Dynamic deep space network flow prediction method and system based on graph attention network

Abstract

The invention discloses a dynamic deep space network flow prediction method and system based on a graph attention network, and relates to the technical field of deep space network communication. The method comprises the following steps: receiving deep space network flow data, preprocessing the network flow data, and dividing the network flow data into a training set and a test set; an RNN recurrent neural network is used as a basic structure, a deep space network traffic prediction model is constructed, and the deep space network traffic prediction model is divided into an encoder part and a decoder part. According to the method, the spatial attention module and the sequence convolution module are respectively used for analyzing the spatial correlation and the temporal correlation of the deep space network flow information, the network flow spatial-temporal correlation is fully analyzed, so that the prediction result is more accurate, the flow correlation between unconnected nodes is considered, the graph generation module is introduced, and the prediction efficiency is improved. An adjacent matrix based on dynamic attributes can be obtained according to current and historical flow information, the matrix reflects correlation among nodes from a long term, and long-term prediction is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of deep space network communication technology, and in particular to a dynamic deep space network traffic prediction method and system based on a graph attention network. Background Art

[0002] Satellite networks play an important role in various fields such as communications, aviation, navigation and military. In recent years, with the advancement of science and technology, human exploration of space has become increasingly in-depth, and artificial satellites and spacecraft are constantly moving into deeper space, which makes deep space communication one of the focuses in the field of scientific research. Compared with ground communication and low-orbit satellite communication, deep space communication faces significant technical challenges, including huge signal transmission delays and link discontinuity, which greatly affect the efficiency and reliability of communication.

[0003] In recent years, traffic prediction technologies have emerged in an endless stream, which can be mainly divided into three categories: statistical models, traditional machine learning models and neural network models. Statistical models rely too much on prior knowledge to obtain parameters and have a fixed model structure that cannot be adaptively changed, which leads to their inability to deeply explore the deep-level characteristics of traffic information; compared with the first two methods, neural network models have stronger nonlinear fitting and adaptive learning capabilities, and can accurately predict the changing trend of traffic. However, the traffic prediction algorithm based on neural networks lacks comprehensive consideration of the dynamic topology of deep space networks and the spatiotemporal correlation of traffic, and the prediction accuracy is poor. For this reason, we propose a dynamic deep space network traffic prediction method and system based on graph attention networks. Summary of the invention

[0004] The purpose of the present invention is to provide a dynamic deep space network traffic prediction method based on graph attention network to solve the problems raised in the above background technology.

[0005] According to a first aspect of the present invention, in order to achieve the above-mentioned purpose, the present invention provides the following technical solution: a dynamic deep space network traffic prediction method based on a graph attention network, comprising the following steps:

[0006] Receive deep space network traffic data, pre-process the network traffic data and divide it into training set and test set;

[0007] A deep space network traffic prediction model is constructed based on the RNN recurrent neural network structure. The deep space network traffic prediction model is divided into an encoder part and a decoder part. The encoder and the decoder both include a graph generation module for generating a dynamic attribute adjacency matrix based on current and historical traffic information, a spatial attention module for capturing the spatial correlation of traffic data, and a sequence processing module for capturing the temporal correlation of traffic data.

[0008] Input the adjacency matrix and the training set into the deep space network traffic prediction model, and train it using the accelerated training algorithm to obtain the trained deep space network traffic prediction model;

[0009] The test set is input into the trained deep space network traffic prediction model for testing to obtain the prediction results.

[0010] Furthermore, the deep space network traffic data is used as follows:

[0011] Based on the traffic monitoring process set up in each network node, the traffic data is periodically uploaded to the control end. The control end integrates the traffic data of each node as the data set of the deep space network traffic prediction model, and divides the data set into training set and test set in the ratio of 7:3.

