Traffic prediction method, device, storage medium and electronic device for unopened flight segments

Abstract

The present application discloses a method, device, storage medium and electronic device for predicting the flow of unopened flight segments. The method comprises: constructing a flight segment network relationship diagram based on navigation flow data, wherein the graph nodes of the flight segment network relationship diagram include at least one of the following: opened flight segments, unopened flight segments, and the connection relationship between the graph nodes represents airport information, wherein the flight segment network relationship diagram includes at least one of the following: flight segment network topology diagram, flight segment network association diagram; inputting the flight segment network relationship diagram into the target graph attention network to obtain the flight segment embedded feature information; inputting the flight segment embedded feature information into the long short-term memory network to obtain the flight segment time feature information; inputting the flight segment time feature information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment. Through the present application, the problem that the flow of undeveloped flight segments cannot be accurately predicted in the related technology is solved.

Description

Background Art

[0002] In the related art, air traffic flow prediction methods can be divided into two categories: model-driven methods and data-driven methods. Model-driven methods refer to methods that build models, such as gravity models and global vector autoregressive models (GVAR). Data-driven methods refer to the use of machine learning to predict air traffic flow through time series data. Data-driven methods have the ability to decipher complex patterns and relationships, and can model complex air traffic systems more flexibly.

[0003] At present, many extended machine learning methods have been proposed, including tree-structured spatiotemporal neural network (TS-STNN), convolutional neural network-long short-term memory (CNN-LSTM), spatiotemporal graph convolutional network (STGCN), etc. These methods have promoted the development of machine learning algorithms in a data-driven way. However, they generally divide the data set from the time dimension and require historical data of all nodes on the graph structure. At the same time, multi-graph structures usually involve related parameter allocation problems, but it is very difficult to manually allocate the optimal parameter combination.

[0004] That is, the existing deep learning method of spatiotemporal integrated graph is a method based on historical data. Due to the lack of historical data, the deep learning method of spatiotemporal integrated graph cannot be directly used to solve the traffic prediction problem of new flight segments.

[0005] There is currently no effective solution to the problem that related technologies cannot accurately predict traffic on undeveloped flight segments. Summary of the invention

[0006] The main purpose of this application is to provide a method, device, storage medium and electronic device for predicting the traffic flow of unopened flight segments, so as to solve the problem in related technologies that it is impossible to accurately predict the traffic flow of undeveloped flight segments.

[0007] In order to achieve the above-mentioned purpose, according to the first aspect of the present application, a method for predicting the flow of unopened flight segments is provided. The method comprises: constructing a flight segment network relationship graph based on the navigation flow data, wherein the graph nodes of the flight segment network relationship graph include at least one of the following: opened flight segments, unopened flight segments, and the connection relationship between the graph nodes represents airport information, wherein the flight segment network relationship graph includes at least one of the following: a flight segment network topology graph, a flight segment network association graph; inputting the flight segment network relationship graph into the target graph attention network to obtain the flight segment embedded feature information; inputting the flight segment embedded feature information into the long short-term memory network to obtain the flight segment time feature information; inputting the flight segment time feature information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment.

[0008] Furthermore, a flight segment network relationship diagram is constructed based on the navigation flow data, including: determining the departure airport information and the destination airport information in the navigation flow data; and constructing a flight segment network relationship diagram based on the flight segment information between the departure airport and the destination airport, wherein at least one of the flight segments between the departure airport and the destination airport has been opened.

[0009] Furthermore, a flight segment network relationship diagram is constructed based on the flight segment information between the departure airport and the destination airport, including: obtaining initial characteristic information of the departure airport information and the destination airport information, wherein the initial characteristic information includes at least one of the following: inter-city distance, economic indicators, airport passenger throughput, population size, and competition factors; and constructing a flight segment network relationship based on the initial characteristic information.

[0010] Furthermore, the flight segment network relationship is a flight segment network association graph, and the flight segment network relationship is constructed according to the initial feature information, including: normalizing the initial feature information to obtain a normalized feature vector, and obtaining a feature dimension of the initial feature information; determining similarity information according to the feature vector and the feature dimension, wherein the similarity information is used to indicate the accuracy of the flight segment network relationship; and determining the flight segment network association graph according to the similarity information.

[0011] Furthermore, the flight segment network relationship graph is input into the target graph attention network to obtain the flight segment embedding feature information, including: determining the node feature set and attention coefficient of the flight segment network relationship graph; inputting the node feature set into the target graph attention network to obtain the flight segment embedding feature information.

[0012] Furthermore, the flight segment embedding feature information is input into the long short-term memory network to obtain the flight segment time feature information, including: selecting historical flight segment embedding feature information of multiple historical moments from the flight segment embedding feature information; inputting the historical flight segment embedding feature information into the long short-term memory network to obtain the flight segment time feature information.

[0013] Furthermore, the flight segment time characteristic information is input into a non-dominated sorting genetic algorithm to obtain flow information of unopened flight segments, including: determining an initial population according to the flight segment time characteristic information, wherein the initial population includes multiple flight segment flow information solutions; processing the initial population according to a preset operation through the non-dominated sorting genetic algorithm to obtain a target population, wherein the target population is the optimal solution among multiple flight segment flow information solutions; and determining the flow information of the unopened flight segments according to the target population.

