A dispatching method and system for matching airline baggage change information

Abstract

The application discloses a kind of scheduling method and system of matching aviation luggage change information, it is related to aviation transport intelligent scheduling technical field, including, real-time acquisition luggage change information in aviation transport scene, passenger itinerary data and airport equipment state data, output original data stream;In the metaverse, the airport operation digital twin is constructed, and the collaborative transfer path is imported into the digital twin environment for space-time pre-play simulation, real-time detection of luggage flow conflict points and marking of physical constraint conflict region coordinate set;By inverse reinforcement learning algorithm, the conflict events in the pre-play simulation are solved, the strategy defect features are traced, and the reward function space is reconstructed, and the gradient compensation parameters are output;Fusion physical constraint conflict region coordinate set and gradient compensation parameters, generate three-dimensional space scheduling instruction set and execute instruction issuing.The application is based on space-time pre-play simulation method, accurately identifies the safety boundary of luggage transfer path, and effectively reduces the risk of physical constraint overrun.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent scheduling of air transportation, and particularly relates to a scheduling method and system for supporting air baggage change information. BACKGROUND

[0002] The development of the air baggage management system has evolved from basic automation to intelligence, in which the graph neural network technology is applied to analyze passenger itinerary data and construct a social relationship model to support baggage scheduling decisions. At the same time, the integration of digital twin technology in airport operations is increasingly mature, through the loading of building information model and Internet of Things data flow, dynamic simulation and conflict detection of the physical environment are realized. The introduction of inverse reinforcement learning algorithm further strengthens the strategy optimization capability, through the inverse solution of state-action pair sample set, the accuracy of reward function reconstruction is improved.

[0003] However, the existing technology still has core defects in processing baggage change information, that is, it lacks efficient group baggage priority dynamic weighting mechanism and real-time conflict simulation optimization capability. This leads to the inability to accurately integrate passenger itinerary correlation in the baggage transfer process, making it difficult to accurately predict physical constraint conflict points through time and space pre-rehearsal, thereby causing path intersection collision probability to rise, equipment response delay to lose control, and sorting carousel volume overload rate to be abnormal, etc. problems, significantly increasing the risk of baggage delay and low resource utilization. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a scheduling method for supporting air baggage change information to solve the problems of insufficient group baggage priority dynamic weighting and real-time conflict simulation optimization in air baggage change scheduling.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a scheduling method for matching airline baggage change information, which comprises: collecting baggage change information, passenger itinerary data and airport equipment state data in an air transportation scene in real time, and outputting an original data stream; inputting the original data stream into a pre-trained graph neural network, constructing a social relationship topology graph by analyzing the co-occurring flights, seat proximity and ticket purchase time interval characteristics in the passenger seat reservation record; triggering a group baggage priority weighting mechanism according to the social relationship topology graph, and generating a collaborative transfer path for the baggage of passengers with itinerary association; constructing an airport operation digital twin in the metaverse, importing the collaborative transfer path into the digital twin environment for time-space pre-play simulation, and detecting and marking the conflict point of the baggage flow and the coordinate set of the physical constraint conflict area in real time; solving the conflict events in the pre-play simulation by inverse reinforcement learning algorithm, tracing the strategy defect characteristics and reconstructing the reward function space, and outputting the gradient compensation parameters; and fusing the coordinate set of the physical constraint conflict area and the gradient compensation parameters to generate a three-dimensional space scheduling instruction set and execute the instruction issuing.

[0008] As a preferred scheme of the scheduling method for matching airline baggage change information, the passenger itinerary data comprises passenger seat reservation records.

[0009] As a preferred scheme of the scheduling method for matching airline baggage change information, the constructing of the social relationship topology graph comprises the following steps:

[0010] Analyzing the passenger seat reservation records in the original data stream, and extracting the co-occurring flight numbers, adjacent seat numbers and ticket purchase time difference values;

[0011] Applying a negative exponential decay function to the ticket purchase time difference value to generate a time decay coefficient, and performing a Euclidean distance inverse calculation on the adjacent seat numbers to generate a spatial proximity;

[0012] Taking the co-occurring flight numbers as the initial weight of the graph edges, generating a dynamic social weight matrix by fusing the time decay coefficient and the spatial proximity, aggregating the passenger node features through a graph convolution network, and outputting the social relationship topology graph.

[0013] As a preferred scheme of the scheduling method for matching airline baggage change information, the generating of the collaborative transfer path for the baggage of passengers with itinerary association comprises the following steps:

[0014] When the dynamic of any passenger node associated in the social relationship topology graph changes, traversing all the baggage transfer states connected with the changed node to obtain the loading state identification code and the current position coordinates;

[0015] If the baggage loading state identification code is not loaded, generating a flight binding instruction to forcibly synchronize the current baggage to the target flight after the change;

[0016] If the luggage loading state identification code is loaded, the shortest transfer time of the current position coordinates of the corresponding luggage to the target boarding gate is calculated, and a virtual synchronization time window constraint is generated;

[0017] According to the dynamic weight value of the connection edge in the social relationship topology graph, the transfer order adjustment coefficient is assigned to the passenger node constrained by the virtual synchronization time window, and the collaborative transfer path is output.