[0012] Furthermore, a graph generation module is used to generate an adjacency matrix based on dynamic attributes, as follows:

[0013] (1) At each time step, the deep space network traffic X t and the hidden state H at the previous time step t-1 Spliced ​​together as input to the graph generation module;

[0014] (2) Generate a dynamic filter DF through graph convolution operation t , and then embedding the randomly initialized nodes with DF t Multiply point by point to get the dynamic node embedding DE, in this way, we get the dynamic node embedding DE1 and DE2 of the source node and target node respectively;

[0015] (3) Finally, node similarity is used to calculate the adjacency matrix DA based on dynamic attributes t .

[0016] Furthermore, the spatial attention module is implemented using the graph attention network GAT, and the spatial correlation of deep space network traffic data is captured using the spatial attention module, as follows:

[0017] (1) GAT uses the attention mechanism to replace the traditional graph convolution operation. Its input is a set of node features, and its output is a set of new node features, where the node features are the traffic information of the satellite nodes.

[0018] (2) In order to map the input features to a high-dimensional space, all nodes share the weight matrix W, and then use the shared attention mechanism β to calculate the attention coefficient between the node and its neighbors. Finally, the normalized attention coefficient m ij The number is put into the graph convolution;

[0019] (3) GAT contains a multi-head attention mechanism for multi-channel convolutional neural networks. Each head sets corresponding parameters independently. The K-head attention mechanism is used in the deep space network traffic prediction model, and the K heads are aggregated by average aggregation.

[0020] Furthermore, the sequence processing module is implemented using GRU, and the sequence processing module is used to capture the time correlation of traffic information, as follows:

[0021] In GRU, there are update gates and reset gates. The update gate determines how much information in the previous state needs to be retained in the current state. The larger the value of the update gate, the more historical information will be saved. The reset gate is responsible for controlling the extent to which the information in the previous state is forgotten or ignored. The larger the value of the reset gate, the more previous state information is discarded, thereby focusing on the new information currently input.

[0022] Furthermore, the working steps of the encoder part of the deep space network traffic prediction model are as follows:

[0023] (1) First, at each time step, the graph generation module obtains an adjacency matrix based on state attributes according to the current traffic information;

[0024] (2) The adjacency matrix based on state attributes and the adjacency matrix based on satellite connection relationships are combined as a new adjacency matrix, which is used together with the flow information in the deep space network at the current moment as the input of the spatial attention module to obtain the processed deep space network flow information;

[0025] (3) The processed deep space network traffic information and the hidden information at historical moments are used as the input of the time series processing module to obtain the processed feature sequence and the hidden output at the current moment.

[0026] Furthermore, the working steps of the decoder part of the deep space network traffic prediction model are as follows:

[0027] (1) First, at each time step, the graph generation module obtains an adjacency matrix based on state attributes according to the hidden output of the previous moment;

[0028] (2) The adjacency matrix based on state attributes and the random topology are combined as a new adjacency matrix, which is used together with the flow information of the hidden output at the previous moment as the input of the spatial attention module to obtain the processed sequence;

[0029] (3) The processed sequence and the hidden output of the previous moment are taken together as the input of the time series processing module to obtain the hidden output of the current moment and the deep space network traffic prediction result of the current moment.

[0030] Furthermore, the accelerated training algorithm is used to train the deep space network traffic prediction model, as follows:

[0031] (1) Initialize the number of cycles cir and the counter count;

[0032] (2) When the number of cycles cir is less than the set number of cycles MT, execute (3)-(12);

[0033] (3) Initialize the hidden state H0 and randomly select a batch from the input

[0034] (4) If the number of loops is an integer multiple of the set step length and the count counter is less than the predicted time Q, the count counter is increased by 1;

[0035] (5) Execute a loop to update H from p from 0 to P-1 p+1 ;

[0036] (6) Initialize the decoder internal signal

[0037] (7) Execute (8)-(10) for q from 0 to count-1;

[0038] (8) Update the predicted value

[0039] (9) Select a random number a;

[0040] (10) If a is less than the sampling value of the cycle, the actual flow rate is used. renew Otherwise, use the predicted value renew

[0041] (11) Update the loss value L, where the loss function is set to the masked mean absolute error;

[0042] (12) Update model parameters and add 1 to the counter;

[0043] (13)End the loop.