[0014] In order to achieve the above-mentioned purpose, according to the second aspect of the present application, a flow prediction device for an unopened flight segment is provided. The device includes: a construction unit, which is used to construct a flight segment network relationship graph based on navigation flow data, wherein the graph nodes of the flight segment network relationship graph include at least one of the following: opened flight segments, unopened flight segments, and the connection relationship between the graph nodes represents airport information, wherein the flight segment network relationship graph includes at least one of the following: a flight segment network topology graph, a flight segment network association graph; a first input unit, which is used to input the flight segment network relationship graph into the target graph attention network to obtain the flight segment embedded feature information; a second input unit, which is used to input the flight segment embedded feature information into the long short-term memory network to obtain the flight segment time feature information; a third input unit, which is used to input the flight segment time feature information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment.

[0015] Furthermore, the construction unit includes: a first determination subunit, used to determine the departure airport information and the destination airport information in the flight traffic data; a construction subunit, used to construct a segment network relationship diagram based on the segment information between the departure airport and the destination airport, wherein at least one of the segments between the departure airport and the destination airport has been opened.

[0016] Furthermore, a sub-unit is constructed, including: an acquisition module for acquiring initial characteristic information of the departure airport information and the destination airport information, wherein the initial characteristic information includes at least one of the following: inter-city distance, economic indicators, airport passenger throughput, population size, and competition factors; a construction module for constructing a segment network relationship based on the initial characteristic information.

[0017] Furthermore, the flight segment network relationship is a flight segment network association graph, and the construction module includes: a processing submodule, which is used to normalize the initial feature information to obtain a normalized feature vector and obtain the feature dimension of the initial feature information; a first determination submodule, which is used to determine similarity information based on the feature vector and the feature dimension, wherein the similarity information is used to represent the accuracy of the flight segment network relationship; and a second determination submodule, which is used to determine the flight segment network association graph based on the similarity information.

[0018] Furthermore, the first input unit includes: a second determination subunit, used to determine the node feature set and attention coefficient of the flight segment network relationship graph; a first input subunit, used to input the node feature set into the target graph attention network to obtain the flight segment embedded feature information.

[0019] Furthermore, the second input unit includes: an acquisition subunit, which is used to select historical flight segment embedding feature information of multiple historical moments from the flight segment embedding feature information; a second input subunit, which is used to input the historical flight segment embedding feature information into the long short-term memory network to obtain the flight segment time feature information.

[0020] Furthermore, the third input unit includes: a third determination subunit, used to determine the initial population according to the time characteristic information of the flight segment, wherein the initial population includes multiple flight segment flow information solutions; a processing subunit, used to process the initial population according to preset operations through a non-dominated sorting genetic algorithm to obtain a target population, wherein the target population is the optimal solution among the multiple flight segment flow information solutions; a fourth determination subunit, used to determine the flow information of the unopened flight segment according to the target population.

[0021] According to a third aspect of an embodiment of the present application, there is provided an electronic device, comprising: a processor; and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned methods for predicting traffic flow for unopened flight segments is implemented.

[0022] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for predicting traffic flow for an unopened flight segment according to any one of the above items is implemented.

[0023] Through this application, the following steps are adopted: constructing a segment network relationship graph based on the navigation flow data, wherein the graph nodes of the segment network relationship graph include at least one of the following: opened segments, unopened segments, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship graph includes at least one of the following: segment network topology graph, segment network association graph; inputting the segment network relationship graph into the target graph attention network to obtain segment embedded feature information; inputting the segment embedded feature information into the long short-term memory network to obtain segment time feature information; inputting the segment time feature information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened segment. Through this application, the problem that the flow of undeveloped segments cannot be accurately predicted in the related art is solved. By constructing the segment network relationship graph, the segment network relationship graph is analyzed and processed step by step through the target graph attention network, the long short-term memory network and the non-dominated sorting genetic algorithm, thereby achieving the effect of accurately predicting the flow of undeveloped segments. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0025] Figure 1 It is a flow chart of a method for predicting traffic flow of an unopened flight segment provided in an embodiment of the present application;

[0026] Figure 2 is a topological diagram of an aviation network provided according to an embodiment of the present application;

[0027] Figure 3 is a schematic diagram of a time step provided according to an embodiment of the present application;

[0028] Figure 4 is a schematic diagram of segmentation of a data set provided according to an embodiment of the present application;

[0029] Figure 5 is a schematic diagram of individual coding provided according to an embodiment of the present application;

[0030] Figure 6 is a schematic diagram of a population iteration process provided according to an embodiment of the present application;

[0031] Figure 7 It is a schematic diagram of the MGAT-LSTM-NSGA-Ⅱ model framework provided according to an embodiment of the present application;

[0032] Figure 8 is a schematic diagram of a flow prediction device for an unopened flight segment provided according to an embodiment of the present application;

[0033] Fig. 9 It is a schematic diagram of the network architecture of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0035] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0036] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0037] According to an embodiment of the present application, a method for predicting traffic flow of an unopened flight segment is provided.

[0038] Figure 1 FIG. 1 is a flow chart of a method for predicting traffic flow of an unopened flight segment according to an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:

[0039] Step S101, constructing a segment network relationship diagram based on the navigation flow data, wherein the graph nodes of the segment network relationship diagram include at least one of the following: opened segments, unopened segments, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship diagram includes at least one of the following: a segment network topology diagram, a segment network association diagram.

