Satellite optical fiber transmission data stream automatic drawing method and system

Abstract

The invention provides an automatic drawing method and system for a satellite optical fiber transmission data stream, and the method comprises the steps: a user adds an e-mail address through software, and the software detects the uploading state of an interface data sheet, and sends a detection result to the user through an e-mail. And after the mail is successfully sent, the system records and displays a mail notification in a specified area. And identifying the interface data sheet by using GNN and graph embedding technologies. And the software establishes a digital model according to an identification result, forms a mapping relation with the constituent elements, and draws an optical fiber transmission data flow statistical graph according to a rule. According to the invention, automatic satellite data flow drawing is realized, and the efficiency of data processing and error detection is improved. The method has remarkable advantages in the aspects of improving the automation degree, accuracy and user friendliness of data processing.

Description

Technical Field

[0001] The present invention relates to the technical field of optical fiber transmission, and specifically, to a method and system for automatically drawing satellite optical fiber transmission data streams. Background Art

[0002] In satellite communication systems, the speed of data transmission and processing is crucial. Traditional satellite data transmission methods pose many challenges, especially in terms of processing speed and data accuracy. Usually, data is transmitted through optical fibers to a laser terminal, and then processed and drawn by the terminal.

[0003] However, due to the extremely high rate of satellite data transmission, when software receives and processes interface data sheets, it often fails to promptly identify and record which files are successfully uploaded and which ones fail to upload. This results in information loss and errors during the data processing.

[0004] Existing technologies also have obvious deficiencies in data recognition. Especially when dealing with complex and large-scale data, it is difficult to ensure both accuracy and efficiency. The content on multiple interface data sheets may be self-contradictory or have printing errors. At this time, conventional extraction methods cannot distinguish the content, and manual intervention is often relied on, which is inefficient and error-prone.

[0005] Therefore, a method is needed to monitor the recognition of interface data sheets and can more accurately identify and process the data on the interface data sheets, which can effectively solve many problems in traditional methods, improve the automation degree of data processing, and significantly enhance the accuracy and efficiency of data recognition. Summary of the Invention

[0006] Aiming at the deficiencies in the prior art, the purpose of the present invention is to provide a method and system for automatically drawing satellite optical fiber transmission data streams.

[0007] According to a method for automatically drawing satellite optical fiber transmission data streams provided by the present invention, it is characterized by including:

[0008] Step S1: Add an email to detect the upload status of the interface data sheet; the uploaded files will be saved in a specified directory on the server, and the user's email address and notification options will be recorded;

[0009] Step S2: Send the detected status result to the user via email;

[0010] Step S3: After the email is successfully sent, record and display past email notifications in another specified area;

[0011] Step S4: After extracting the content on the interface data sheet form, represent each data point and the relationships between the data points in the interface data sheet as a graph structure, and construct an identification model using the GNN and graph embedding methods;

[0012] Step S5: Determine whether there are errors on the current interface data sheet. If so, immediately notify the user in the form of a log and give a prediction plan to correct the errors or supplement the missing data; if not, execute Step S6;

[0013] Step S6: Standardize the names of the device names and signal contents through which the optical fiber transmission data flow statistical chart flows, establish a digital model and form a mapping relationship with the constituent elements, and draw the optical fiber transmission data flow statistical chart according to the rules.

[0014] Preferably, when the upload status of all files is recorded, an email including the upload status-related information corresponding to each file is generated;

[0015] The upload status-related information includes file name, upload time, and upload result;

[0016] If the file is a failed file, the content of the email also includes the failed file name, the specific error reason, and all corresponding solutions.

[0017] Preferably, in Step S2, the generated email is sent to the user through an SMTP server or a third-party mail server.

[0018] Preferably, in the graph structure, nodes represent each data item, including device name, device code, and signal content;

[0019] Edges represent the relationships between the data items, including signal transmission paths or logical associations;

[0020] Further capture the interdependencies and interaction relationships between data items according to the content in the graph.

[0021] Preferably, after constructing the graph structure, GNN performs feature embedding on the nodes in the graph through a neural network, including:

[0022] Step S4.1: Represent the attributes of each node as a multi-dimensional vector to capture the feature information of the node;

[0023] Step S4.2: Through a multi-layer message passing mechanism, GNN will perform information propagation and aggregation between the nodes and edges of the graph;

[0024] Step S4.3: Each node will receive information from its neighbor nodes at each layer, and the information includes the features of the neighbor nodes and the relationship weights between them.