[0044] According to a second aspect of the present invention, the present invention provides a dynamic deep space network traffic prediction system based on a graph attention network, which is used to implement the above-mentioned dynamic deep space network traffic prediction method based on a graph attention network, comprising:

[0045] A preprocessing module is used to receive deep space network traffic data, and divide the network traffic data into a training set and a test set after preprocessing;

[0046] A construction module is used to construct a deep space network traffic prediction model based on an RNN recurrent neural network as a basic structure, wherein the deep space network traffic prediction model is divided into an encoder part and a decoder part, wherein the encoder and the decoder both include a graph generation module for generating a dynamic attribute adjacency matrix based on current and historical traffic information, a spatial attention module for capturing spatial correlation of traffic data, and a sequence processing module for capturing temporal correlation of traffic data;

[0047] A training module is used to input the adjacency matrix and the training set into the deep space network traffic prediction model, and train it using an accelerated training algorithm to obtain a trained deep space network traffic prediction model;

[0048] The prediction output module is used to input the test set into the trained deep space network traffic prediction model for testing and obtain the prediction results.

[0049] According to a third aspect of the present invention, the present invention provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the memory stores a computer program capable of running on the processor, and when the processor loads and executes the computer program, the above-mentioned dynamic deep space network traffic prediction method based on graph attention network is adopted.

[0050] The present invention has at least the following beneficial effects:

[0051] 1. The present invention uses a spatial attention module and a sequence convolution module to analyze the spatial correlation and temporal correlation of deep space network traffic information respectively. The full analysis of the spatiotemporal correlation of network traffic makes the prediction results more accurate, which is suitable for solving the problem of frequent link interruptions in deep space networks.

[0052] 2. The present invention takes into account the existence of traffic correlation between unconnected nodes and introduces a graph generation module. This module obtains an adjacency matrix based on dynamic attributes according to current and historical traffic information. This matrix reflects the correlation between nodes in the long term. Therefore, the present invention can effectively cope with long-term prediction tasks.

[0053] 3. In the prediction model, the adjacency matrix is ​​set as the weighted adjacency matrix based on dynamic attributes and the adjacency matrix based on connection relationships. Therefore, during the prediction process, the present invention can adaptively change the weighted values ​​to adapt to different prediction tasks.

[0054] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the process of the method of the present invention;

[0056] Figure 2 It is a structural schematic diagram of the deep space network traffic prediction model in the present invention;

[0057] Figure 3 It is a structural schematic diagram of the image generation module in the present invention;

[0058] Figure 4 Schematic diagram of the structure of the spatial attention module in the present invention;

[0059] Figure 5 It is a structural schematic diagram of a single sequence processing module in the present invention;

[0060] Figure 6 It is a schematic diagram of the flow of the accelerated training algorithm in the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0062] In actual deep space networks, there is a large transmission delay, so it is difficult and costly to perform real-time state-aware congestion control on the network. Therefore, using a dynamic deep space network traffic prediction method based on a graph attention network to predict the traffic state of the deep space network in advance can effectively achieve congestion control, optimize traffic scheduling, and ensure efficient and stable communications.

[0063] See also Figure 1-Figure 6 The present invention provides a technical solution: a dynamic deep space network traffic prediction method based on a graph attention network, comprising the following steps:

[0064] S1. Receive deep space network traffic data, pre-process the network traffic data and divide it into training set and test set;

[0065] The Deep Space Network traffic data is used as follows:

[0066] Based on the traffic monitoring process set up by each network node, the traffic data is periodically uploaded to the control end, which integrates the traffic data of each node as the data set of the deep space network traffic prediction model and divides the data set into a training set and a test set in a ratio of 7:3;

[0067] S2. Based on the RNN recurrent neural network as the basic structure, a deep space network traffic prediction model is constructed. The deep space network traffic prediction model is divided into an encoder part and a decoder part. The encoder and the decoder both include a graph generation module for generating a dynamic attribute adjacency matrix based on current and historical traffic information, a spatial attention module for capturing the spatial correlation of traffic data, and a sequence processing module for capturing the temporal correlation of traffic data, such as Figure 2 As shown,