[0018] As a preferred scheme of the scheduling method of the matching aviation luggage change information, the real-time luggage flow conflict point is detected and the physical constraint conflict region coordinate set is marked, and the specific steps are as follows,

[0019] In the meta-universe environment, the airport building information model and the real-time device Internet of Things data stream are loaded, and a dynamic digital twin is constructed;

[0020] The collaborative transfer path is imported into the dynamic digital twin, and the luggage transfer space-time simulation is performed, the sorting disc volume overload rate, the path intersection collision probability and the device response delay time length are calculated;

[0021] The physical engine is activated to simulate the luggage stacking and dumping effect and the conveyor belt vibration offset, and the physical constraint conflict region coordinate set exceeding the safety threshold is marked.

[0022] As a preferred scheme of the scheduling method of the matching aviation luggage change information, the gradient compensation parameter is output, and the specific steps are as follows,

[0023] The physical constraint conflict region coordinate set and the digital twin environment parameter are combined to generate a state-action pair sample set, which is input data of the inverse reinforcement learning algorithm;

[0024] The probability distribution of the state-action pair sample set is calculated by the maximum entropy probability model, and the reward function corresponding to the optimal strategy is deduced;

[0025] The weight distribution defect of the reward function is analyzed, the device response delay weight and the abnormal path intersection penalty coefficient causing the conflict are identified, and the reward function space structure is reconstructed;

[0026] The device fault simulation noise and the path random offset are injected into the state-action pair sample set to generate an adversarial perturbation sample, and the strategy network model is trained to output the gradient compensation parameter.

[0027] As a preferred scheme of the scheduling method of the matching aviation luggage change information, the three-dimensional space scheduling instruction set is generated and the instruction is issued, and the specific steps are as follows,

[0028] According to the gradient compensation parameter, the start time offset value of the sorting device motor is calculated, and the motor control instruction is generated;

[0029] Based on the physical constraint conflict region coordinate set, the A* algorithm is used to re-plan the ground staff travel path, and a path navigation point coordinate sequence is output.

[0030] The motor control instruction and the path navigation point coordinate sequence are aligned according to the timestamp, a structured instruction data packet is generated, the motor control instruction is sent to the baggage sorting controller through the message queue, and the path navigation point coordinate sequence is transmitted to the ground staff handheld terminal through the wireless protocol.

[0031] In a second aspect, the present application provides a scheduling system for matching aviation baggage change information, comprising a data acquisition module, a graph construction module, a path generation module, a simulation rehearsal module, a strategy optimization module and an instruction scheduling module; the data acquisition module is used to acquire baggage change information, passenger itinerary data and airport equipment state data in real time in an aviation transportation scene, and output an original data stream; the graph construction module is used to input the original data stream into a pre-trained graph neural network, and construct a social relationship topology graph by analyzing the co-occurring flights, seat proximity and ticket purchase time interval features in the passenger seat reservation record; the path generation module is used to trigger a group baggage priority weighting mechanism according to the social relationship topology graph, and generate a collaborative transfer path for the passenger baggage with itinerary association; the simulation rehearsal module is used to construct an airport operation digital twin in the metaverse, import the collaborative transfer path into the digital twin environment for spatio-temporal pre-rehearsal simulation, and detect and mark a physical constraint conflict region coordinate set in real time; the strategy optimization module is used to inversely solve the conflict events in the pre-rehearsal simulation through a reinforcement learning algorithm, trace the strategy defect features and reconstruct the reward function space, and output gradient compensation parameters; and the instruction scheduling module is used to fuse the physical constraint conflict region coordinate set and the gradient compensation parameters, generate a three-dimensional space scheduling instruction set and execute the instruction issuing.

[0032] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, any step of the scheduling method for matching aviation baggage change information according to the first aspect of the present application is implemented.

[0033] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, any step of the scheduling method for matching aviation baggage change information according to the first aspect of the present application is implemented.

[0034] The present application has the beneficial effects that: the present application constructs an airport operation digital twin in a metaverse environment, accurately simulates the luggage stacking and dumping effect and the conveyor belt vibration offset by using a physics engine, dynamically calculates the sorting disc volume overload rate, path intersection collision probability and equipment response delay time in real time, and generates a high-precision three-dimensional conflict area coordinate set; based on a time-space pre-performance simulation method, accurately identifies the safety boundary of the luggage transfer path, and effectively reduces the risk of physical constraint overrun. By integrating real-time device Internet of Things data streams, the spatio-temporal consistency of conflict detection is significantly improved, reliable data support is provided for gradient compensation parameter generation, thereby enhancing the robustness and execution efficiency of the scheduling instruction. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0036] Fig. 1 The flowchart of the scheduling method matched with the airline baggage change information.

[0037] Fig. 2 The flowchart of the social relationship topology graph construction.

[0038] Fig. 3 The flowchart of the digital twin conflict detection.

[0039] Fig. 4 The flowchart of the three-dimensional scheduling instruction generation. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0041] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0042] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0043] REFERENCE Figs. 1-4For an embodiment of the present application, the embodiment provides a scheduling method for matching airline baggage change information, comprising the following steps:

[0044] S1: Real-time collection of baggage change information, passenger itinerary data and airport equipment state data in the aviation transportation scene, outputting raw data stream.