[0040] Among them, the navigation flow data can be a data set of ships, aircraft or other navigation tools moving along a specific route (segment) within a certain period of time, which includes the starting point, end point, segments passed, navigation time, speed, load capacity and other information of the navigation tool. This application takes aircraft navigation as an example for explanation. Among them, the network relationship diagram is a graphical representation method used to describe the connection relationship between different entities (such as segments, nodes). The segment network topology diagram describes the physical connection relationship or structural layout between segments. The segment network association diagram focuses more on describing the association relationship between segments in terms of traffic, passenger demand, flight arrangements, etc.

[0041] In some embodiments, constructing a segment network relationship diagram based on navigation traffic data can be obtained through the following steps: determining the origin airport information and the destination airport information in the navigation traffic data; constructing a segment network relationship diagram based on the segment information between the origin airport and the destination airport, wherein at least one of the segments between the origin airport and the destination airport has been opened.

[0042] For example, each flight segment in the flight network is considered as a node, and each airport is considered as a connection between the flight segments. is the flight segment network topology diagram, where is a node set, , N is the total number of nodes in the graph structure, and E is the arc set. Figure 2 As shown, the node Indicates segment BD, which indicates an undeveloped segment. Segment BD is associated with segments BC, AD, and AB through airports B and D. Figure 2 As shown, point Indicates flight segment BD, which is not in the flight plan and is a new flight segment, that is, an unopened flight segment in this case. Flight segment BD is connected to flight segments BC and AB due to the relationship of airport B. , ; Segments BD and AD are connected due to airport D .

[0043] Since segment BD has not been developed yet, there is no historical demand data for this segment, so the airline cannot fit its future demand based on the historical demand data of segment BD. However, the historical demand for segments AD, AC, AB, and BC is known, and the characteristic information of all segments can also be obtained based on relevant statistical reports. This application uses appropriate technology combined with the historical demand of known segments to predict the future demand of new segment BD, in order to evaluate whether the new segment BD is worth developing, and can also promote optimization and improvement in many aspects, providing strong support for the sustainable development of airlines.

[0044] In some embodiments, a flight segment network relationship diagram is constructed based on the flight segment information between the departure airport and the destination airport, which can be obtained by the following steps: obtaining initial characteristic information of the departure airport information and the destination airport information, wherein the initial characteristic information includes at least one of the following: inter-city distance, economic indicators, airport passenger throughput, population size, and competition factors; and constructing a flight segment network relationship based on the initial characteristic information.

[0045] For example, the distance between the departure point and the destination ( ) indicates the nature of the route ( ), GDP per capita ( ) represents consumer characteristics, airport throughput ( ) represents the characteristics of local residents, and the population size ( ) reflects the market potential, competitive factors ( ) represents the saturation of the market. represents the feature of node i, which is defined by five features. Finally, the feature matrix of the graph is a matrix with N rows and 5 columns, denoted as .

[0046] ;

[0047] Since the nodes in the flight segment network diagram represent the flight segments with the origin airport and the determined airport, the characteristics of GDP per capita (economic indicator), airport throughput and population size are all determined by the relevant data of the origin and destination. The characteristic matrix of the flight segment network relationship structure diagram is a matrix with N rows and 5 columns, denoted as .

[0048] This application improves the accuracy of predicting subsequent undeveloped flight segments by jointly defining the flight segment network relationship diagram using the above five characteristics.

[0049] It should be noted that in order to define the adjacent relationship between flight segments, the higher the similarity, the stronger the adjacency of the flight segments. Figure 1 Together, we constructed a flight segment network relationship diagram, where the topology diagram is constructed based on the flight network in real life. Representation Node With Node The topological relationship between the two flight segments is shown in Figure 1, where symbols o and o' represent departure airports, and symbols d and d' represent destination airports. The topological graph represents the diversion and confluence relationship between routes. If the origins of the two flight segments are the same, there is a diversion effect on the traffic, and the weight is set to 1. If the destinations of the two flight segments are the same, there may be confluence, so the weight is set to 0.5. Otherwise, the weight is set to 0. The main implementation logic of the above is as follows:

[0050] .

[0051] Furthermore, the flight segment network relationship is a flight segment network association graph, and the flight segment network relationship is constructed according to the initial feature information, which can be obtained through the following steps: normalizing the initial feature information to obtain a normalized feature vector, and obtaining the feature dimension of the initial feature information; determining similarity information according to the feature vector and the feature dimension, wherein the similarity information is used to indicate the accuracy of the flight segment network relationship; and determining the flight segment network association graph according to the similarity information.

[0052] For example, in order to reduce the impact of differences in sample data in different feature dimensions, vectors with different features are normalized. As shown in the following formula, represents the normalized vector, is the minimum value of the characteristic dimension m, is the maximum value of the feature dimension m. M is the set of feature dimensions, and the normalized feature vector formula is: make Representation Node With Node At the similarity of feature dimension m, the Gaussian kernel calculates the initial similarity, where the initial similarity formula is as follows:

[0053] ;

[0054] in, Used to measure the difference between two points. Represents the similarity coefficient at m dimension, affecting

[0055] Similarity value between flight segments. When the similarity coefficient is large enough, the similarity always tends to 1. Therefore, it is necessary to determine the node and nodes The comprehensive similarity between

[0056] as follows: ;

[0057] As shown in the above formula, different dimensions m The initial similarity Multiplying them together gives the comprehensive similarity. When the similarity approaches 1, the impact on the comprehensive similarity is small. Therefore, this formula can eliminate the false high similarity caused by too many similarity coefficients. If the similarity on a certain feature dimension has little impact on the learning results, we can increase the corresponding similarity coefficient to ignore the differences between the points on that feature dimension.

