Data processing method, device and medium for associating flights and parking spaces

Abstract

The present invention relates to the field of aviation data processing technology, and in particular to a data processing method, device, and medium for associating flights and parking stands, comprising: obtaining an initial conflict region group set and an initial association feasibility matrix; if the current conflict region group set is not empty, selecting a conflict region group from the current conflict region group set as the current region group to be processed, and deleting the region group; generating a network graph corresponding to the current region group to be processed as the current network graph to be processed; based on the current network graph to be processed and a pre-trained graph neural network model, obtaining the maximum weighted independent node set corresponding to the network graph as the association result corresponding to the conflict region group, and then updating the current association feasibility matrix. The present invention obtains a parking stand allocation operation by obtaining the maximum weighted independent set of the network graph, which can effectively improve the docking rate, computational efficiency, and soft preference of the parking stand allocation.

Description

Technical Field

[0001] The present invention relates to the technical field of aviation data processing, and in particular to a data processing method, device and medium for associating flights and parking spaces. Background Art

[0002] At large hub airports, parking stand allocation is a key factor affecting airport operational efficiency. Reasonable parking stand allocation not only improves flight turnaround efficiency, but also reduces flight delays, enhances the passenger experience, and optimizes airport resource utilization. The parking stand allocation problem requires comprehensive consideration of complex constraints such as aircraft type matching, flight arrival times, parking stand conflicts, and flight stopover times to ensure the safety and stability of flight operations. However, due to the continuous expansion of airport scale and the continuous growth in the number of flights, existing parking stand allocation methods face many challenges in practical applications, including low docking rates, low computational efficiency, and lack of generalization. Summary of the Invention

[0003] In view of the above technical problems, the technical solution adopted by the present invention is:

[0004] According to a first aspect of the present invention, there is provided a data processing method for associating flights and parking spaces, the method comprising the following steps:

[0005] S1. Obtain an initial conflict region group set and an initial association feasibility matrix corresponding to a target geographic region. The initial conflict region group set includes multiple conflict region groups that do not conflict with each other. The initial conflict region group set is obtained based on an association region conflict relationship matrix of the target geographic region. The initial association feasibility matrix is ​​used to indicate whether associations between association regions and associated objects are permitted. The association regions are parking stands, and the associated objects are flights.

[0006] S2: If the current conflicting region group set is not empty, select a conflicting region group from the current conflicting region group set as the current region group to be processed, and delete the current region group to be processed from the current conflicting region group set; the initial value of the current conflicting region group set is the initial conflicting region group set.

[0007] S3, generating a network diagram corresponding to the current group of regions to be processed as the current network diagram to be processed; the nodes in the network diagram represent the association operations between the associated regions and the associated objects, and the two nodes corresponding to the two conflicting association operations are connected by a connecting line.

[0008] S4, based on the current network graph to be processed and the pre-trained graph neural network model, obtain the mapping value corresponding to each node in the network graph, obtain the node mapping value set corresponding to the network graph, and based on the node mapping value set corresponding to the network graph and the connection relationship between the nodes, obtain the maximum weighted independent node set corresponding to the network graph, and use it as the association result of the current conflict area group to be processed; wherein, the mapping value of each node is used to represent the predicted probability that the node belongs to the maximum weighted independent set, and there is no connecting line between any two nodes in the maximum weighted independent node set.

[0009] S5, based on the association result corresponding to the current group of regions to be processed, the current association feasibility matrix is ​​updated, and S2 is executed; the initial value of the current association feasibility matrix is ​​the initial association feasibility matrix.

[0010] According to a second aspect of the present invention, an electronic device is provided, comprising a processor and a memory; the processor is configured to execute the steps of the method according to the first aspect of the present invention by calling a program or instruction stored in the memory.

[0011] According to a third aspect of the present invention, there is provided a computer-readable storage medium storing a program or instructions, wherein the program or instructions enable a computer to execute the steps of the method according to the first aspect of the present invention.

[0012] The present invention has at least the following beneficial effects:

[0013] An embodiment of the present invention provides a data processing method for associating flights and parking stands, including: obtaining an initial set of conflicting area groups and an initial association feasibility matrix corresponding to a target geographic area; selecting a conflicting area group from the current set of conflicting area groups as the current area group to be processed, and deleting the current area group to be processed; generating a network graph of the current area group to be processed as the current network graph to be processed; obtaining a mapping value corresponding to each node in the network graph based on the current network graph to be processed and a pre-trained graph neural network model, obtaining a node mapping value set corresponding to the network graph; and obtaining a maximum weighted independent node set corresponding to the network graph based on the node mapping value set and the connection relationship between nodes in the network graph as the association result for the corresponding conflicting area group, and updating the current association feasibility matrix based on the association result. The present invention models the parking stand allocation problem as a combinatorial optimization problem on the network graph corresponding to each conflicting stand group, and obtains the maximum weighted independent set of the network graph using the pre-trained graph neural network model to obtain the parking stand allocation operation. This method can effectively improve the docking rate and computational efficiency of parking stand allocation, while better meeting soft preference requirements.