[0045] Among them, the raw data stream refers to the raw data stream with timestamp and device association label generated by spatiotemporal alignment and semantic annotation of baggage change information, passenger itinerary data and airport equipment state data;

[0046] S1.1: The baggage change information includes baggage state change events (such as the change of baggage from "unloaded" state to "loaded" state) and current position coordinates of baggage. During the collection process, the event log or sensor data stream is directly read to ensure the real-time nature of the data.

[0047] The passenger itinerary data includes passenger seat reservation records, which contain specific attributes: flight number, seat number and ticket purchase time. When collecting, directly query or subscribe to the update event of the seat reservation device.

[0048] The airport equipment state data includes the current running state of the equipment (such as "normal" or "fault") and the physical location information (such as the device coordinates). When collecting, directly pull the device state report or event push.

[0049] S1.2: Add a uniform timestamp to each data point collected (including baggage change information, passenger itinerary data and airport equipment state data). The timestamp uses system clock synchronization (such as Network Time Protocol NTP) to ensure that all data points have a consistent UTC time reference.

[0050] Based on the added timestamp, all data points are sorted in time sequence to ensure that the events at the same time are consistent in position in the data stream;

[0051] S1.3: Associate the data points with the physical location, specifically: associate the baggage change information with the device coordinates of the occurrence location; associate the passenger itinerary data with the flight information; associate the flight information with the gate device coordinates; the airport equipment state data directly uses its own device coordinates, outputting the spatiotemporally aligned data point set, each data point containing timestamp, position coordinates and data content.

[0052] It should be noted that the data content specifically refers to the structured information field after semantic annotation, including the state change type in the baggage change information (such as "loading state change"), the flight binding relationship in the passenger itinerary data (such as "flight CA1234") and the running parameter in the airport equipment state data (such as "conveyor belt speed 0.5 m / s").

[0053] The data points after spatio-temporal alignment are semantically labeled, and device association labels are added.

[0054] Among them, the semantic labeling automatically generates semantic labels based on data content. Specifically, the luggage change information adds "change type" labels and "device association labels"; the passenger itinerary data adds "itinerary association labels" and "device association labels"; and the airport device status data adds "device association labels". Label generation uses simple rules (such as string matching or dictionary mapping).

[0055] For example, the luggage change information is labeled: the "change type" label is "loading status change", and the "device association label" is "sorting machine 001";

[0056] The passenger itinerary data is labeled: the "itinerary association label" is "flight binding", and the "device association label" is "boarding gate A12";

[0057] The airport device status data is labeled: the "device association label" is "conveyor belt 005";

[0058] Label generation rule: in the luggage status change event, if the status value contains "loading", the "change type" label is set to "loading status change"; the device identifier is directly used as the "device association label".

[0059] S1.4: Integrate all data points after processing to generate the original data stream.

[0060] The original data stream is structured into a field list: timestamp, device association label, data content, and semantic label, and is transmitted in real time through a data stream pipeline (such as Kafka or RabbitMQ) for subsequent steps.

[0061] S2: Input the original data stream into the pre-trained graph neural network, and construct a social relationship topology graph by analyzing the co-occurring flights, seat proximity, and ticket purchase time interval features in the passenger reservation records.

[0062] S2.1: Analyze the passenger reservation records in the original data stream, and extract co-occurring flight numbers, adjacent seat numbers, and ticket purchase time differences.

[0063] Specifically, from the original data stream, filter data entries with semantic labels "itinerary association labels", and extract all passenger reservation records;

[0064] Among them, each passenger reservation record contains a passenger unique identifier, a flight number, a seat number, and a ticket purchase time;

[0065] Group the passenger reservation records for each flight, and count the passengers who appear in the same flight to generate co-occurring flight numbers;

[0066] Query the flight seat physical layout database, calculate the physical adjacency of the passenger seat numbers of the same flight, and generate adjacent seat numbers.

[0067] Calculate the absolute difference (unit: seconds) of the ticket purchase timestamp for each pair of passengers on the same flight, and generate the ticket purchase time difference value.

[0068] S2.2: Apply a negative exponential decay function to the ticket purchase time difference value to generate a time decay coefficient, and perform a Euclidean distance inverse calculation on the adjacent seat numbers to generate spatial proximity.

[0069] Specifically, for the ticket purchase time difference value , apply a negative exponential decay function to calculate the time decay coefficient , the expression is:

[0070]

[0071] In the formula, is a predefined decay factor used to control the decay rate, is a natural constant representing the base number of the exponential function;

[0072] It should be noted that the predefined decay factor refers to fitting an exponential decay curve through historical flight passenger ticket purchase time difference data, and using the maximum likelihood estimation method to optimize the decay factor to obtain the optimal value.

[0073] Convert the adjacent seat numbers to physical coordinates (through the flight seat layout database), calculate the Euclidean distance between the two seats, the expression is:

[0074] ;

[0075] In the formula, represents the horizontal coordinate of passenger A's seat, represents the horizontal coordinate of passenger B's seat, represents the vertical coordinate of passenger A's seat, represents the vertical coordinate of passenger B's seat;

[0076] wherein passenger A and passenger B represent two passenger nodes connected by a dynamic weight edge in the social relationship topology graph, and the relevance of passenger A and passenger B is defined by the co-occurrence of flights, seat proximity, and ticket purchase time interval characteristics.

[0077] Apply the Euclidean distance inverse function to output the scalar value spatial proximity , the expression is:

[0078] ;

[0079] In the formula, This represents extremely small positive numbers (such as 0.01) and is used to prevent... Division by zero error.