[0058] Finally, in order to make the comprehensive similarity of each node sufficiently different from other nodes, the comprehensive similarity is normalized. The formula of the normalized comprehensive similarity is as follows.

[0059] .

[0060] in, Indicate point i The smallest value among the comprehensive similarities with other points. Similarly, Indicate point i The point with the largest value of comprehensive similarity with other points is i The comprehensive similarity with each point is normalized to a value between 0 and 1 for subsequent model learning.

[0061] That is, the similarity information of this case is determined based on the initial similarity, the comprehensive similarity, and the normalized comprehensive similarity. The neighbor matrix of the flight segment network relationship graph is determined based on the similarity information, and the neighbor matrix is ​​defined as . Neighboring relationship The expression formula is as follows:

[0062] (8)

[0063] Among them, the point i The normalized comprehensive similarity with other points is multiplied by its corresponding weight. When there is a topological connection between nodes, the adjacent relationship is determined by the weight in the topological connection. The normalized comprehensive similarity in the associated connection When the similarity value between nodes is relatively large, although there is no topological connection, there is a chain connection between the nodes.

[0064] This application can effectively improve the accuracy of predicting subsequent undeveloped flight segments through the flight segment network relationship diagram obtained by the above definition.

[0065] Step S102: input the flight segment network relationship graph into the target graph attention network to obtain the flight segment embedding feature information.

[0066] Among them, the target graph attention network can be a multi-graph attention network (MGAT). After obtaining the relationship between nodes, this case constructs a flight segment network relationship connection diagram, and further explores the spatial graph information by proposing a multi-graph attention network (MGAT).

[0067] In some optional embodiments, the flight segment network relationship graph is input into the target graph attention network to obtain the flight segment embedded feature information, which can be obtained through the following steps: determining the node feature set and attention coefficient of the flight segment network relationship graph; inputting the node feature set into the target graph attention network to obtain the flight segment embedded feature information.

[0068] Exemplarily, the core component of MGAT is the graph attention layer, which defines the node features of layer l as a set, e.g., the node feature set , use it as the input of MGAT, and get the initial attention coefficient of the node , the coefficient The following formula example:

[0069] ;

[0070] Among them, the symbol is the attention mechanism, is a shared weight matrix used to transform the input

[0071] Convert to embed. express l Layer Node i The characteristic data of Indicates l Layer Node j feature data.

[0072] The attention coefficient can be not only the initial attention coefficient, but also the normalized coefficient of the initial attention coefficient. Obtained by the following formula:

[0073] ;

[0074] in, is an activation function, the softmax function is used to initialize the attention system

[0075] The numbers are normalized.

[0076] Therefore, the node feature set is input into the target graph attention network, and the propagation process of MGAT can be expressed by the following formula.

[0077]

[0078] in, express l+1 Layer Node i The characteristic data of Represents the activation function, using the ReLU function to confirm that the output value is greater than 0. For Node i The set of adjacent nodes of j Representation Node i The adjacent nodes of is the normalized attention coefficient, is the weight matrix, represents the feature data of node j in layer l. Finally, we get the embedded feature information of the flight segment , and take this embedded information as the output of the spatial block MGAT and record it as .

[0079] In this case, the node feature set is input into the target graph attention network, which can make full use of the graph structure information and the advantages of the attention mechanism, improve the expressive power and computational efficiency of the model, and thus achieve better performance in various application scenarios.

[0080] Step S103, inputting the flight segment embedding feature information into the long short-term memory network to obtain the flight segment time feature information.

[0081] Among them, the flight segment time feature information includes the features for predicting flight time and time trend analysis. By embedding the flight segment feature information into deep learning models such as LSTM, we can fully explore the time dependency and complex relationships in the flight segment data, thereby obtaining more accurate flight segment time feature information.

[0082] Specifically, the flight segment embedding feature information is input into the long short-term memory network to obtain the flight segment time feature information, which can be obtained through the following steps: selecting historical flight segment embedding feature information of multiple historical moments from the flight segment embedding feature information; inputting the historical flight segment embedding feature information into the long short-term memory network to obtain the flight segment time feature information.

[0083] For example, the output of the spatial module MGAT is used as the input of the temporal module LSTM (Long Short-Term Memory). This component is used to capture temporal information based on LSTM. Assume that there is a historical period with L moments of flight segment embedding feature information. Each time the MGAT block is used L times to obtain the corresponding spatial aggregation information, the input information of LSTM is obtained, which is expressed as Emb , .