[0014] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 A flowchart of a data processing method for associating flights and parking spaces provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0019] It should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of the steps can be performed in parallel, concurrently, or simultaneously. In addition, the order of the steps can be rearranged. A process can be terminated when its operation is completed, but can also have additional steps not included in the accompanying drawings. A process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0020] An embodiment of the present invention provides a data processing method for associating flights and parking stands, which is used to perform an operation of associating flights and parking stands under the following assumptions:

[0021] Assumption 1: Aircraft towing is not considered, that is, the aircraft cannot change parking positions during parking.

[0022] Assumption 2: To prevent the situation where there are no available parking spaces for an aircraft, it is assumed that there exists a virtual remote parking space with infinite capacity that satisfies various constraints.

[0023] Assumption 3: Assume that the arrival and departure times of flights are fixed, and do not consider the uncertainty of the arrival and departure times of flights.

[0024] Assumption 4: Flight cancellations and last-minute aircraft changes are not considered.

[0025] like Figure 1 As shown, the data processing method for associating flights and parking spaces provided by the embodiment of the present invention may include the following steps:

[0026] S1, obtaining an initial conflict region group set and an initial association feasibility matrix corresponding to the target geographical area.

[0027] In an embodiment of the present invention, the target geographical area may be an airport, the associated area may be a parking space, and the associated object may be a flight.

[0028] In the embodiment of the present invention, before S1, the following steps are further included:

[0029] Get the associated region information and associated object information of the target geographic region.

[0030] The associated area information may include the location number, area identifier, area model, and area attribute information of the associated area.

[0031] The associated area's location number represents the associated area's location within the target geographic area. The area identifier includes a first area identifier and a second area identifier. The associated area with the first area identifier is the first associated area, and the associated area with the second area identifier is the second associated area. The first and second area identifiers can be represented by different numbers, such as 1 and 0, respectively. In one specific example, the first associated area is a near-camera position, and the second associated area is a far-camera position. The area model represents the size of the associated area.

[0032] The attribute information of the associated area may include the area location identifier, the target field allowed for association, and the flight mission types allowed for association. The target field may be a two-character airline code. The area location identifier is used to indicate whether the associated area belongs to the area for associated objects used to dock and perform international missions or the area for associated objects used to dock and perform domestic missions, that is, whether it belongs to an international flight stand or a domestic flight stand. If the area location identifier is 1, it means that the associated area belongs to the area for associated objects used to dock and perform international missions. If the area location identifier is 0, it means that the associated area belongs to the area for associated objects used to dock and perform domestic missions.

[0033] The associated object information may include the associated object's ID, carrier model, planned arrival time, planned departure time, object attribute information, VIP object identification, overnight object identification, area selection weight, etc.

[0034] The associated object ID is the associated object's unique identifier, such as a flight number. The associated object's carrier model is the associated object's carrier model, i.e., the aircraft model. The planned arrival time is the planned arrival time at the associated area, and the planned departure time is the planned departure time from the associated area. The VIP object identifier indicates whether the associated object is a VIP object, i.e., whether it is a VIP flight. If the VIP object identifier is 1, it indicates a VIP object; if the VIP object identifier is 0, it indicates a non-VIP object. The overnight object identifier indicates whether the associated object is an overnight object, i.e., whether it is an overnight flight. If the overnight object identifier is 1, it indicates an overnight object; if the overnight object identifier is 0, it indicates a non-overnight object. The associated object's area selection weight represents the weight of the associated object's selection for different associated areas, i.e., the degree to which the associated object prefers different associated areas. For example, an airport may prefer VIP aircraft to dock at close gates, but aircraft with long layovers or overnight stays may prefer to be assigned to far gates. The more an associated object prefers to dock in a certain area, the greater the area selection weight assigned to that area. The area selection weight can be expressed as a decimal between 0 and 5. The regional selection weight reflects the soft preference of flights.

[0035] The object attribute information of the associated object may include the corresponding target field, object attribute identifier, flight mission type, etc. The flight mission type is used to characterize the type of mission performed by the associated object, such as a cargo mission or a training flight mission. Different mission types can be represented by different identifier values. The object attribute identifier is used to indicate whether the associated object is an object performing an international mission or an object performing a domestic mission. If the object attribute identifier is 1, it indicates that the associated object is an object performing an international mission, that is, an international flight. If the object attribute identifier is 0, it indicates that the associated object is an object performing a domestic mission, that is, a domestic flight.

[0036] In the embodiment of the present invention, the initial association feasibility matrix is ​​used to represent whether association is allowed to be established between the association area and the association object, and can be determined based on the association area information, the association object information and the association constraint information.