[0068] (S21) With respect to the technical solution of this embodiment, as Figure 3 As shown in the figure, the graph generation module generates an adjacency matrix based on dynamic attributes according to the current and historical traffic information, as follows:

[0069] (S21.1) At each time step, the deep space network traffic X t and the hidden state H at the previous time step t-1 Stitched together as input to the graph generation module;

[0070] (S21.2) Generate a dynamic filter DF through graph convolution operation t , and then embedding the randomly initialized nodes with DF t Multiply point by point to get the dynamic node embedding DE, in this way, we get the dynamic node embedding DE1 and DE2 of the source node and target node respectively;

[0071] (S21.3) Finally, the node similarity is used to calculate the adjacency matrix DA based on dynamic attributes t ;

[0072] (S22) Figure 4 As shown in Figure 1, the spatial attention module is implemented using the graph attention network GAT, and the spatial correlation of deep space network traffic data is captured using the spatial attention module, as follows:

[0073] (S22.1) GAT uses the attention mechanism to replace the traditional graph convolution operation. Its input is a set of node features, and its output is a set of new node features, where the node features are the traffic information of the satellite nodes.

[0074] (S22.2) In order to map the input features to a high-dimensional space, all nodes share the weight matrix W, and then use the shared attention mechanism β to calculate the attention coefficient between the node and its neighbors. Finally, the normalized attention coefficient m ij The number is put into the graph convolution;

[0075] (S22.3) GAT includes a multi-head attention mechanism for multi-channel convolutional neural networks, each head sets corresponding parameters independently, and adopts a K-head attention mechanism in the deep space network traffic prediction model, and aggregates the K heads by average aggregation;

[0076] (S24) The sequence processing module is implemented using GRU, and the sequence processing module is used to capture the time correlation of traffic information, as follows:

[0077] like Figure 5 As shown, an update gate r is set in GRU t and reset gate z t , they determine the state update of information. Specifically, the update gate determines how much information in the previous state needs to be retained in the current state. The larger the value of the update gate, the more historical information will be preserved. On the other hand, the reset gate is responsible for controlling the extent to which the information of the previous state is forgotten or ignored. The larger the value of the reset gate, the more the model tends to discard more previous state information and focus on the new information currently input.

[0078] It should be noted that the working steps of the encoder part are as follows:

[0079] (1) First, at each time step, the graph generation module obtains an adjacency matrix based on state attributes according to the current traffic information;

[0080] (2) The adjacency matrix based on state attributes and the adjacency matrix based on satellite connection relationships are combined as a new adjacency matrix, which is used together with the flow information in the deep space network at the current moment as the input of the spatial attention module to obtain the processed deep space network flow information;

[0081] (3) The processed deep space network traffic information and the hidden information at the historical moment are used as the input of the time series processing module to obtain the processed feature sequence and the hidden output at the current moment;

[0082] The working steps of the decoder part are as follows:

[0083] (4) First, at each time step, the graph generation module obtains an adjacency matrix based on state attributes according to the hidden output of the previous moment;

[0084] (5) The adjacency matrix based on state attributes and the random topology are combined as a new adjacency matrix, which is used together with the flow information of the hidden output at the previous moment as the input of the spatial attention module to obtain the processed sequence;

[0085] (6) The processed sequence and the hidden output at the previous moment are used as the input of the time series processing module to obtain the hidden output at the current moment and the deep space network traffic prediction result at the current moment;

[0086] S3. Input the adjacency matrix and training set into the deep space network traffic prediction model, and use the accelerated training algorithm to train it to obtain the trained deep space network traffic prediction model, such as Figure 6 As shown, the details are as follows:

[0087] (1) Initialize the number of cycles cir and the counter count;

[0088] (2) When the number of cycles cir is less than the set number of cycles MT, execute (3)-(12);

[0089] (3) Initialize the hidden state H0 and randomly select a batch from the input

[0090] (4) If the number of loops is an integer multiple of the set step length and the count counter is less than the predicted time Q, the count counter is increased by 1. Figure 6 Steps 4-6;