[0084] Specific as Figure 3 As shown in Figure 1, the LSTM process is a rolling prediction process. If the time step is 3, the time window is used. embeddings in to train and predict time windows Different from the traditional traffic prediction model, this application not only divides the data set with the time dimension, but also divides the point set, where 80% of the points are used as training data and 20% of the points are used as test data. The points in the training data set are regarded as existing flight segments with historical flow information. Then the information points in the training data set are used in the time window Train the prediction model, specifically, embed the corresponding nodes in the time window As input, the streaming data used corresponds to the nodes of the training dataset in the time window is used to evaluate the training loss. Finally, using the time window The prediction model is tested with the information of each point in the test data set, and the flight time characteristic information of this case is obtained. In order to make the segmentation of the data set clearer, it can be Figure 4 As shown, 80% of the points are used as training data and 20% of the points are used as test data.

[0085] Step S104, inputting the flight segment time characteristic information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment.

[0086] Due to the interaction between similarity coefficients, it is difficult to manually select the appropriate coefficient combination. NSGA-Ⅱ (Non-dominated Sorting Genetic Algorithms-Ⅱ) is an algorithm that can randomly generate a population and find the optimal solution. Therefore, in order to make the results easy to interpret and further discuss the impact of each similarity coefficient, this case uses NSGA-Ⅱ to find the best coefficient combination of MGAT. The flight segment time characteristic information is input into the non-dominated sorting genetic algorithm, which can efficiently obtain the flow information of the unopened flight segments.

[0087] Specifically, the flight segment time characteristic information is input into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment, which can be obtained through the following steps: determining an initial population according to the flight segment time characteristic information, wherein the initial population includes multiple flight segment flow information solutions; processing the initial population according to a preset operation through the non-dominated sorting genetic algorithm to obtain a target population, wherein the target population is the optimal solution among multiple flight segment flow information solutions; and determining the flow information of the unopened flight segment according to the target population.

[0088] Exemplarily, the above steps can be divided into:

[0089] First, the initial population size of NSGA-Ⅱ is determined to be 5×P, and each individual is randomly generated using 0-1 coding, and each individual represents a coefficient combination. Figure 5 As shown, a gene segment corresponds to the corresponding coefficient in the flight segment time characteristic information, and a coefficient corresponds to the P gene segment. These coefficients are obtained through multiple flight segment flow information solutions. The coefficient combination is used as the input of MGAT, and the evaluation index is returned. That is, the initial population is processed according to the preset operation through the non-dominated sorting genetic algorithm to obtain the target population.

[0090] Secondly, the initial population is processed mainly according to fitness, and individuals are sorted first. In order to produce better offspring, the selection mechanism is set to the higher the fitness, the higher the probability of being selected. The selected individuals are considered to be the parents of the new individuals. The parent individuals are crossover and mutation operations are performed based on the random principle, and then the non-dominant sorting strategy is adopted to retain the original population and the new population. The size of the new population is consistent with the origin. The specific population iteration operations are as follows Figure 6 shown.

[0091] Finally, the above steps are repeated until the number of iterations reaches the maximum number i, that is, the traffic information of the unopened flight segments is predicted.

[0092] That is, the design of the NSGA-Ⅱ module is not only conducive to the automatic solution of key parameters, but also makes the MGAT-LSTM-NSGA-Ⅱ model more interpretable. Figure 7As shown, this application uses the characteristic information of its own segment and adjacent segments and the historical traffic data of existing segments to predict the future traffic of new segments. The MGAT-LSTM-NSGA-Ⅱ model is divided into three modules: MGAT, LSTM and NSGA-Ⅱ. First, a segment network diagram is constructed based on the characteristic relationship of the aviation network. Secondly, taking the segment network diagram as input, MGAT is used to aggregate the network information from the spatial dimension, and the LSTM model is used to predict the segment traffic from the time dimension. Finally, the NSGA-Ⅱ model is designed to optimize the key parameters in MGAT. NSGA-Ⅱ generates the parameters of the MGAT association graph, and MGAT-LSTM returns the corresponding fitness value to NSGA-Ⅱ.

[0093] In summary, the flow prediction method for unopened flight segments provided in the embodiment of the present application is to construct a flight segment network relationship diagram based on navigation flow data, wherein the graph nodes of the flight segment network relationship diagram include at least one of the following: opened flight segments, unopened flight segments, and the connection relationship between the graph nodes represents airport information, wherein the flight segment network relationship diagram includes at least one of the following: flight segment network topology diagram, flight segment network association diagram; the flight segment network relationship diagram is input into the target graph attention network to obtain the flight segment embedded feature information; the flight segment embedded feature information is input into the long short-term memory network to obtain the flight segment time feature information; the flight segment time feature information is input into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment. Through this application, the problem that the flow of undeveloped flight segments cannot be accurately predicted in the related art is solved. By constructing a flight segment network relationship diagram, the flight segment network relationship diagram is analyzed and processed step by step through the target graph attention network, the long short-term memory network and the non-dominated sorting genetic algorithm, thereby achieving the effect of accurately predicting the flow of undeveloped flight segments.

[0094] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0095] The embodiment of the present application also provides a flow prediction device for an unopened flight segment. It should be noted that the flow prediction device for an unopened flight segment in the embodiment of the present application can be used to execute the flow prediction method for an unopened flight segment provided in the embodiment of the present application. The flow prediction device for an unopened flight segment provided in the embodiment of the present application is introduced below.

[0096] Figure 8 FIG. 8 is a schematic diagram of a flow prediction device 800 for an unopened flight segment according to an embodiment of the present application. Figure 8As shown, the device includes: a construction unit 801, a first input unit 802, a second input unit 803, and a third input unit 804.