[0037] In an embodiment of the present invention, the association constraint information is used to constrain the carrier model and object attribute information of the associated object to match the regional model and regional attribute information of the associated area, that is, the carrier model of the associated object needs to be less than or equal to the regional model of the associated area, and the object attribute information and the regional attribute information need to be consistent. In this way, based on the association constraint information, according to the regional model and regional attribute information of the associated area in the associated area information and the carrier model and object attribute information of the associated object in the associated object, the matched associated areas and associated objects can be associated to obtain an initial association feasibility matrix, that is, to obtain an initial feasible domain matrix. It is known to those skilled in the art that in the initial association feasibility matrix, the same associated object may be associated with multiple associated areas, and the same associated area may also be associated with multiple associated objects.

[0038] In an embodiment of the present invention, the initial association feasibility matrix may be a two-dimensional matrix of size N×M. N is the number of associated objects, and M is the number of associated regions. The elements in the initial association feasibility matrix are composed of a first association identifier and a second association identifier. The first association identifier indicates that the corresponding associated region and the associated object are allowed to be associated, and the second association identifier indicates that the corresponding associated region and the associated object are not allowed to be associated. In an exemplary embodiment, the first association identifier may be 1, and the second association identifier may be 0. For example, if the element R in the k1th row and k2th column of the initial association feasibility matrix is k1k2 =1, indicating that the k1th associated object is allowed to be associated with the k2th associated area. The value of k1 ranges from 1 to N, and the value of k2 ranges from 1 to M.

[0039] In this embodiment of the present invention, the initial conflict region group set includes multiple conflict region groups that do not conflict with each other. Specifically, the same initial conflict region group is composed of conflicting conflict regions, and there is no conflict between different initial conflict region groups, that is, there is no conflict between associated regions in different conflict region groups.

[0040] In an embodiment of the present invention, the existence of a conflict between two associated areas means that there may be a potential taxiing conflict between two associated objects respectively associated with the two associated areas. Taxiing conflicts may include double push-out conflicts, double slide-in conflicts, and push-out and slide-in conflicts. Among them, a double push-out conflict refers to when flights parked at two adjacent parking stands have similar departure times, and a conflict may occur during the push-out process. A double slide-in conflict refers to when flights parked at two adjacent parking stands have similar arrival times, and a conflict may occur during the taxiing process. A slide-in push-out conflict refers to when flights parked at two adjacent parking stands have similar arrival times, and a conflict may occur during the taxiing process.

[0041] In this embodiment of the present invention, the associated areas in the same initial conflict area group share the same taxiway and have the same area identifier. That is, all the close stands in the close stand group are close stands, and all the far stands in the far stand group are far stands. In this embodiment of the present invention, each conflict area group is composed of the position numbers of the corresponding associated areas. The initial conflict area group set can be obtained based on the associated area conflict relationship matrix of the target geographic area. The associated area conflict relationship matrix can be expressed as:

[0042] .

[0043] Among them, B is the conflict relationship matrix of the associated region, L s2s1 is the conflict identification value between the s2th association area and the s1th association area. If there is a conflict between the s2th association area and the s1th association area, L s2s1 =1, otherwise, L s2s1 =0, and the values ​​of s1 and s2 are both 1 to M. It can be seen that B is a symmetric matrix and all diagonal elements are 1.

[0044] In this embodiment of the present invention, the target geographic area's association area conflict relationship matrix represents whether a conflict exists between two adjacent association areas. This matrix is ​​a pre-determined matrix. Whether a conflict exists between two adjacent association areas can be determined based on the locations of the two adjacent association areas and prior knowledge, i.e., whether a potential taxiing conflict exists between the two adjacent association areas.

[0045] In this embodiment of the present invention, the initial conflict region group set may be obtained by the following steps:

[0046] S10: Based on the M associated regions, obtain M initialization region sets. Each element in the initialization region set is a position number of the corresponding associated region. Set a counter e=1.

[0047] S11, if e≤M, execute S12, otherwise, execute S15.

[0048] S12: If the position number corresponding to the e-th initialized region set exists in the current merged region set list, it means that the merge process has already been performed and no further processing is required. Set e=e+1 and execute S11. If it does not exist, execute S13. The initial value of the current merged region set list is empty.

[0049] S13: Based on B, obtain associated regions that conflict with associated regions corresponding to the e-th initialization region set. If conflicting associated regions are obtained, use the obtained associated regions as conflicting regions for the e-th initialization region set and execute S14. If no conflicting associated regions are obtained, add the e initialization region sets to the current merged region set list, set e=e+1, and execute S11.

[0050] In this embodiment of the present invention, if L exists in B es3 =1 and e≠s3, it means that there is an associated area that conflicts with the associated area corresponding to the e-th initialized area set, namely the s3-th associated area, and the value of s3 ranges from 1 to M.

[0051] S14, if the position number corresponding to the conflicting area of ​​the e-th initialization area set exists in the current merged area set list, merge the e-th initialization area set with the merged area set to which the corresponding conflicting area belongs, that is, add the elements in the e-th initialization area set to the merged area set to which the corresponding conflicting area belongs; otherwise, that is, if the position number corresponding to the conflicting area of ​​the e-th initialization area set does not exist in the current merged area set list, merge the e-th initialization area set with the initialization area set corresponding to the corresponding associated area, and add the merged merged area set to the current merged area set list, set e=e+1, and execute S11.