[0091] (5) Execute a loop to update H from 0 to P-1 p+1 ,like Figure 6 Steps 7-9;

[0092] (6) Initialize the decoder internal signal

[0093] (7) Execute (8)-(10) for q from 0 to count-1;

[0094] (8) Figure 6 Step 12 in step 12 updates the predicted value

[0095] (9) Figure 6 In step 13, a random number a is selected;

[0096] (10) If a is less than the sampling value of the cycle, the actual flow rate is used. renew Otherwise, use the predicted value renew like Figure 6 Steps 14-18;

[0097] (11) Update the loss value L, where the loss function is set to the masked mean absolute error, as Figure 6 Step 20;

[0098] (12) Update model parameters and add 1 to the counter;

[0099] (13)End the loop.

[0100] In order to reduce the time and resource consumption in the training process, the present invention proposes an accelerated training algorithm. First, in the forward propagation process before the backward propagation, not all the steps of the decoder are calculated, only the first few steps are calculated, and this step is recorded as a count; during the training process, the count is continuously increased until the predicted sequence length is reached. This training method significantly reduces the training time and saves a lot of memory resources in the early stage (lines 4-6). In addition, combining the sampling and training processes can further improve the training performance (lines 14-18);

[0101] S4. Input the test set into the trained deep space network traffic prediction model for testing to obtain the prediction results.

[0102] In summary, this embodiment proposes a dynamic deep space network traffic prediction method based on graph attention network (TPDS-GAN), which comprehensively considers the spatiotemporal correlation between the deep space network topology dynamics and network traffic; considering that there is also traffic correlation between unconnected nodes, this embodiment introduces a graph generation module, which obtains an adjacency matrix based on dynamic attributes according to current and historical traffic information. The matrix reflects the correlation between nodes in the long term, so this embodiment can effectively cope with long-term prediction tasks; at the same time, in the deep space network traffic prediction model of this embodiment, the adjacency matrix is ​​set as the weighted adjacency matrix based on dynamic attributes and the adjacency matrix based on connection relationships, so the weighted value can be adaptively changed during the prediction process to adapt to different prediction tasks; finally, this embodiment uses the spatial attention module and the sequence convolution module to analyze the spatial correlation and temporal correlation of deep space network traffic information respectively, and the full analysis of the spatiotemporal correlation of network traffic makes the prediction results more accurate.

[0103] Embodiment 2:

[0104] This embodiment provides a dynamic deep space network traffic prediction system based on a graph attention network, which is used to implement the above-mentioned dynamic deep space network traffic prediction method based on a graph attention network, including:

[0105] A preprocessing module is used to receive deep space network traffic data, and divide the network traffic data into a training set and a test set after preprocessing;

[0106] A construction module is used to construct a deep space network traffic prediction model based on an RNN recurrent neural network as a basic structure, wherein the deep space network traffic prediction model is divided into an encoder part and a decoder part, wherein the encoder and the decoder both include a graph generation module for generating a dynamic attribute adjacency matrix based on current and historical traffic information, a spatial attention module for capturing spatial correlation of traffic data, and a sequence processing module for capturing temporal correlation of traffic data;

[0107] A training module is used to input the adjacency matrix and the training set into the deep space network traffic prediction model, and train it using an accelerated training algorithm to obtain a trained deep space network traffic prediction model;

[0108] The prediction output module is used to input the test set into the trained deep space network traffic prediction model for testing and obtain the prediction results.

[0109] Specifically, the above-mentioned preprocessing module, construction module, training module and prediction output module can be embedded in a computer processing system. The computer calls the above-mentioned modules to complete the task of predicting the deep space network traffic according to the above-mentioned dynamic deep space network traffic prediction method based on graph attention network; the above-mentioned preprocessing module, construction module, training module and prediction output module can perform operations according to the specific steps given in the dynamic deep space network traffic prediction method based on graph attention network.