[0080] S2.3: Using co-occurring flight numbers as initial weights for graph edges, a dynamic social weight matrix is ​​generated by fusing time decay coefficients and spatial proximity. Passenger node features are aggregated through a graph convolutional network to output a social relationship topology graph.

[0081] Specifically, co-occurring flight numbers are used as the basic weights of graph edges to construct an initial adjacency matrix between passenger nodes. The initial adjacency matrix is ​​then dynamically weighted by fusing time decay coefficients and spatial proximity to generate a dynamic social weight matrix. ;

[0082] The total number of flights for each passenger in the passenger booking record is counted and used as the initial node feature;

[0083] Perform graph convolution on the initial node features to output the updated node feature matrix, as shown in the expression:

[0084] ;

[0085] In the formula, Indicates the first The node feature matrix of the layer This represents a non-linear activation function (such as ReLU). Degree matrix The negative one-half power (standardized term). Indicates the first The node feature matrix of the layer Indicates the first The trainable parameter matrix of the layer;

[0086] Based on passenger unique identifiers, updated node feature matrix, and dynamic social weight matrix The values ​​are used to construct the social relationship topology graph structure and output the social relationship topology graph.

[0087] S3: Based on the social relationship topology, trigger the group baggage priority weighting mechanism to generate collaborative transfer paths for passenger baggage with itinerary connections.

[0088] S3.1: When the flight status associated with any passenger node in the social relationship topology changes, traverse all baggage transfer statuses connected to the changed node to obtain the loading status identifier code and the current location coordinates.

[0089] Specifically, the system monitors the flight status associated with passenger nodes in real time. When the flight number, gate, or departure time of any passenger node changes, the corresponding passenger node is marked as the changed node.

[0090] In the social relationship topology, query all passenger nodes connected with the change node by dynamic weight edges (i.e., passengers with itinerary association).

[0091] Through the unique luggage identifier (binding the unique passenger identifier), obtain the loading state identifier code ("unloaded" or "loaded") of each luggage from the luggage change information. Through the unique luggage identifier, obtain the current position coordinates of the luggage (such as the sorting carousel coordinates) from the airport equipment state data, and output all luggage state sets associated with the change node.

[0092] S3.2: If the luggage loading state identifier code is unloaded, generate a flight binding instruction to forcibly synchronize the current luggage to the target flight after the change.

[0093] Specifically, for the luggage state set, filter the luggage items with the loading state identifier code "unloaded", and extract the ID of the luggage-owning passenger to generate an unloaded luggage list.

[0094] The ID of the luggage-owning passenger refers to the unique identity code assigned by the airline to each passenger in the passenger reservation record, usually composed of alphanumeric characters (such as ABC123). The unique identity code is bound with the passenger's passport information and flight itinerary data, ensuring accurate association between the passenger and the checked baggage during the luggage handling process.

[0095] Among them, the unloaded luggage list contains the unique luggage identifier (identifying the luggage itself) and the associated unique passenger identifier (identifying the luggage-owning passenger).

[0096] Create a structured instruction for each unloaded luggage, and output a set of flight binding instructions.

[0097] S3.3: If the luggage loading state identifier code is loaded, calculate the shortest transfer time from the current position coordinates of the corresponding luggage to the target boarding gate, and generate a virtual synchronization time window constraint.

[0098] Specifically, for the luggage state set, filter the items with the loading state identifier code "loaded" to generate a loaded luggage list.

[0099] According to the target flight number after the change, query the airport flight database to obtain the target boarding gate coordinates.

[0100] It should be noted that the airport flight database refers to a structured data set that stores flight dynamic information and airport geographic spatial data, including the association relationship between flight number, planned take-off and landing time, assigned boarding gate number, and boarding gate three-dimensional coordinates, which is updated in real time through the airline's operation platform or airport ground management platform.

[0101] Calculate the shortest path distance from the current position of the luggage to the target coordinate of the boarding gate using the A* algorithm Output the shortest transfer time value of each piece of luggage The expression is:

[0102] ;

[0103] In the formula, is the average speed of the luggage conveyor belt;

[0104] It should be noted that the average speed of the luggage conveyor belt is determined by measuring the actual running speed of the airport conveyor belt (such as multiple sampling and averaging), and the example value is usually 0.5 meters per second (the specific value needs to be calibrated according to the equipment parameters of the airport).

[0105] When generating the time window constraint, take the boarding gate closing time in the flight plan as the basis, form a time difference value between the shortest luggage transfer time and the boarding gate closing time, determine the latest time boundary for the luggage to arrive at the boarding gate, and thus define the effective time range for the luggage transfer operation.

[0106] S3.4: According to the dynamic weight value of the connection edge in the social relationship topology graph, assign a transfer order adjustment coefficient to the passenger node constrained by the virtual synchronization time window, and output the collaborative transfer path.

[0107] Specifically, extract the dynamic social weight matrix corresponding weight value between the passenger node and the change node constrained by the time window from the social relationship topology graph Calculate the adjustment coefficient The expression is:

[0108] ;

[0109] In the formula, represents the sum of the weight values of all associated nodes;

[0110] Sort the constrained passenger nodes in descending order of the adjustment coefficient, and output the priority The higher the priority, the higher the priority;

[0111] For each piece of luggage, generate a collaborative transfer path in combination with the time window constraint and the priority;

[0112] It should be noted that high-priority luggage: assigns an earlier conveyor belt period and a shorter path; low-priority luggage: allows the use of a backup path or a delayed period.