[0052] S15 , obtaining the initial conflicting region group set based on the merged region sets in the current merged region set list, that is, taking a merged region set in the current merged region set list as a conflicting region group.

[0053] In another embodiment of the present invention, the initial conflict area group set may also be obtained based on the aircraft stand distribution map of the target geographical area. Specifically, associated areas belonging to the same taxiway and within a set range may be regarded as a conflict area group.

[0054] S2: If the current conflicting region group set is not empty, select a conflicting region group from the current conflicting region group set as the current region group to be processed, and delete the current region group to be processed from the current conflicting region group set, wherein the initial value of the current conflicting region group set is the initial conflicting region group set.

[0055] Furthermore, S2 specifically includes:

[0056] If at least one first conflicting region group exists in the current conflicting region group set, the first conflicting region group having the smallest association sum value and the largest region selection weight sum value is selected from the current conflicting region group set as the current region group to be processed. That is, if the current conflicting region group set includes multiple first conflicting region groups having the smallest association sum values, the first associated region group having the largest region selection weight sum value is selected as the current region group to be processed.

[0057] If only at least one second conflicting region group exists in the current conflicting region group set, that is, the current conflicting region group set does not contain the first conflicting region group but only contains the second conflicting region group, then the second conflicting region group with the smallest association sum value and the largest region selection weight sum value is selected from the current conflicting region group set as the current region group to be processed. In other words, if the current conflicting region group set includes multiple second conflicting region groups with the smallest association sum values, then the second associated region group with the largest region selection weight sum value is selected as the current region group to be processed.

[0058] The associated areas in the first conflict area group have a first area identifier, for example, the first conflict area group is a close camera group, and the associated areas in the second conflict area group have a second area identifier, for example, the second conflict area group is a far camera group.

[0059] In this embodiment of the present invention, the association sum value of a conflicting region group is equal to the sum of the association values ​​corresponding to all associated regions within the region group. The association value of an associated region is used to represent the number of associated objects actually associated with the associated region, that is, the number of associated objects allowed to be associated with the associated region. The region selection weight sum value of a conflicting region group is equal to the sum of the region selection weights of all associated objects corresponding to the conflicting region group. The region selection weight of an associated object is used to represent the degree of preference of the associated object for the associated region.

[0060] In the embodiment of the present invention, when selecting the conflicting area group to be processed, the close stand group with a small number of assignable flights and the largest selection weight is preferentially selected, thereby improving stand allocation efficiency and soft preference.

[0061] S3: Generate a network graph corresponding to the current group of regions to be processed as the current group of regions to be processed. Further, in S3, based on the feasible associated operation constraint information and the associated feasibility matrix corresponding to the current group of regions to be processed, generate a network graph corresponding to the current group of regions to be processed.

[0062] In an embodiment of the present invention, the feasible associated operation constraint information is used to constrain the associated operation to be an associated operation that is allowed to be executed. The feasible associated operation constraint information includes a first constraint condition, a second constraint condition, and a third constraint condition.

[0063] In the embodiment of the present invention, the first constraint condition is used to constrain each associated object to be associated with at most one associated area. The first constraint condition can be expressed as: M k=1 x uk ≤1, where x uk is the decision variable indicating whether to associate the u-th associated object with the k-th associated region, x uk The value of is the first association identifier or the second association identifier, the value of u is 1 to N, and the value of k is 1 to M.

[0064] In an embodiment of the present invention, the second constraint condition is used to constrain the existence of a minimum buffer time interval between two adjacent associated objects associated with the same associated area. That is, the time interval between the arrival time of the latter flight and the departure time of the previous flight of two adjacent flights assigned to the same parking position should be greater than the minimum buffer time interval to ensure that there is no conflict between the flights.

[0065] The second constraint can be expressed as: uk +x vk +Q uv ≤2. Where x vk is the decision variable indicating whether to associate the vth associated object with the kth associated region, x vk The value of Q is the first association identifier or the second association identifier. uv is the time overlap conflict identifier between the u-th associated object and the v-th associated object, if (t u a -t v d -ρ)(t v a -t d u -ρ) ≥ 0, then Q uv =1, indicating that there is a time overlap conflict between the u-th associated object and the v-th associated object, otherwise, that is, (t u a -t v d -ρ)(t v a -t d u -ρ)<0,Q uv =0, indicating that there is no time overlap conflict between the u-th associated object and the v-th associated object. u a is the planned arrival time of the u-th associated object, t u d is the planned departure time of the u-th associated object, tv a is the planned arrival time of the vth associated object, t v d is the planned departure time of the vth associated object, ρ is the minimum buffer time interval, which can be determined based on the actual situation of the target geographical area, and the value of v ranges from 1 to N.

[0066] In this embodiment of the present invention, the third constraint is used to ensure that no collision occurs between two associated objects associated with two adjacent associated areas. Airport ground traffic conditions are complex and unpredictable. When flights at adjacent parking stands arrive and depart close together, collisions are more likely to occur on the same taxiway. Therefore, collisions between flights at adjacent parking stands should be avoided.

