting train operational status data and executing railway-side resource adjustment commands; and a cross-network collaborative scheduling agent for receiving the aircraft operational status data and train operational status data, generating cross-network scheduling decisions, and issuing scheduling commands to the aviation resource agent agent and the railway resource agent agent; wherein the aviation resource agent agent and the railway resource agent agent interact asynchronously with the cross-network collaborative scheduling agent through a standardized message protocol.

4. The method according to claim 3, characterized in that, The process involves retrieving aircraft timetables, train timetables, aircraft capacity, train capacity, and their respective maintenance window information from a preset database. An initial set of fixed-binding candidate pairs is obtained by comparing the time overlap intervals with the capacity ratio, including: By comparing the overlapping intervals of aircraft and trains on the timeline with timetable data, a set of timetable overlap intervals is obtained. For each overlapping time interval, obtain the aircraft capacity and train capacity within the corresponding time period; The capacity ratio is obtained by calculating the ratio of aircraft capacity to train capacity within the time overlap interval. If the capacity ratio is within a preset reasonable range, the aircraft and train combination corresponding to the overlapping interval at that time will be retained. Based on the maintenance window information, determine whether there is a maintenance conflict between the aircraft and the train within the time overlap interval. If a maintenance conflict exists, the combination is removed. By combining the retained aircraft and trains, an initial set of fixed-binding candidate pairs is obtained.

5. The method according to claim 3, characterized in that, The process involves acquiring real-time passenger flow monitoring data and historical delay handling records for branches transitioning to semi-fixed binding candidate pairs. It then calculates the allowable time fluctuation adjustment window using a dynamic seat release mechanism and standby train activation conditions to determine the flexible matching range between aircraft and trains under semi-fixed binding, including: The distribution of delays and the average delay duration of each binding candidate pair are statistically analyzed by recording historical delays. The trigger threshold for activating standby trains is calculated based on the distribution of delays and the average delay duration. If the real-time passenger flow monitoring data exceeds the preset passenger flow threshold, the backup train will be activated. A dynamic seat release mechanism is adopted, which releases seat resources at the corresponding time based on the activated spare trains and real-time passenger flow monitoring data. By utilizing the released seat resources and the time slot adjustment window, the adjustable time range for aircraft and trains under semi-fixed binding is calculated. Determine the flexible matching range between aircraft and trains within an adjustable time frame.

6. The method according to claim 3, characterized in that, The initial fixed-binding candidate pair set is calculated using strict time alignment. Candidate pairs with reserved time between aircraft and trains at shared physical station transfer channels that is less than the minimum transfer standard are eliminated, resulting in preliminary fixed-binding pairs that satisfy strict one-to-one correspondence of fixed-binding times, including: Extract aircraft time, train time, physical station information and transfer channel data from the initial set of fixed binding candidate pairs; Compare the time differences between the arrival and departure times of aircraft and the arrival and departure times of trains at the physical stations one by one; If the time difference is less than the preset minimum standard threshold, the candidate pair is marked as not meeting the condition; Filter out all candidate pairs marked as not meeting the conditions, and retain candidate pairs that meet the time alignment and time allowance requirements; For the retained candidate pairs, determine the usage status of the transfer channels. If there are conflicts or the channels are unavailable, remove the relevant candidate pairs to obtain the optimized set of fixed binding pairs.

7. The method according to claim 3, characterized in that, The process involves determining whether there is overlap or conflict between the aircraft maintenance window and the train maintenance window based on the initial fixed binding pair. If the overlap duration exceeds a preset synchronization coordination threshold, the binding pair is transferred to the semi-fixed binding processing branch, resulting in a classification set of fixed binding pair set and semi-fixed binding candidate pair set, including: For each initial fixed binding pair, obtain the start and end times of the aircraft maintenance window and the train maintenance window; Determine if there is an overlap between the start and end times of the aircraft maintenance window and the train maintenance window to obtain the overlapping interval; Calculate the duration corresponding to the overlapping intervals to obtain the overlapping duration; Determine whether the overlap duration exceeds the preset synchronization coordination threshold. If it does, then divide the initial fixed binding pair into the semi-fixed binding candidate pair set. If the overlap duration does not exceed the synchronization coordination threshold, the initial fixed binding pair will be retained in the fixed binding pair set. Through the above partitioning operation, the classification results of the fixed binding pair set and the semi-fixed binding candidate pair set are obtained.

8. The method according to claim 3, characterized in that, The process of unifying fixed and semi-fixed binding pairs into a cross-network resource view, and generating a complete set of binding relationships based on the automatic generation rules of baggage direct-check tags for fixed binding pairs and the requirements of backup shuttle vehicle dispatch channels for semi-fixed binding pairs, includes: This is for fixed-binding methods to read the contents of baggage direct-attachment tags; Binding rules for fixed binding pairs are automatically generated based on the contents of the baggage direct-check tag; This addresses the requirements for reading the backup shuttle vehicle dispatch channel in semi-fixed binding; Binding constraints for semi-fixed binding pairs are generated based on the requirements of the backup shuttle vehicle dispatch channel; The binding rules of fixed binding pairs and the binding constraints of semi-fixed binding pairs are merged into a unified set of binding rules. By performing matching and validation on all binding pairs in the cross-network resource view using a unified set of binding rules, a complete set of binding relationships is determined.

9. The method according to claim 3, characterized in that, The process of extracting the real-time inventory synchronization and locking status of aircraft and train ticketing systems from the complete binding relationship set, as well as the proportional matching status of capacity, and judging whether service continuity meets the standards through inventory linkage release and fine-tuning results under disturbance simulation scenarios includes: By inputting inventory change data into a disturbance simulation scenario, the inventory release value between aircraft and trains is obtained. Based on the proportional matching status between inventory release values ​​and transportation capacity, rules are used to determine if the release exceeds a preset threshold, and then a fine-tuning process is initiated. For the fine-tuning process, the current locked status data is obtained, and the adjusted inventory synchronization status is calculated through proportional matching relationships. Determine the difference between the adjusted inventory synchronization status and the original locked status to obtain the fine-tuning results; Based on the results of the fine-tuning and the inventory release values ​​under the disturbance simulation scenario, it is determined whether the service continuity meets the standards.

10. The method according to claim 3, characterized in that, If the ticket inventory release is interrupted or the capacity adjustment exceeds the elastic matching range in the disturbance simulation scenario, the process will backtrack to the semi-fixed binding candidate pair processing branch. The latest real-time passenger flow monitoring data and historical delay processing records will be incorporated to recalculate the dynamic seat release mechanism and time slot adjustment window, resulting in a final cross-network fixed and semi-fixed fusion collaborative binding scheme, including: Determine whether the coordinated release of ticketing inventory has been interrupted or whether the fine-tuning of transportation capacity has exceeded the elastic matching range; If the release of ticket inventory is interrupted or the capacity adjustment exceeds the elastic matching range, the candidate pair set is obtained from the semi-fixed binding candidate pair processing branch. Obtain current passenger flow distribution data from the real-time passenger flow monitoring data interface, and obtain the corresponding route delay duration sequence from the historical delay processing record database; The dynamic number of seats to be released is calculated based on passenger flow distribution data and delay time sequence to obtain the current set of seats that can be released. For the set of releasable seats and the set of semi-fixed binding candidate pairs, the binding time range of each candidate pair is determined by the time floating adjustment window, and the adjusted binding time set is obtained. The final collaborative binding scheme is determined by matching the adjusted binding time set with the cross-network fixed binding rules.

Images