[0025] Preferably, an attention mechanism is introduced in the message passing process. When each node receives neighbor information, it adjusts the amount of information received according to the importance weights of the neighbor nodes;

[0026] During the training process of the GNN, noise is added or some data is randomly discarded to simulate data missing and abnormal situations.

[0027] According to a satellite optical fiber transmission data stream automatic mapping system provided by the present invention, it includes:

[0028] Module M1: Add an email, detect the upload status of the interface data sheet; the uploaded file will be saved in a specified directory on the server, and the user's email address and notification options will be recorded;

[0029] Module M2: Send the detected status result to the user by email;

[0030] Module M3: After the email is sent successfully, record and display the past email notifications in another specified area;

[0031] Module M4: After extracting the content on the interface data sheet form, represent each data point and the relationship between the data points in the interface data sheet as a graph structure, and construct an identification model by using GNN and graph embedding;

[0032] Module M5: Judge whether there is an error on the current interface data sheet. If so, immediately notify the user in the form of a log and give a prediction scheme for correcting the error or supplementing the missing data; if not, execute Module M6;

[0033] Module M6: Standardize the names of the device names and signal contents passed by the optical fiber transmission data stream statistical chart, establish a digital model and form a mapping relationship with the constituent elements, and draw the optical fiber transmission data stream statistical chart according to the rules.

[0034] Preferably, in the graph structure, the nodes represent each data item, including the device name, device code, and signal content;

[0035] The edges represent the relationships between the data items, including the signal transmission path or logical association;

[0036] Further capture the interdependent and interactive relationships between the data items according to the content in the graph.

[0037] Preferably, after the graph structure is constructed, the GNN performs feature embedding on the nodes in the graph through a neural network, including:

[0038] Module M4.1: Represent the attributes of each node as a multi-dimensional vector to capture the feature information of the node;

[0039] Module M4.2: Through a multi-layer message passing mechanism, the GNN will perform information propagation and aggregation between the nodes and edges of the graph;

[0040] Module M4.3: Each node will receive information from its neighboring nodes at each layer, and the information includes the features of the neighboring nodes and the relationship weights between them.

[0041] Preferably, an attention mechanism is introduced during the message passing process. When each node receives neighbor information, it will adjust the amount of received information according to the importance weights of the neighboring nodes;

[0042] During the training process of the GNN, noise is added or some data is randomly discarded to simulate data missing and abnormal situations.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention monitors the software through email notification to perform secondary discrimination on the extracted interface data sheet data by BERT, and automatically draws the path through which the satellite data is transmitted along the optical fiber by the software. The data improves the efficiency of data flow drawing and management through visualization.

[0045] 2. The present invention further improves the reliability of satellite data transmission by setting the threshold of transmission delay and the selection of alternative paths through statistical methods.

[0046] 3. The present invention can capture complex interdependencies by using the feature embedding and message passing mechanism of the GNN, while the attention mechanism adjusts the amount of information according to the importance weights of neighboring nodes during the information aggregation process, thereby improving the recognition accuracy of the model. In addition, by introducing noise during the GNN training process, the robustness of the model to data defects is enhanced. Once incorrect data is found, the system will immediately notify the administrator and provide corrective or supplementary measures to ensure data quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] By reading the following detailed description of the non-limiting embodiments with reference to the accompanying drawings, other features, objectives and advantages of the present invention will become more obvious:

[0048] Figure 1 It is a schematic flow chart of the working method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0049] The present invention will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that those of ordinary skill in the art can make several changes and improvements without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0050] Example 1

[0051] A method for automatically drawing satellite optical fiber transmission data streams provided by the present invention transmits satellite data to a laser terminal through an optical fiber and draws a graph through self-developed software. As Figure 1 shown, it includes:

[0052] Step S1: Add an email and detect the upload status of the interface data sheet. The user adds a valid email address in the specified area and adds a checkbox to ask the user whether to detect the upload status of the interface data sheet and send it to the server. If not, it will not connect to the server and directly proceed to the step of identifying the interface data sheet; if so, the number of filled email addresses is no more than 5. The uploaded files will be saved in the specified directory of the server, and the user's email address and notification options will be recorded. Next, the server will process the file upload and record the upload status of each file, including successful files and failed files.