[0067] The third constraint can be expressed as: uk +x vg +H uv +L kg ≤3. Among them, x vg is the decision variable indicating whether to associate the vth associated object with the gth associated region, x vg The value of is the first association identifier or the second association identifier, and the value of g is 1 to M. uv The collision identification value between the u-th associated object and the v-th associated object. If min(|t u a -t v a |,|t u d -t v d |,|t u a -t v d |,|t u d -t v a |)≤τ,H uv =1, indicating that there is a collision between the u-th associated object and the v-th associated object, otherwise, H uv =0, indicating that there is no collision between the uth associated object and the vth associated object. min() represents the minimum value, and || represents the absolute value operation. τ is the preset time interval, which can be determined based on the actual situation of the target geographical area. kg is the conflict identification value between the kth associated area and the gth associated area, which can be obtained based on the associated area conflict relationship matrix.

[0068] The association feasibility matrix corresponding to the current group of regions to be processed is determined based on the current association feasibility matrix. That is, the association relationships between the associated regions and associated objects corresponding to the current group of regions to be processed are extracted from the current association feasibility matrix to form the association feasibility matrix corresponding to the group of regions to be processed. The initial value of the current association feasibility matrix is ​​the initial association feasibility matrix.

[0069] In this embodiment of the present invention, a node in a network diagram represents an association operation between an association region and an association object. The number of nodes in the currently processed network diagram is equal to the number of first association identifiers in the corresponding association feasibility matrix. That is, the association relationship corresponding to one first association identifier is one association operation. The node name can be the identifier of the corresponding association operation, and the identifier of the association operation can be composed of the position number of the corresponding association region and the ID of the association object.

[0070] In addition, a connecting line is used to connect the two nodes corresponding to two conflicting association operations in the network graph. Specifically, if the association operations corresponding to the two nodes do not satisfy the feasible association operation constraint information, that is, do not satisfy any of the first to third constraints, it indicates that the corresponding association operations of the two nodes are in a conflicting relationship. A connecting line is used to connect the two nodes, that is, an edge exists. Otherwise, no connecting line is used to connect the two nodes, that is, no edge exists. The connecting line is an undirected connecting line, that is, the network graph is an undirected graph.

[0071] S4, based on the current network graph to be processed and the pre-trained graph neural network model, obtain the mapping value corresponding to each node in the network graph, obtain the node mapping value set corresponding to the network graph, and based on the node mapping value set corresponding to the network graph and the connection relationship between the nodes, obtain the maximum weighted independent node set corresponding to the network graph, and use it as the association result of the current conflict area group to be processed; wherein, there is no connecting line between any two nodes in the maximum weighted independent node set.

[0072] Furthermore, S4 may specifically include:

[0073] S401, input the node feature matrix and adjacency matrix corresponding to the current network graph to be processed into a pre-trained graph neural network model to obtain the mapping value of each node in the current network graph to be processed.

[0074] In this embodiment of the present invention, the mapping value of each node is used to represent the predicted probability that the node belongs to the maximum weighted independent set. The mapping value can range from (0 to 1). The closer the mapping value of a node is to 1, the more likely it is that the node belongs to the maximum weighted independent node. The closer the mapping value is to 0, the less likely it is that the node belongs to the maximum weighted independent node.

[0075] In an embodiment of the present invention, the node feature matrix corresponding to the current network graph to be processed is composed of the initial feature vectors of all corresponding nodes. The initial feature vector of each node may include the number of associated regions that the associated object allows to be associated, i.e., the number of aircraft parking spaces at which the corresponding flight can dock; the number of associated objects that the associated region allows to be associated, i.e., the number of flights that can dock at the corresponding aircraft spaces; the weight of the node; the degree of the node; the number of two-hop neighbors of the node; and the average degree of the one-hop neighbors of the node.

[0076] In this embodiment of the present invention, the degree of a node refers to the number of edges directly connected to the node. The weight of a node is the region selection weight of the corresponding associated object. The average degree of a node's one-hop neighbors is the average degree of the node's directly connected neighbor nodes, which is equal to the average degree of all one-hop neighbors of the node.

[0077] In the embodiment of the present invention, the initial feature vector of a node comprehensively captures the basic attributes, local structural information and connection characteristics of the node.

[0078] In an embodiment of the present invention, the adjacency matrix corresponding to the current network graph to be processed is used to describe the adjacency relationship between nodes in the current network graph to be processed. If there is an edge between two nodes, the corresponding element value in the adjacency matrix is ​​1, otherwise, it is 0.

[0079] S402 , sorting all nodes corresponding to the current network graph to be processed in descending order of mapping values ​​to obtain a sorted node set as an initial sorted node set.