[0113] S4: Build an airport operation digital twin in the metaverse, import the collaborative transfer path into the digital twin environment for spatiotemporal pre-visualization simulation, and real-time detect luggage flow conflict points and mark the physical constraint conflict region coordinate set.

[0114] S4.1: Load airport building information model and real-time device IoT data stream into metaverse environment, build dynamic digital twin.

[0115] Specifically, airport three-dimensional layout data is obtained from the airport flight database. The airport three-dimensional layout data includes building structure, device location and path network;

[0116] Subscribe to real-time device IoT data stream using data stream pipeline;

[0117] Integrate airport three-dimensional layout data and real-time device IoT data stream into metaverse environment, establish real-time mapping relationship in in-memory database (such as Redis), form digital twin real-time update device state and luggage location.

[0118] S4.2: Import collaborative transfer path into dynamic digital twin, and perform luggage transfer space-time simulation to calculate sorting carousel volume overload rate, path intersection collision probability and device response delay time.

[0119] Specifically, the collaborative transfer path is loaded as input into the dynamic digital twin.

[0120] Start luggage transfer space-time simulation, which simulates the process of luggage moving along the path based on timestamp sequence sorting. In the simulation: simulate the process of luggage moving from the current location to the target boarding gate, stepwise update the luggage location (every millisecond step to reflect real-time dynamics). Calculate the sorting carousel volume overload rate: monitor the number of luggage on each sorting carousel, and compare it with the maximum capacity of the sorting carousel, output the overload rate value. Calculate the path intersection collision probability: detect whether the coordinates of the luggage at the same time overlap, overlap indicates collision event, output probability value. Calculate the device response delay time: record the start time and actual completion time of the device processing instruction, output the delay value.

[0121] Detect luggage flow conflict points, which are the positions and time points in the simulation where the overload rate is too high, the collision probability is abnormal or the delay time is out of limit, output the conflict point list. The fields of the conflict point list include conflict type (such as "overload"), current luggage position coordinates (sorting carousel or path intersection) and timestamp.

[0122] S4.3: Activate physical engine to simulate luggage stacking and tipping effect and conveyor belt vibration offset, and mark physical constraint conflict area coordinate set exceeding safety threshold.

[0123] Wherein, the physical engine refers to a software component that simulates physical phenomena (such as based on Newtonian mechanics), and the parameter settings include luggage weight, friction coefficient and conveyor belt vibration parameters.

[0124] Specifically, enable the physical engine function to execute the luggage stack collapse effect and conveyor belt vibration offset simulation: when the coordinates of multiple pieces of luggage are close to each other, calculate the stack height and stability. If the stack height exceeds the stability angle, predict the collapse event through the physical engine function, and at the same time, in the simulation of the conveyor belt vibration offset, apply random vibration disturbance to the conveyor belt position coordinates, and calculate the offset amplitude by integral operation.

[0125] Monitor whether the parameters in the physical engine simulation exceed the safety threshold, which includes the stack height limit and the offset amplitude limit, where the stack height limit is the maximum allowed value of the ratio of the luggage stack height to the bottom width, and the offset amplitude limit is the maximum allowed value of the ratio of the conveyor belt vibration displacement to the conveyor belt width; when it is monitored that any parameter exceeds the limit, record the three-dimensional space coordinates and time stamp of the corresponding conflict point.

[0126] Combine the results of the collapse event and the vibration offset overrun, and associate the luggage current position coordinates and time stamps in the conflict point list to generate a physical constraint conflict region coordinate set;

[0127] The physical constraint conflict region coordinate set is a three-dimensional coordinate set, including the conflict type (such as stack collapse), three-dimensional coordinates, and associated time stamps.

[0128] S5: Reverse the conflict events in the pre-simulation through the inverse reinforcement learning algorithm, trace the strategy defect characteristics, and reconstruct the reward function space, output the gradient compensation parameters.

[0129] S5.1: Combine the physical constraint conflict region coordinate set with the digital twin environment parameters to generate a state-action pair sample set as input data for the inverse reinforcement learning algorithm.

[0130] Specifically, align the physical constraint conflict region coordinate set with the digital twin environment parameters according to the time stamp. After alignment, each conflict point is associated with the digital twin environment parameter value at the corresponding time;

[0131] State: composed of conflict point coordinates, sorting carousel volume overload rate at the associated time, path intersection collision probability, and device response delay length; Action: defined as the instruction related to the conflict point in the collaborative transfer path (such as path navigation point coordinate sequence or motor control instruction).

[0132] Each sample is a two-tuple (state, action), and the state-action pair sample set is output.

[0133] S5.2: Calculate the probability distribution of the state-action pair sample set through the maximum entropy probability model, and inversely deduce the optimal strategy corresponding to the reward function.

[0134] Specifically, the state features (device response delay duration, path intersection collision probability, and sorting carousel volume overload rate) are extracted from the state-action pair sample set as input items of the reward function.