[0097] Specifically, the construction unit 801 is used to construct a segment network relationship graph based on the flight flow data, wherein the graph nodes of the segment network relationship graph include at least one of the following: an opened segment and an unopened segment, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship graph includes at least one of the following: a segment network topology graph and a segment network association graph;

[0098] The first input unit 802 is used to input the flight segment network relationship graph into the target graph attention network to obtain the flight segment embedding feature information;

[0099] The second input unit 803 is used to input the flight segment embedding feature information into the long short-term memory network to obtain the flight segment time feature information;

[0100] The third input unit 804 is used to input the flight segment time characteristic information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment.

[0101] The flow prediction device for unopened flight segments provided in the embodiment of the present application is used to construct a flight segment network relationship diagram based on navigation flow data through a construction unit 801, wherein the graph nodes of the flight segment network relationship diagram include at least one of the following: opened flight segments, unopened flight segments, and the connection relationship between the graph nodes represents airport information, wherein the flight segment network relationship diagram includes at least one of the following: a flight segment network topology diagram, a flight segment network association diagram; a first input unit 802 is used to input the flight segment network relationship diagram into a target graph attention network to obtain flight segment embedded feature information; a second input unit 803 is used to input the flight segment embedded feature information into a long short-term memory network to obtain flight segment time feature information; a third input unit 804 is used to input the flight segment time feature information into a non-dominated sorting genetic algorithm to obtain flow information of unopened flight segments, thereby solving the problem in the related art that it is impossible to accurately predict the flow of undeveloped flight segments. By constructing a flight segment network relationship diagram, the flight segment network relationship diagram is analyzed and processed step by step through the target graph attention network, long short-term memory network and non-dominated sorting genetic algorithm, thereby achieving the effect of accurately predicting the traffic of undeveloped flight segments.

[0102] Optionally, in the traffic prediction device for unopened segments provided in an embodiment of the present application, a construction unit includes: a first determination subunit, used to determine the departure airport information and destination airport information in the navigation traffic data; a construction subunit, used to construct a segment network relationship diagram based on the segment information between the departure airport and the destination airport, wherein at least one of the segments between the departure airport and the destination airport has been opened.

[0103] Optionally, in the traffic prediction device for unopened flight segments provided in an embodiment of the present application, a sub-unit is constructed, including: an acquisition module for acquiring initial characteristic information of the departure airport information and the destination airport information, wherein the initial characteristic information includes at least one of the following: inter-city distance, economic indicators, airport passenger throughput, population size, and competition factors; a construction module for constructing a flight segment network relationship based on the initial characteristic information.

[0104] Optionally, in the traffic prediction device for unopened flight segments provided in an embodiment of the present application, the flight segment network relationship is a flight segment network association diagram, and a construction module includes: a processing submodule, used to normalize the initial feature information to obtain a normalized feature vector, and obtain the feature dimension of the initial feature information; a first determination submodule, used to determine similarity information based on the feature vector and the feature dimension, wherein the similarity information is used to indicate the accuracy of the flight segment network relationship; a second determination submodule, used to determine the flight segment network association diagram based on the similarity information.

[0105] Optionally, in the traffic prediction device for unopened flight segments provided in an embodiment of the present application, the first input unit includes: a second determination subunit, used to determine the node feature set and attention coefficient of the flight segment network relationship graph; and a first input subunit, used to input the node feature set into the target graph attention network to obtain the flight segment embedded feature information.

[0106] Optionally, in the traffic prediction device for an unopened flight segment provided in an embodiment of the present application, the second input unit includes: an acquisition subunit, used to select historical flight segment embedded feature information of multiple historical moments from the flight segment embedded feature information; a second input subunit, used to input the historical flight segment embedded feature information into a long short-term memory network to obtain the flight segment time feature information.

[0107] Optionally, in the traffic prediction device for an unopened segment provided in an embodiment of the present application, the third input unit includes: a third determination subunit, used to determine an initial population based on segment time characteristic information, wherein the initial population includes multiple segment traffic information solutions; a processing subunit, used to process the initial population according to a preset operation through a non-dominated sorting genetic algorithm to obtain a target population, wherein the target population is the optimal solution among multiple segment traffic information solutions; and a fourth determination subunit, used to determine the traffic information of the unopened segment based on the target population.

[0108] The traffic prediction device for unopened flight segments includes a processor and a memory. The above-mentioned construction unit 801, first input unit 802, second input unit 803, third input unit 804, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0109] The processor includes a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the flow prediction of unopened flight segments can be performed by adjusting kernel parameters.

[0110] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0111] In the exemplary embodiment of the present application, a computer storage medium capable of implementing the above method is also provided. A program product capable of implementing the above method of the present specification is stored thereon. In some possible embodiments, various aspects of the present application can also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps described in the "Exemplary Method" section of the present specification according to various exemplary embodiments of the present application, such as the following steps: constructing a segment network relationship graph based on the navigation flow data, wherein the graph nodes of the segment network relationship graph include at least one of the following: opened segments, unopened segments, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship graph includes at least one of the following: segment network topology graph, segment network association graph; inputting the segment network relationship graph into the target graph attention network to obtain segment embedded feature information; inputting the segment embedded feature information into the long short-term memory network to obtain segment time feature information; inputting the segment time feature information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened segment.