[0080] S403, select the node with the maximum mapping value from the current sorted node set as the current candidate node. If there is no connection line between the current candidate node and any node in the current maximum weighted independent node set, execute S404; otherwise, execute S405; the initial state of the current maximum weighted independent node set is empty; the initial value of the current sorted node set is the initial sorted node set.

[0081] S404: Add the current candidate node to the current maximum weighted independent node set, and delete the current candidate node and its corresponding one-hop neighbor from the current sorted node set; if the current sorted node set is not an empty set, execute S403; if the current sorted node set is an empty set, execute S406.

[0082] S405: Delete the current candidate node from the current sorting node set. If the current sorting node set is not an empty set, execute S403; if the current sorting node set is an empty set, execute S406.

[0083] S406: Taking the current maximum weighted independent node set as the association result corresponding to the current area group to be processed, a parking space allocation plan corresponding to the current area group to be processed is obtained, and then executing S5.

[0084] S5: Based on the association result corresponding to the current group of regions to be processed, the current association feasibility matrix is ​​updated, and S2 is executed.

[0085] In an embodiment of the present invention, the current association feasibility matrix can be specifically updated based on the feasible association operation constraint information, that is, in the current association feasibility matrix, the association object corresponding to the association result corresponding to the current group of regions to be processed is only allowed to be associated with the corresponding association region, and the association identifiers of other association operations that have a conflicting relationship with the association result are all set to the second association identifier.

[0086] Those skilled in the art know that, in S2, if the current conflict area group set is empty, it means that the association operation between all associated objects and associated areas has been completed, that is, all conflict area groups have obtained association results, then the subsequent steps are not executed, the association results corresponding to all conflict area groups are directly obtained, and the current control program is exited.

[0087] Furthermore, the method provided by the embodiment of the present invention further includes: visually displaying the association results of all conflicting area groups, that is, displaying the flights allocated to each parking stand.

[0088] Furthermore, in an embodiment of the present invention, the graph neural network model may be any graph neural network model structure in the prior art, which will not be described in detail here. Preferably, the graph neural network model may include a node embedding layer, a graph convolutional network layer, a first graph attention network layer, a second graph attention network layer, and a fully connected output layer connected in sequence.

[0089] In an embodiment of the present invention, the node embedding layer is used to map the initial feature vector of each node into a feature vector of a first predetermined dimension, thereby obtaining a corresponding node embedding feature. The feature vector of the first predetermined dimension may be, for example, a 64-dimensional vector. The node embedding feature can enrich the feature representation of the node and facilitate the capture of the structural characteristics of the nodes in the graph.

[0090] Specifically, the node embedding layer maps the received initial feature vector through a fully connected layer, and the dimension is changed from the original feature dimension d in Mapped to a 64-dimensional vector. The specific mapping process can be expressed as:

[0091] .

[0092] Among them, z iis the initial feature vector of the i-th node in the current network graph to be processed, i ranges from 1 to H, and H is the number of nodes in the current network graph to be processed. （0） is the weight matrix of the node embedding layer, b （0） is the bias vector of the node embedding layer. In this embodiment of the present invention, W （0） For size d in ×64 matrix, b （0） is a 64-dimensional vector, σ() is the ReLU activation function, For z i The corresponding node embedding features.

[0093] Furthermore, in an embodiment of the present invention, the graph convolutional network layer is used to map node embedding features into hidden features of a second set dimension as graph convolution features to learn the local topological structure of the graph. In an embodiment of the present invention, the second set dimension may be 32 dimensions.

[0094] Specifically, the graph convolution network layer is used to update the node's own features based on the symmetric normalization aggregation rule of the graph convolution and the features of the neighboring nodes. In the embodiment of the present invention, the graph convolution feature of the i-th node obtained by the graph convolution network layer is The following conditions must be met:

[0095] .

[0096] Where N(i) represents the number of one-hop neighbors of the i-th node in the current network graph to be processed, NC(i) is the number of nodes in the merged node set formed by the i-th node and its one-hop neighbors, that is, NC(i) = N(i) + 1, the value of j ranges from 1 to NC(i), and N(j) is the number of one-hop neighbors of the j-th node. W （1） is the weight matrix of the graph convolutional network layer, b （1） is the bias vector of the graph convolutional network layer. In this embodiment of the present invention, W （1） is a matrix of size 64×32, b （1） is a 32-dimensional vector. Node embedding feature for the jth node in the merged node set.

[0097] Furthermore, in an embodiment of the present invention, the first graph attention network layer is used to calculate the attention weights between a node and its neighboring nodes through a multi-head attention mechanism, and concatenate the outputs of each attention head to obtain node features of a third set dimension as the first attention feature. The third set dimension can be 128 dimensions.

[0098] Specifically, the first attention feature of the i-th node obtained by the first graph attention network layer is The following conditions must be met:

[0099]

[0100] Among them, concat() represents the splicing operation, represents the dth attention head pair passing through the first graph attention network layer Execute the attention mechanism, d ranges from 1 to E1, where E1 is the number of attention heads in the first graph attention network layer. The following conditions must be met:

[0101] .