[0135] The state-action pair sample set is input into the maximum entropy probability model. The maximum entropy probability model is a standard algorithm for inverse reinforcement learning to back-propagate the reward function. Specifically, based on the co-occurrence frequency of states and actions in the state-action pair sample set, a probability distribution function is fitted, the probability distribution of the state-action pair sample set is calculated, and the reward function form that optimizes the probability distribution is solved by maximizing the likelihood of the state-action pair sample set. By comparing the state feature values and weight distribution when the conflict event occurs, defects such as low device response delay weight or insufficient path intersection collision penalty coefficient are identified.

[0136] The reward function output is a linear weighted expression, and the weight variable corresponds to the conflict detection index (such as device response delay weight, path intersection collision probability weight).

[0137] S5.3: Analyze the weight distribution defects of the reward function, identify the device response delay weight and abnormal path intersection penalty coefficient that cause the conflict, and reconstruct the reward function space structure.

[0138] Specifically, the weight distribution of the reward function is analyzed, and the weight value corresponding to the device response delay duration and the penalty coefficient corresponding to the path intersection collision probability are extracted.

[0139] If the device response delay weight is lower than the preset weight threshold, it is marked as a defect feature.

[0140] If the abnormal path intersection penalty coefficient deviates from the historical normal range, it is marked as a defect feature.

[0141] The device response delay weight is increased to the preset weight threshold, the abnormal path intersection penalty coefficient is corrected to the median of the historical normal range, and the other weights remain unchanged, and the reconstructed reward function expression is output.

[0142] It should be noted that when determining the preset weight threshold, the P95 quantile value of the device response delay weight distribution in the historical operation data is used as the reference, and the historical normal range is determined by the interquartile range (IQR) of the abnormal path intersection penalty coefficient in the past three months. For example, the preset threshold range of the device response delay weight may be between 0.15 and 0.35, and the historical normal range of the abnormal path intersection penalty coefficient may fluctuate within the interval of 0.05-0.25.

[0143] S5.4: Inject device fault simulation noise and path random offset into the state-action pair sample set to generate adversarial perturbation samples and train the strategy network model to output gradient compensation parameters.

[0144] Specifically, device failure simulation noise is superimposed on the device response delay duration feature of the state-action pair sample set; random position offsets are injected into the path navigation point coordinate sequence in the action field to generate an expanded set of adversarial perturbation samples;

[0145] A graph convolution network structure is used as the strategy network model, the input is the state field of the adversarial perturbation sample set, and the output is the predicted value of the action field;

[0146] It should be noted that the training process of using the graph convolution network structure as the strategy network model is as follows: a graph data structure containing a node feature matrix and an adjacency matrix is constructed, wherein the node feature matrix is composed of state parameters of the adversarial perturbation sample set (including device response delay duration, path intersection collision probability, etc. Dimensional data), and the adjacency matrix quantifies the association strength between state parameters; the node feature matrix is processed through multi-layer graph convolution operation, each layer performs linear transformation and feature propagation with the adjacency matrix, and adopts ReLU activation function; finally, the processed node feature matrix is mapped to the predicted value of the action parameter, and the mean square error is used as the loss function during training, the network parameters are optimized through back propagation, and the Adam optimizer is used to complete the convergence of the strategy network model.

[0147] The gradient value of the loss function of the strategy network model to the state parameters contained in the node feature matrix is calculated, the gradient value is normalized to a compensation parameter vector, and the gradient compensation parameter is output. The gradient compensation parameter includes a device response delay compensation coefficient, a path intersection compensation coefficient, and a volume overload compensation coefficient.

[0148] S6: Fuse the physical constraint conflict region coordinate set and the gradient compensation parameter to generate a three-dimensional space scheduling instruction set and execute the instruction issuing.

[0149] S6.1: According to the gradient compensation parameter, calculate the start time offset value of the motor of the sorting device, and generate the motor control instruction.

[0150] Specifically, the product of the device response delay compensation coefficient and the device response delay duration is taken as the time compensation amount, and the time compensation amount is converted into a clock period adjustment value in combination with the pre-stored motor acceleration characteristic parameter;

[0151] The clock period adjustment value is written into the PLC timer register to generate the motor control instruction;

[0152] The pre-stored motor acceleration characteristic parameter refers to the inherent performance data obtained through motor factory test, including the start time-speed curve under rated voltage, the maximum allowable acceleration slope, and the inertia compensation coefficient, which is usually stored in the device controller in the form of a lookup table, and is used to accurately calculate the clock period compensation value.

[0153] The motor control instruction includes the sorting equipment number, the adjusted start timestamp, and the execution priority.

[0154] S6.2: Based on the physical constraint conflict area coordinate set, the A* algorithm is used to re-plan the ground crew travel path, and the path navigation point coordinate sequence is output.

[0155] Specifically, the physical constraint conflict area coordinate set is read, and the path planning parameters are set.

[0156] The path planning parameters include: start point: the current position coordinate of the ground crew (the airport equipment state data contains the physical position information); end point: the coordinate of the nearest safe point outside the conflict area (calculated through the boundary of the coordinate set); obstacle: all points in the physical constraint conflict area coordinate set.

[0157] Using the same A* algorithm in step S3.3, the A* algorithm path planning is performed, and the path navigation point coordinate sequence is output.

[0158] S6.3: Align the motor control instruction with the path navigation point coordinate sequence according to the timestamp, generate a structured instruction data packet, send the motor control instruction to the baggage sorting controller through the message queue, and transmit the path navigation point coordinate sequence to the ground crew handheld terminal through the wireless protocol.