[0112] In an optional implementation: determining the departure airport information and the destination airport information in the flight traffic data; constructing a segment network relationship diagram based on the segment information between the departure airport and the destination airport, wherein at least one of the segments between the departure airport and the destination airport has been opened.

[0113] In an optional implementation: initial characteristic information of the departure airport information and the destination airport information is obtained, wherein the initial characteristic information includes at least one of the following: distance between cities, economic indicators, airport passenger throughput, population size, and competition factors; and a flight segment network relationship is constructed based on the initial characteristic information.

[0114] In an optional implementation: normalizing the initial feature information to obtain a normalized feature vector, and obtaining a feature dimension of the initial feature information; determining similarity information based on the feature vector and the feature dimension, wherein the similarity information is used to represent the accuracy of the network relationship of the flight segments; and determining a flight segment network association graph based on the similarity information.

[0115] In an optional implementation: determining a node feature set and an attention coefficient of a flight segment network relationship graph; inputting the node feature set into a target graph attention network to obtain flight segment embedding feature information.

[0116] In an optional implementation: historical flight segment embedding feature information of multiple historical moments is selected from the flight segment embedding feature information; the historical flight segment embedding feature information is input into a long short-term memory network to obtain the flight segment time feature information.

[0117] In an optional implementation: an initial population is determined based on the time characteristic information of the flight segment, wherein the initial population includes multiple flight segment flow information solutions; the initial population is processed by a non-dominated sorting genetic algorithm according to preset operations to obtain a target population, wherein the target population is the optimal solution among the multiple flight segment flow information solutions; and flow information of unopened flight segments is determined based on the target population.

[0118] In an optional embodiment, the embodiment of the present application may also include a program product for implementing the above method, which may adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present application is not limited thereto, and in this document, a readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus, or device.

[0119] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0120] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0121] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0122] Program code for performing the operations of the present application may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0123] In addition, in an exemplary embodiment of the present application, an electronic device capable of implementing the above method is also provided.

[0124] Those skilled in the art will appreciate that various aspects of the present application may be implemented as a system, method or program product. Therefore, various aspects of the present application may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to as "circuit", "module" or "system" herein.

[0125] Refer to the following Fig. 9 hereinafter describes an electronic device 900 according to this embodiment of the present application. Fig. 9 The electronic device 900 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0126] like Fig. 9 As shown, the electronic device 900 is in the form of a general computing device. The components of the electronic device 900 may include but are not limited to: at least one processing unit 910, at least one storage unit 920, a bus 930 connecting different system components (including the storage unit 920 and the processing unit 910), and a display unit 940.

[0127] The storage unit stores program code, and the program code can be executed by the processing unit 910, so that the processing unit 910 executes the steps described in the above "Exemplary Method" section of this specification according to various exemplary embodiments of the present application. For example, the processing unit 910 can execute the following steps: construct a segment network relationship graph based on the navigation flow data, wherein the graph nodes of the segment network relationship graph include at least one of the following: opened segments, unopened segments, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship graph includes at least one of the following: segment network topology graph, segment network association graph; input the segment network relationship graph into the target graph attention network to obtain segment embedded feature information; input the segment embedded feature information into the long short-term memory network to obtain segment time feature information; input the segment time feature information into the non-dominated sorting genetic algorithm to obtain the flow information of the unopened segment.

[0128] In an optional implementation: determining the departure airport information and the destination airport information in the flight traffic data; constructing a segment network relationship diagram based on the segment information between the departure airport and the destination airport, wherein at least one of the segments between the departure airport and the destination airport has been opened.

[0129] In an optional implementation: initial characteristic information of the departure airport information and the destination airport information is obtained, wherein the initial characteristic information includes at least one of the following: distance between cities, economic indicators, airport passenger throughput, population size, and competition factors; and a flight segment network relationship is constructed based on the initial characteristic information.

[0130] In an optional implementation: normalizing the initial feature information to obtain a normalized feature vector, and obtaining a feature dimension of the initial feature information; determining similarity information based on the feature vector and the feature dimension, wherein the similarity information is used to represent the accuracy of the network relationship of the flight segments; and determining a flight segment network association graph based on the similarity information.

[0131] In an optional implementation: determining a node feature set and an attention coefficient of a flight segment network relationship graph; inputting the node feature set into a target graph attention network to obtain flight segment embedding feature information.

[0132] In an optional implementation: historical flight segment embedding feature information of multiple historical moments is selected from the flight segment embedding feature information; the historical flight segment embedding feature information is input into a long short-term memory network to obtain the flight segment time feature information.

[0133] In an optional implementation: an initial population is determined based on the time characteristic information of the flight segment, wherein the initial population includes multiple flight segment flow information solutions; the initial population is processed by a non-dominated sorting genetic algorithm according to preset operations to obtain a target population, wherein the target population is the optimal solution among the multiple flight segment flow information solutions; and flow information of unopened flight segments is determined based on the target population.

[0134] The storage unit 920 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 9201 and / or a cache storage unit 9202 , and may further include a read-only storage unit (ROM) 9203 .