[0102] in, is the normalized attention weight of the d-th attention head obtained based on the graph convolution features of the i-th node and the j1-th node, which can be obtained based on the existing method. is the weight matrix of the d-th attention head of the first graph attention network layer, is the graph convolution feature of the j1th node obtained by the graph convolutional network layer, and the value of j1 ranges from 1 to N(i).

[0103] In the embodiment of the present invention, E1=4.

[0104] Furthermore, in an embodiment of the present invention, the second graph attention network layer is used to adopt the graph attention mechanism, and close the splicing operation, and adopt the mean or weighted sum method to reduce the node features of the third set dimension to the node features of the second set dimension as the second attention feature. Specifically, the second attention feature of the i-th node obtained by the second graph attention network layer is The following conditions must be met:

[0105] .

[0106] in, represents the c-th attention head pair passing through the second graph attention network layer Execute the graph attention mechanism, where c ranges from 1 to E2, where E2 is the number of attention heads in the second graph attention network layer. In this embodiment of the present invention, E2=2.

[0107] In an embodiment of the present invention, the fully connected output layer is used to map the node features of the second set dimension to a 1-dimensional output through the fully connected layer, indicating the probability that the node belongs to the maximum weighted independent node set, thereby completing the node prediction task. Specifically, the output features of the fully connected output layer satisfy the following formula:

[0108] .

[0109] Among them, p iis the mapping value of the i-th node obtained by the fully connected output layer. Sigmoid() is an activation function used to map the output to a mapping value between 0 and 1. （3） is the weight matrix of the fully connected output layer, b （3） is the bias vector of the fully connected output layer, W （3） is a matrix of size 32×1.

[0110] In an embodiment of the present invention, a pre-trained graph neural network model can be trained using a supervised learning method. The training samples can be derived from a data set of historical records and labels generated based on the solution results of a commercial optimization solver (such as a Gurobi solver). Specifically, the training sample can be a network graph including node features and corresponding node labels. The node label indicates whether the node belongs to the maximum weighted independent node set of the corresponding network graph, where a node label of 1 indicates that the node belongs to the maximum weighted independent node set, and a node label of 0 indicates that the node does not belong to the set. The nodes of the maximum weighted independent node set usually come from associated operations that have been performed in history, such as nodes that have successfully allocated seats for flights. Using this labeled sample data, the pre-built original graph neural network model can be trained so that the model learns the mapping relationship between node features and the independent set to which they belong. During the training process, the cross entropy loss function can be used as the optimization objective function.

[0111] Those skilled in the art know that any method of training a pre-built original graph neural network model using a sample network graph marked with node features and node labels falls within the scope of protection of the present invention.

[0112] In the embodiment of the present invention, since a pre-trained graph neural network model is used to obtain the maximum weighted independent node set of the network graph, the computing efficiency can be improved and the bridging rate can be maximized.

[0113] An embodiment of the present invention also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are configured to execute the method described in the embodiment of the present invention.

[0114] An embodiment of the present invention further provides a computer-readable storage medium storing computer-executable instructions, wherein the computer instructions are used to execute the method described in the embodiment of the present invention.

[0115] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in the present invention can be achieved. This is not limited herein.

[0116] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A data processing method for associating flights and parking spaces, characterized in that: The method comprises the following steps: S1. Obtain an initial conflict region group set and an initial association feasibility matrix corresponding to a target geographic region. The initial conflict region group set includes multiple conflict region groups that do not conflict with each other. The initial conflict region group set is obtained based on the association region conflict relationship matrix of the target geographic region. The initial association feasibility matrix is ​​used to indicate whether associations can be established between association regions and association objects. The association regions are parking stands, and the association objects are flights. S2, if the current conflicting region group set is not empty, selecting a conflicting region group from the current conflicting region group set as the current region group to be processed, and deleting the current region group to be processed from the current conflicting region group set; the initial value of the current conflicting region group set is the initial conflicting region group set; S3, generating a network graph corresponding to the current group of regions to be processed as the current network graph to be processed; the nodes in the network graph represent the association operations between the associated regions and the associated objects, and the two nodes corresponding to the two conflicting association operations are connected by a connecting line; S4, based on the current network graph to be processed and a pre-trained graph neural network model, obtain the mapping value corresponding to each node in the network graph, obtain the node mapping value set corresponding to the network graph, and based on the node mapping value set corresponding to the network graph and the connection relationship between the nodes, obtain the maximum weighted independent node set corresponding to the network graph, and use it as the association result of the current conflict area group to be processed; wherein the mapping value of each node is used to represent the predicted probability that the node belongs to the maximum weighted independent set, and there is no connecting line between any two nodes in the maximum weighted independent node set; S5, based on the association result corresponding to the current group of regions to be processed, the current association feasibility matrix is ​​updated, and S2 is executed; the initial value of the current association feasibility matrix is ​​the initial association feasibility matrix.