[0159] Specifically, the motor control instruction and the path navigation point coordinate sequence are aligned according to the timestamp, a structured instruction data packet is generated, and the instruction is issued. The fields of the structured instruction data packet include: instruction type (motor control / path navigation), target device / personnel identification, execution time, and content data.

[0160] It should be noted that the timestamp of the motor control instruction is matched with the time window of the path navigation point; the motor control instruction is sent to the baggage sorting controller through the message queue (such as Kafka). The path navigation point coordinate sequence is transmitted to the ground crew handheld terminal through the wireless protocol (such as Wi-Fi 6).

[0161] The embodiment also provides a scheduling system matched with the luggage change information, comprising a data acquisition module, a graph construction module, a path generation module, a simulation rehearsal module, a strategy optimization module and an instruction scheduling module; the data acquisition module is used for collecting luggage change information, passenger itinerary data and airport equipment state data in an air transportation scene in real time, and outputting an original data stream; the graph construction module is used for inputting the original data stream into a pre-trained graph neural network, constructing a social relationship topology graph by analyzing co-occurring flights, seat proximity and ticket purchase time interval features in passenger seat reservation records; the path generation module is used for triggering a group luggage priority weighting mechanism according to the social relationship topology graph, and generating a collaborative transfer path for passenger luggage with itinerary association; the simulation rehearsal module is used for constructing an airport operation digital twin in a metaverse, importing the collaborative transfer path into a digital twin environment for space-time pre-rehearsal simulation, and detecting luggage flow conflict points and marking a physical constraint conflict region coordinate set in real time; the strategy optimization module is used for inversely solving conflict events in the pre-rehearsal simulation through an inverse reinforcement learning algorithm, tracing a strategy defect feature and reconstructing a reward function space, and outputting a gradient compensation parameter; and the instruction scheduling module is used for fusing the physical constraint conflict region coordinate set and the gradient compensation parameter, generating a three-dimensional space scheduling instruction set and executing instruction issuing.

[0162] The embodiment also provides a computer device suitable for the scheduling method of the luggage change information, comprising a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the scheduling method of the luggage change information.

[0163] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used for providing computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device. The input device can also be an external keyboard, touchpad or mouse, etc.

[0164] The embodiment also provides a storage medium on which a computer program is stored, the program being executed by a processor to implement the scheduling method for realizing change information of matched airline baggage as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disk.

[0165] To sum up, the application constructs an airport operation digital twin in a meta-universe environment, accurately simulates the luggage stacking and dumping effect and the conveyor belt vibration and deviation by using a physics engine, dynamically calculates the sorting disc volume overload rate, path intersection collision probability, and equipment response delay time in real time, and generates a high-precision three-dimensional conflict area coordinate set; based on a time-space pre-rehearsal simulation method, accurately identifies the safety boundary of the luggage transfer path, and effectively reduces the risk of exceeding the physical constraints. By integrating real-time device Internet of Things data streams, the spatio-temporal consistency of conflict detection is significantly improved, reliable data support is provided for gradient compensation parameter generation, and the robustness and execution efficiency of the scheduling instruction are enhanced.

[0166] It should be noted that the above embodiments are only used to illustrate the technical solutions of the application and not to limit the application. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the application, and all should be covered in the scope of the claims of the application.

Claims

1. A scheduling method for supporting airline baggage change information, characterized in that: include, Real-time collection of baggage change information, passenger itinerary data, and airport equipment status data in air transport scenarios, and output of raw data stream; The raw data stream is input into a pre-trained graph neural network, and a social relationship topology graph is constructed by analyzing the co-occurring flights, seat proximity, and ticket purchase time interval features in passenger booking records. Based on the social relationship topology, a group baggage priority weighting mechanism is triggered to generate collaborative transfer paths for baggage of passengers with related travel itineraries; In the metaverse, a digital twin of airport operations is constructed. Collaborative transfer routes are imported into the digital twin environment for spatiotemporal simulation, real-time detection of baggage flow conflict points, and marking of the coordinate set of physically constrained conflict areas. The specific steps are as follows. By loading airport building information model and real-time equipment IoT data stream into the metaverse environment, a dynamic digital twin is constructed; The collaborative transfer path is imported into a dynamic digital twin, and a spatiotemporal simulation of baggage transfer is performed to calculate the sorting carousel volume overload rate, the probability of path intersection collision, and the equipment response delay time. Activate the physics engine to simulate the effects of luggage stacking and tipping, as well as conveyor belt vibration and offset, and mark the coordinate set of physical constraint conflict areas that exceed the safety threshold; The physical constraint conflict area coordinate set includes: enabling the physics engine function to simulate baggage stacking and tipping effects and conveyor belt vibration offset; if the stacking height exceeds the stable angle, predicting the tipping event through the physics engine function, and simultaneously applying random vibration disturbances to the conveyor belt position coordinates during the conveyor belt vibration offset simulation, calculating the offset amplitude through integration; monitoring whether the parameters in the physics engine simulation exceed safety thresholds, including stacking height limits and offset amplitude limits; when a parameter exceeds either limit, recording the three-dimensional spatial coordinates and timestamp of the corresponding conflict point; merging the tipping event and the result of exceeding the vibration offset limit, associating them with the current position coordinates and timestamp of the baggage in the conflict point list, and generating the physical constraint conflict area coordinate set; the physical constraint conflict area coordinate set is a three-dimensional coordinate set, including the conflict type, three-dimensional coordinates, and associated timestamps; By using the inverse reinforcement learning algorithm to solve the conflict events in the pre-simulation, the source of policy defect features is traced and the reward function space is reconstructed, and gradient compensation parameters are output. By integrating the coordinate set of the conflicting regions of physical constraints with gradient compensation parameters, a three-dimensional spatial scheduling instruction set is generated and the instructions are issued.