[0053] Step S2: Send the detected status result to the user by email. When the upload status of all files is recorded, an email including the upload status-related information corresponding to each file is generated. The upload status-related information includes the file name, upload time, and upload result. If the file is a failed file, the content of the email also includes the failed file name, specific error reason, and possible solutions. Specifically, the email is titled "Satellite Optical Fiber Transmission Data Stream Automatic Drawing File Upload Status Report". At the beginning, it briefly explains the purpose and content of the email. The main body of the email details the upload status of each file, including the file name, upload time, upload result (success or failure). If the upload fails, the reason or error message for the failure is provided. The end part provides some follow-up suggestions and the contact information of the sender. In addition, the email system supports rich text format, and further uses text formatting tools such as bold, italic, and color to highlight the content of the uploaded failed files and the error reasons.

[0054] The generated email is sent to the user through the SMTP server. If the email sending fails, the software will record this failure event and try to resend it. After failing again, the system administrator will be notified immediately. Third-party email servers can also be used. These services usually provide API interfaces, which are convenient for integration into applications and can also handle email bounce and spam issues.

[0055] Step S3: After the email is successfully sent, record and display past email notifications in another designated area. Each time an email is successfully sent, a new record is created in the designated area, and the record contains the detailed information of the email. The detailed information includes the sending time, the recipient's email address, the email content summary, and the email sending status. The email sending status includes sent successfully, failed, or the number of retries. These information are displayed in the form of a list in the designated area. Users can easily browse and retrieve historical email notifications through this interface, and a search and filtering function based on partial email information is provided. These records are sorted chronologically by default, and the latest email notification is displayed at the front.

[0056] Step S4: After extracting the content on the interface data sheet form, represent each data point and the relationships between the data points in the interface data sheet as a graph structure, and use the GNN and graph embedding methods to construct an identification model. After constructing the graph structure, GNN performs feature embedding on the nodes in the graph through a neural network. Represent the attributes of each node as a multi-dimensional vector to capture the feature information of the node. Next, through a multi-layer message passing mechanism, GNN will perform information propagation and aggregation between the nodes and edges of the graph. Each node will receive information from its neighbor nodes at each layer, and the information includes the features of the neighbor nodes and the relationship weights between them. The nodes in the graph structure represent each data item, including device name, device code, signal content, and the edges represent the relationships between the data items, including signal transmission paths or logical associations. Further capture the interdependent and interactive relationships between data items according to the content in the graph.

[0057] An attention mechanism is introduced during the message passing process. When each node receives neighbor information, it will adjust the amount of information received according to the importance weight of the neighbor node. During the training process of GNN, some noise is deliberately added or some data is randomly discarded to simulate data missing and abnormal situations.

[0058] Step S5: Determine whether there are errors on the current interface data sheet. If so, immediately notify the user in the form of a log and give a prediction plan to correct the error or supplement the missing data; if not, execute Step S6. Specifically, if an error is identified, classify the identified error, including data missing, format error, and logical inconsistency. Each type of error will generate a detailed log record, clearly indicating the type of the error and the specific location where it occurred, and circled in red, describe in detail the specific content and impact of the error, and record the specific time when the error was identified. At the same time, give specific correction steps according to the specific error. For example, if a certain field is missing, the system will automatically complete it. If a certain data is missing, the system will provide a predicted value, and the user can choose to directly use these predicted values or further verify them before using. Each error log will be attached with a unique identifier for traceability.

[0059] Step S6: Standardize the names of the devices through which the optical fiber transmission data flow statistical chart passes and the signal content, establish a digital model and form a mapping relationship with the constituent elements, and draw the optical fiber transmission data flow statistical chart according to the rules. Lock a certain data frame at fixed intervals of time, calculate the difference between the time when the data frame is transmitted from one module and the receiving time of another module, so as to obtain the transmission delay. Then draw a histogram, adjust the parameter to set an appropriate number of intervals to observe the frequency distribution of the data. Next, draw a box plot. The median line in the box plot represents the median, the upper and lower edges of the box represent the first and third quartiles respectively, the "whiskers" outside the box extend to the minimum and maximum values of the non-outliers, and the points beyond the whiskers are marked as outliers. Combine the position where the data frequency in the histogram drops significantly to calculate a reasonable threshold. And display this delay in the figure, and mark it as red if it exceeds the threshold.