[0110] It should be noted that it should be understood that the division of the various modules of the above system is only the division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated, and these modules can all be implemented in the form of software calling through processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software calling through processing elements, and some modules can be implemented in the form of hardware. For example, the preprocessing module can be a separately established processing element, or it can be integrated in a certain chip of the above device. In addition, it can also be stored in the memory of the above device in the form of program code, and called and executed by a certain processing element of the above device. The implementation of other modules is similar. In addition, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each module above can be completed by the hardware integrated logic circuit in the processor element or the instructions in the form of software.

[0111] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASIC), or one or more digital singnal processors (DSP), or one or more field programmable gate arrays (FPGA). For another example, when a module is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processors that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0112] Embodiment three:

[0113] The present invention provides a terminal device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the memory stores a computer program capable of running on the processor, and when the processor loads and executes the computer program, the above-mentioned dynamic deep space network traffic prediction method based on graph attention network is adopted.

[0114] It should be noted that the terminal device can be a computer device such as a desktop computer, a laptop computer or a cloud server, and the terminal device includes but is not limited to a processor and a memory. For example, the terminal device can also include input and output devices, network access devices and buses.

[0115] Furthermore, the processor may adopt a central processing unit (CPU). Of course, according to actual usage, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. may also be adopted. The general-purpose processor may adopt a microprocessor or any conventional processor, etc., and the present application does not impose any restrictions on this.

[0116] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0117] For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances. When an element is referred to as being "assembled on", "installed on", "fixed on" or "set on" another element, it can be directly on the other element or there can also be a centered element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be a centered element at the same time. The terms "vertical", "horizontal", "up", "down", "left", "right" and similar expressions used herein are only for illustrative purposes and are not intended to be the only implementation method.

[0118] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0119] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

Claims

1. A dynamic deep space network traffic prediction method based on graph attention network, characterized by: The following steps are involved: Receive deep space network traffic data, pre-process the network traffic data and divide it into training set and test set; A deep space network traffic prediction model is constructed based on the RNN recurrent neural network structure. The deep space network traffic prediction model is divided into an encoder part and a decoder part. The encoder and the decoder both include a graph generation module for generating a dynamic attribute adjacency matrix based on current and historical traffic information, a spatial attention module for capturing the spatial correlation of traffic data, and a sequence processing module for capturing the temporal correlation of traffic data. Input the adjacency matrix and the training set into the deep space network traffic prediction model, and train it using the accelerated training algorithm to obtain the trained deep space network traffic prediction model; The test set is input into the trained deep space network traffic prediction model for testing to obtain the prediction results.

2. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 1 is characterized in that: The Deep Space Network traffic data is used as follows: Based on the traffic monitoring process set up in each network node, the traffic data is periodically uploaded to the control end. The control end integrates the traffic data of each node as the data set of the deep space network traffic prediction model, and divides the data set into training set and test set in the ratio of 7:

3.

3. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 2 is characterized in that: Use the graph generation module to generate an adjacency matrix based on dynamic attributes, as follows: (1) At each time step, the deep space network traffic X t and the hidden state H at the previous time step t-1 Stitched together as input to the graph generation module; (2) Generate a dynamic filter DF through graph convolution operation t , and then embedding the randomly initialized nodes with DF t Multiply point by point to get the dynamic node embedding DE, in this way, we get the dynamic node embedding DE1 and DE2 of the source node and target node respectively; (3) Finally, node similarity is used to calculate the adjacency matrix DA based on dynamic attributes t .

4. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 3 is characterized in that: The spatial attention module is implemented using the graph attention network GAT, and the spatial correlation of deep space network traffic data is captured using the spatial attention module, as follows: (1) GAT uses the attention mechanism to replace the traditional graph convolution operation. Its input is a set of node features, and its output is a set of new node features, where the node features are the traffic information of the satellite nodes. (2) In order to map the input features to a high-dimensional space, all nodes share the weight matrix W, and then use the shared attention mechanism β to calculate the attention coefficient between the node and its neighbors. Finally, the normalized attention coefficient m ij The number is put into the graph convolution; (3) GAT contains a multi-head attention mechanism for multi-channel convolutional neural networks. Each head sets corresponding parameters independently. The K-head attention mechanism is used in the deep space network traffic prediction model, and the K heads are aggregated by average aggregation.

5. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 4 is characterized in that: The sequence processing module is implemented using GRU, and the sequence processing module is used to capture the time correlation of traffic information, as follows: In GRU, there are update gates and reset gates. The update gate determines how much information in the previous state needs to be retained in the current state. The larger the value of the update gate, the more historical information will be saved. The reset gate is responsible for controlling the extent to which the information in the previous state is forgotten or ignored. The larger the value of the reset gate, the more previous state information is discarded, thereby focusing on the new information currently input.

6. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 5 is characterized in that: The working steps of the encoder part of the deep space network traffic prediction model are as follows: (1) First, at each time step, the graph generation module obtains an adjacency matrix based on state attributes according to the current traffic information; (2) The adjacency matrix based on state attributes and the adjacency matrix based on satellite connection relationships are combined as a new adjacency matrix, which is used together with the flow information in the deep space network at the current moment as the input of the spatial attention module to obtain the processed deep space network flow information; (3) The processed deep space network traffic information and the hidden information at historical moments are used as the input of the time series processing module to obtain the processed feature sequence and the hidden output at the current moment.

7. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 6 is characterized by: The working steps of the decoder part of the deep space network traffic prediction model are as follows: (1) First, at each time step, the graph generation module obtains an adjacency matrix based on state attributes according to the hidden output of the previous moment; (2) The adjacency matrix based on state attributes and the random topology are combined as a new adjacency matrix, which is used together with the flow information of the hidden output at the previous moment as the input of the spatial attention module to obtain the processed sequence; (3) The processed sequence and the hidden output of the previous moment are taken together as the input of the time series processing module to obtain the hidden output of the current moment and the deep space network traffic prediction result of the current moment.

8. The method for dynamic deep space network traffic prediction based on graph attention network according to claim 6 is characterized in that: The accelerated training algorithm is used to train the deep space network traffic prediction model. The specific method is as follows: (1) Initialize the number of cycles cir and the counter count; (2) When the number of cycles cir is less than the set number of cycles MT, execute (3)-(12); (3) Initialize the hidden state H0 and randomly select a batch from the input (4) If the number of loops is an integer multiple of the set step length and the count counter is less than the predicted time Q, the count counter is increased by 1; (5) Execute a loop to update H from p from 0 to P-1 p+1 ; (6) Initialize the decoder internal signal (7) Execute (8)-(10) for q from 0 to count-1; (8) Update the predicted value (9) Select a random number a; (10) If a is less than the sampling value of the cycle, the actual flow rate is used. renew Otherwise, use the predicted value renew (11) Update the loss value L, where the loss function is set to the masked mean absolute error; (12) Update model parameters and add 1 to the counter; (13)End the loop.

9. A dynamic deep space network traffic prediction system based on a graph attention network, used to implement the dynamic deep space network traffic prediction method based on a graph attention network as described in any one of claims 1 to 8, characterized in that: include: A preprocessing module is used to receive deep space network traffic data, and divide the network traffic data into a training set and a test set after preprocessing; A construction module is used to construct a deep space network traffic prediction model based on an RNN recurrent neural network as a basic structure, wherein the deep space network traffic prediction model is divided into an encoder part and a decoder part, wherein the encoder and the decoder both include a graph generation module for generating a dynamic attribute adjacency matrix based on current and historical traffic information, a spatial attention module for capturing spatial correlation of traffic data, and a sequence processing module for capturing temporal correlation of traffic data; A training module is used to input the adjacency matrix and the training set into the deep space network traffic prediction model, and train it using an accelerated training algorithm to obtain a trained deep space network traffic prediction model; The prediction output module is used to input the test set into the trained deep space network traffic prediction model for testing and obtain the prediction results.

10. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: The memory stores a computer program that can be run on the processor. When the processor loads and executes the computer program, the dynamic deep space network traffic prediction method based on the graph attention network described in any one of claims 1 to 8 is adopted.

Images