2. The scheduling method for supporting air baggage change information as described in claim 1, characterized in that: The passenger travel data includes passenger reservation records.

3. The scheduling method for supporting air baggage change information as described in claim 1, characterized in that: The specific steps for constructing the social relationship topology graph are as follows: Parse the passenger reservation records in the raw data stream to extract co-occurring flight numbers, adjacent seat numbers, and ticket purchase time differences; A time decay coefficient is generated by applying a negative exponential decay function to the ticket purchase time difference, and spatial proximity is generated by calculating the inverse Euclidean distance between adjacent seat numbers. Using co-occurring flight numbers as initial weights for graph edges, a dynamic social weight matrix is ​​generated by fusing time decay coefficients and spatial proximity. Passenger node features are aggregated through a graph convolutional network to output a social relationship topology graph.

4. The scheduling method for supporting air baggage change information as described in claim 3, characterized in that: The specific steps for generating collaborative transfer routes for passenger baggage with itinerary associations are as follows: When the flight status associated with any passenger node in the social relationship topology changes, traverse all baggage transfer statuses connected to the changed node to obtain the loading status identifier code and the current location coordinates; If the baggage loading status code is "not loaded", a flight binding instruction will be generated to forcibly synchronize the current baggage to the changed target flight; If the baggage loading status code is "loaded", calculate the shortest transfer time from the current location coordinates of the corresponding baggage to the boarding gate of the target flight, determine the virtual synchronization time window and use it as a constraint condition; Based on the dynamic weight values ​​of the connecting edges in the social relationship topology graph, transfer order adjustment coefficients are assigned to passenger nodes constrained by the virtual synchronization time window, and a collaborative transfer path is output.

5. The scheduling method for supporting air baggage change information as described in claim 4, characterized in that: The specific steps for setting the output gradient compensation parameters are as follows: The coordinate set of the conflict region of physical constraints is combined with the parameters of the digital twin environment to generate a state-action pair sample set, which is used as input data for the inverse reinforcement learning algorithm; The probability distribution of state-action pairs on the sample set is calculated using the maximum entropy probability model, and the reward function corresponding to the optimal policy is derived from this. The study analyzes the weight allocation defects of the reward function, identifies the device response delay weights and abnormal path crossover penalty coefficients that cause conflicts, and reconstructs the spatial structure of the reward function. Inject device fault simulation noise and random path offsets into the state-action pair sample set to generate adversarial perturbation samples, and train the policy network model to output gradient compensation parameters.

6. The scheduling method for supporting air baggage change information as described in claim 5, characterized in that: The specific steps for generating and executing the three-dimensional spatial scheduling instruction set are as follows. Based on the gradient compensation parameters, the start-up time offset of the sorting equipment motor is calculated, and motor control commands are generated. Based on the coordinate set of the conflict area of ​​physical constraints, the A* algorithm is used to replan the travel path of ground staff and output the coordinate sequence of the path navigation point. The motor control commands and the path navigation point coordinate sequence are aligned with the timestamps to generate a structured command data packet. The motor control commands are sent to the baggage sorting controller through a message queue, while the path navigation point coordinate sequence is transmitted to the ground staff handheld terminal through a wireless protocol.

7. A scheduling system for supporting airline baggage change information, based on the scheduling method for supporting airline baggage change information as described in any one of claims 1 to 6, characterized in that: It includes a data acquisition module, a map construction module, a path generation module, a simulation and pre-play module, a strategy optimization module, and an instruction scheduling module; The data acquisition module is used to collect baggage change information, passenger itinerary data and airport equipment status data in the air transport scenario in real time, and output raw data stream; The graph construction module is used to input the raw data stream into a pre-trained graph neural network and construct a social relationship topology graph by analyzing the co-occurring flights, seat proximity, and ticket purchase time interval features in the passenger booking records. The path generation module is used to trigger a group baggage priority weighting mechanism based on the social relationship topology graph to generate collaborative transfer paths for passenger baggage with itinerary associations; The simulation pre-play module is used to construct a digital twin of airport operations in the metaverse, import the collaborative transfer path into the digital twin environment for spatiotemporal pre-play simulation, and detect baggage flow conflict points in real time and mark the coordinate set of physical constraint conflict areas. The policy optimization module is used to solve the conflict events in the pre-simulation by using the inverse reinforcement learning algorithm, trace the source of policy defect features and reconstruct the reward function space, and output gradient compensation parameters. The instruction scheduling module is used to fuse the coordinate set of the physical constraint conflict region with the gradient compensation parameters, generate a three-dimensional spatial scheduling instruction set, and execute the instruction issuance.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the scheduling method for the accompanying air baggage change information as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the scheduling method for the accompanying air baggage change information as described in any one of claims 1 to 6.

Images