[0135] The storage unit 920 may also include a program / utility 9204 having a set (at least one) of program modules 9205, such program modules 9205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0136] Bus 930 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0137] The electronic device 900 may also communicate with one or more external devices 1000 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 900, and / or communicate with any device that enables the electronic device 900 to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication may be performed through an input / output (I / O) interface 950. In addition, the electronic device 900 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter 960. As shown, the network adapter 960 communicates with other modules of the electronic device 900 through a bus 930. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 900, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0138] Through the description of the above implementation methods, it is easy for those skilled in the art to understand that the example implementation methods described here can be implemented by software, or by combining software with necessary hardware. Therefore, the technical solution according to the implementation method of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation method of the present application.

[0139] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present application, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0140] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application. The specification and embodiments are to be regarded as exemplary only, and the true scope and spirit of the present application are indicated by the claims.

Claims

1. A method for predicting traffic flow of unopened flight segments, characterized in that: include: Constructing a segment network relationship graph based on the flight traffic data, wherein the graph nodes of the segment network relationship graph include: opened segments and unopened segments, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship graph includes: a segment network topology graph and a segment network association graph; Inputting the flight segment network relationship graph into the target graph attention network to obtain the flight segment embedding feature information; Inputting the flight segment embedding feature information into a long short-term memory network to obtain flight segment time feature information; Inputting the flight segment time characteristic information into a non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment; Among them, the segment network relationship diagram is constructed based on the navigation flow data, including: Determining the origin airport information and the destination airport information in the navigation flow data; constructing the flight segment network relationship graph based on the flight segment information between the origin airport and the destination airport, wherein at least one of the flight segments between the origin airport and the destination airport has been opened; The step of constructing the flight segment network relationship diagram based on the flight segment information between the departure airport and the destination airport includes: Acquire initial characteristic information of the departure airport information and the destination airport information, wherein the initial characteristic information includes at least one of the following: inter-city distance, economic index, airport passenger throughput, population, and competition factor; Constructing the flight segment network relationship diagram according to the initial feature information; The step of constructing the flight segment network relationship diagram according to the initial feature information includes: Normalizing the initial feature information to obtain a normalized feature vector and acquiring a feature dimension of the initial feature information; Determining similarity information according to the feature vector and the feature dimension, wherein the similarity information is used to represent the accuracy of the network relationship of the flight segments; The flight segment network relationship diagram is determined according to the similarity information.

2. The method according to claim 1, characterized in that The flight segment network relationship graph is input into the target graph attention network to obtain the flight segment embedding feature information, including: Determining a node feature set and an attention coefficient of the flight segment network relationship graph; The node feature set is input into the target graph attention network to obtain the flight segment embedding feature information.

3. The method according to claim 1, characterized in that The flight segment embedding feature information is input into the long short-term memory network to obtain the flight segment time feature information, including: Selecting historical flight segment embedding feature information at multiple historical moments from the flight segment embedding feature information; The historical flight segment embedding feature information is input into the long short-term memory network to obtain the flight segment time feature information.

4. The method according to claim 1, characterized in that Inputting the flight segment time characteristic information into a non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment, including: Determining an initial population according to the flight segment time characteristic information, wherein the initial population includes a plurality of flight segment flow information solutions; Processing the initial population according to a preset operation by a non-dominated sorting genetic algorithm to obtain a target population, wherein the target population is an optimal solution among the multiple flight segment flow information solutions; The flow information of the unopened flight segment is determined according to the target population.

5. A flow prediction device for an unopened flight segment, characterized in that: include: A construction unit is used to construct a segment network relationship graph based on the navigation flow data, wherein the graph nodes of the segment network relationship graph include: opened segments and unopened segments, and the connection relationship between the graph nodes represents airport information, wherein the segment network relationship graph includes: a segment network topology graph and a segment network association graph; A first input unit, used to input the flight segment network relationship graph into the target graph attention network to obtain the flight segment embedding feature information; A second input unit is used to input the flight segment embedding feature information into the long short-term memory network to obtain the flight segment time feature information; A third input unit is used to input the flight segment time characteristic information into a non-dominated sorting genetic algorithm to obtain the flow information of the unopened flight segment; Wherein, the construction unit comprises: a first determination subunit, used to determine the origin airport information and the destination airport information in the navigation flow data; A construction subunit, configured to construct the flight segment network relationship diagram based on the flight segment information between the origin airport and the destination airport, wherein at least one of the flight segments between the origin airport and the destination airport has been opened; The construction subunit includes: an acquisition module for acquiring initial characteristic information of the departure airport information and the destination airport information, wherein the initial characteristic information includes at least one of the following: inter-city distance, economic indicators, airport passenger throughput, population, and competition factors; A construction module, used for constructing the flight segment network relationship diagram according to the initial feature information; Wherein, the building blocks include: A processing submodule, used for normalizing the initial feature information to obtain a normalized feature vector and acquire a feature dimension of the initial feature information; A first determination submodule, configured to determine similarity information according to the feature vector and the feature dimension, wherein the similarity information is used to represent the accuracy of the network relationship of the flight segments; The second determining submodule is used to determine the flight segment network relationship diagram according to the similarity information.

6. A computer-readable storage medium, characterized in that: The storage medium includes a stored program, wherein the program executes the traffic prediction method for an unopened flight segment according to any one of claims 1 to 4.

7. An electronic device, characterized in that: include: One or more processors, a memory, a display device and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the traffic prediction method for the unopened flight segment as described in any one of claims 1 to 4.

Images