[0060] If the delay is continuously shown in red for a long time, then prepare to enable the standby route, conduct standby path inspection, physical connection inspection and signal quality detection, and finally conduct connection stability testing. If the packet loss rate is low and the delay is within an acceptable range, the standby path is considered stable and reliable.

[0061] After confirming that there is no problem with the standby path, the system will update the routing table, change the target path of data transmission to the standby path, notify the sender and receiver to update the transmission path, and resynchronize the data transmission status.

[0062] During the switching process, the system will also record detailed fault information, including the time and location of the fault occurrence, the type of fault, and the entire process of the switching path.

[0063] The present invention aims to significantly improve the efficiency and accuracy of data transmission and processing. This method for automatically drawing the optical fiber transmission data flow of satellites greatly improves the efficiency and accuracy of data processing through multi-step automated processing and intelligent recognition technology. In terms of user experience, users can easily add email addresses and select whether to enable the notification function in the software. The system automatically receives and processes the uploaded files, and generates a detailed status report email to be sent to the users. This mechanism not only ensures the transparency of data upload, but also improves the reliability of the system through the email resending and administrator notification mechanisms. Even when the email sending fails, relevant personnel can be notified in time to avoid data loss.

[0064] In terms of data plotting and error detection, the present invention utilizes graph neural networks (GNNs) and graph embedding techniques to represent each data point and its relationships in the interface data sheet as a graph structure. This makes the dependency relationships and logical associations of the data more intuitive and easier to analyze. The feature embedding and message passing mechanisms of GNNs can capture complex interdependencies, while the attention mechanism adjusts the amount of information according to the importance weights of neighbor nodes during the information aggregation process, thereby improving the recognition accuracy of the model. In addition, by introducing noise during the GNN training process, the robustness of the model to data defects is enhanced. Once error data is detected, the system will immediately notify the administrator and provide corrective or supplementary measures to ensure data quality.

[0065] During the plotting process of the optical fiber transmission data stream, the standardized device names, signal content, and the establishment of digital models ensure data consistency and accuracy. The statistical graph of the optical fiber transmission data stream drawn according to the rules makes the transmission path of the data stream and the device interaction relationships more intuitive, facilitating monitoring and management. Overall, this method not only improves the degree of automation of data processing but also significantly enhances the ability of data anomaly detection and error correction, ensuring high-quality and reliable data. At the same time, the flexible notification and recording mechanism enables users to keep track of the data processing status at any time, greatly enhancing the practicality of the system and the user experience.

[0066] Furthermore, by locking a certain data frame at fixed intervals and calculating the transmission delay by taking the difference between the transmission time of the data frame from one module and the reception time of another module, precise monitoring of data transmission delay is achieved. By plotting a histogram, users can observe the frequency distribution of the data after adjusting the appropriate number of intervals in the parameter settings, while the box plot provides more detailed statistical analysis, where the median line represents the median, the upper and lower edges of the box represent the first and third quartiles respectively, the "whiskers" outside the box extend to the minimum and maximum non-outlier values, and the points beyond the whiskers are marked as outliers. Combining with the position where the data frequency significantly drops in the histogram, a reasonable threshold is calculated and the delay data is displayed in the graph. The delay data exceeding the threshold will be marked in red.

[0067] Through this method, users can intuitively identify abnormal situations of transmission delay. If the delay is continuously shown in red for a long time, the system will automatically activate the backup route. The activation of the backup route includes checking the backup path, physical connection, and signal quality, and finally conducting a connection stability test. If the packet loss rate is low and the delay is within an acceptable range, the backup path is considered stable and reliable. After confirming that the backup path is problem-free, the system will update the routing table, change the target path of data transmission to the backup path, and notify the sender and receiver to update the transmission path and resynchronize the data transmission status.

[0068] During the handover process, the system will record the fault information in detail, including the time and location of the fault, the type of fault, and the entire process of the handover path. This method not only improves the reliability and stability of data transmission, but also can respond quickly when a fault occurs to ensure that data transmission is not interrupted. Through automated delay monitoring and standby path switching, the present invention significantly improves the intelligence level of the data transmission system, reduces manual intervention, improves the overall efficiency and reliability of the system, and has significant application value and promotion prospects.