2. The method according to claim 1, characterized in that S2 specifically includes: If there is at least one first conflicting region group in the current conflicting region group set, selecting the first conflicting region group having the smallest association sum value and the largest region selection weight sum value from the current conflicting region group set as the current region group to be processed; If there is only at least one second conflicting region group in the current conflicting region group set, selecting the second conflicting region group having the smallest association sum value and the largest region selection weight sum value from the current conflicting region group set as the current region group to be processed; The associated areas in the first conflicting area group have a first area identifier, and the associated areas in the second conflicting area group have a second area identifier. The association sum value of the conflicting area group is equal to the sum of the association values ​​corresponding to all the associated areas in the area group. The association value of the associated area is used to represent the number of associated objects allowed to be associated with the associated area. The area selection weight sum value of the conflicting area group is equal to the sum of the area selection weights of all associated objects allowed to be associated with the conflicting area group. The area selection weight of the associated object is used to represent the degree of preference of the associated object for the associated area.

3. The method according to claim 1, characterized in that In S3, based on the feasible association operation constraint information and the association feasibility matrix corresponding to the current group of regions to be processed, a network diagram corresponding to the current group of regions to be processed is generated; wherein the association feasibility matrix corresponding to the current group of regions to be processed is determined based on the current association feasibility matrix; the feasible association operation constraint information includes a first constraint condition, a second constraint condition, and a third constraint condition, wherein the first constraint condition is used to constrain each association object to be associated with at most one association region, the second constraint condition is used to constrain the existence of a minimum buffer time interval between two adjacent association objects associated with the same association region, and the third constraint condition is used to constrain the existence of no collision conflict between two association objects respectively associated with two adjacent association regions.

4. The method according to claim 1, wherein S4 specifically includes: S401: Input the node feature matrix and adjacency matrix corresponding to the current network graph to be processed into a pre-trained graph neural network model to obtain the mapping value of each node in the current network graph to be processed; S402, sorting all nodes corresponding to the current network graph to be processed in descending order of mapping values ​​to obtain a sorted node set as the initial sorted node set; S403: Select the node with the maximum mapping value from the current sorted node set as the current candidate node. If there is no connection line between the current candidate node and any node in the current maximum weighted independent node set, execute S404; otherwise, execute S405; the initial state of the current maximum weighted independent node set is empty; the initial value of the current sorted node set is the initial sorted node set; S404: Add the current candidate node to the current maximum weighted independent node set, and delete the current candidate node and its corresponding one-hop neighbor from the current sorted node set. If the current sorted node set is not empty, execute S403; if the current sorted node set is empty, execute S406. S405: Delete the current candidate node from the current sorting node set. If the current sorting node set is not an empty set, execute S403; if the current sorting node set is an empty set, execute S406. S406: Use the current maximum weighted independent node set as the association result corresponding to the current area group to be processed, and execute S5.

5. The method according to claim 4, characterized in that The node feature matrix corresponding to the current network graph to be processed is formed by the initial feature vectors of all nodes. The initial feature vector of each node includes the number of associated regions that the associated object is allowed to associate with, the number of associated objects that the associated region is allowed to associate with, the weight of the node, the degree of the node, the number of two-hop neighbors of the node, and the average degree of the one-hop neighbors of the node. The weight of the node is the area selection weight of the corresponding associated object.

6. The method according to claim 1, characterized in that The initial association feasibility matrix is ​​determined based on association constraint information, where the association constraint information is used to constrain the carrier model and object attribute information of the associated object to match the region model and region attribute information of the associated region, respectively.

7. The method according to claim 1, characterized in that The initial conflict region group set is obtained by the following steps: S10, based on the M associated areas in the target geographical area, obtain M initialization area sets; the elements in each initialization area set are the position numbers of the corresponding associated areas; set the counter e=1; S11, if e≤M, execute S12, otherwise, execute S15; S12: If the position number corresponding to the e-th initialized region set exists in the current merged region set list, set e=e+1 and execute S11; if not, execute S13; the initial value of the current merged region set list is empty; S13, based on the associated area conflict relationship matrix of the target geographic area, obtaining associated areas that conflict with the associated areas corresponding to the e-th initialization area set; if conflicting associated areas are obtained, the obtained associated areas are used as conflicting areas of the e-th initialization area set, and S14 is executed; if no conflicting associated areas are obtained, the e initialization area sets are added to the current merged area set list, e=e+1 is set, and S11 is executed; S14: If the position number corresponding to the conflicting area of ​​the e-th initialization area set exists in the current merged area set list, merge the e-th initialization area set with the merged area set to which the conflicting area belongs; otherwise, merge the e-th initialization area set with the initialization area set corresponding to the associated area, add the merged area set obtained by the merger to the current merged area set list, set e=e+1, and execute S11; S15: Obtain the initial conflicting region group set based on the merged region set in the current merged region set list.

8. An electronic device, characterized in that: The method comprises a processor and a memory; the processor is used to execute the steps of the method according to any one of claims 1 to 7 by calling the program or instructions stored in the memory.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store a program or instruction, and the program or instruction enables a computer to execute the steps of the method according to any one of claims 1 to 7.

Images