[0069] In summary, the present invention overcomes the problems of slow data processing speed, low accuracy, and excessive manual intervention in traditional methods, and realizes the automated drawing of satellite optical fiber transmission data streams. It provides a more reliable and intelligent solution for satellite data transmission, and has significant application value and promotion prospects.

[0070] Embodiment 2

[0071] The present invention also provides a satellite optical fiber transmission data stream automatic drawing system, which can be realized by executing the process steps of the satellite optical fiber transmission data stream automatic drawing method. That is, those skilled in the art can understand the satellite optical fiber transmission data stream automatic drawing method as the preferred embodiment of the satellite optical fiber transmission data stream automatic drawing system.

[0072] According to a satellite optical fiber transmission data stream automatic drawing provided by the present invention, it is characterized in that it includes:

[0073] Module M1: Add an email, detect the upload status of the interface data sheet; the uploaded file will be saved in the specified directory of the server, and the user's email address and notification options will be recorded.

[0074] Module M2: Send the detected status result to the user by email. The module M2 sends the generated email to the user through the SMTP server or a third-party email server.

[0075] Module M3: Record and display past email notifications in another specified area after the email is sent successfully. When the upload status of all files is recorded, an email including the upload status-related information corresponding to each file is generated. The upload status-related information includes the file name, upload time, and upload result. If the file is a failed file, the content of the email also includes the failed file name, the specific error reason, and all corresponding solutions.

[0076] Module M4: After extracting the content on the interface data sheet, represent each data point and the relationships between the data points in the interface data sheet as a graph structure, and construct an identification model using the GNN and graph embedding methods. In the graph structure, nodes represent each data item, including device name, device code, and signal content. Edges represent the relationships between the data items, including signal transmission paths or logical associations. Further capture the interdependent and interactive relationships between data items based on the content in the graph.

[0077] After constructing the graph structure, GNN performs feature embedding on the nodes in the graph through a neural network, including: Module M4.1: Represent the attributes of each node as a multi-dimensional vector to capture the feature information of the node. Module M4.2: Through a multi-layer message passing mechanism, GNN propagates and aggregates information between the nodes and edges in the graph. Module M4.3: Each node receives information from its neighbor nodes at each layer, and the information includes the features of the neighbor nodes and the relationship weights between them. An attention mechanism is introduced during the message passing process. When each node receives neighbor information, it adjusts the amount of received information according to the importance weights of the neighbor nodes. During the training process of GNN, noise is added or some data is randomly discarded to simulate data missing and abnormal situations.

[0078] Module M5: Determine whether there are errors on the current interface data sheet. If so, immediately notify the user in the form of a log and give a prediction plan to correct the errors or supplement the missing data; if not, execute Module M6.

[0079] Module M6: Standardize the names of the device names and signal content through which the optical fiber transmission data flow statistical chart passes, establish a digital model and form a mapping relationship with the constituent elements, and draw the optical fiber transmission data flow statistical chart according to the rules.

[0080] Those skilled in the art know that in addition to implementing the system and its various devices, modules, and units provided by the present invention in the form of pure computer-readable program code, the method steps can be logically programmed to enable the system and its various devices, modules, and units provided by the present invention to be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers, etc., to achieve the same functions. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be regarded as the structure within the hardware component; the devices, modules, and units for implementing various functions can also be regarded as both software modules for implementing the method and the structure within the hardware component.

[0081] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for automatically drawing satellite optical fiber transmission data streams, characterized in that: include: Step S1: Add email and check the upload status of interface data sheet; Uploaded files are saved in a specified directory on the server, and the user's email address and notification preferences are recorded; Step S2: Send the detection status result to the user via email; Step S3: after the email is successfully sent, record and display the past email notifications in another designated area; Step S4: after extracting the content on the interface data sheet, each data point in the interface data sheet and the relationship between the data points are represented as a graph structure, and a recognition model is constructed by using GNN and graph embedding; Step S5: Determine whether there is an error on the current interface data sheet. If so, immediately notify the user in the form of a log and provide a prediction plan for correcting the error or supplementing the missing data; if not, execute step S6; Step S6: standardize the device names and signal contents through which the optical fiber transmission data flow statistics flow, establish a digital model and form a mapping relationship with the constituent elements, and draw the optical fiber transmission data flow statistics according to the rules.

2. The satellite optical fiber transmission data stream automatic drawing method according to claim 1, characterized in that: When the upload status of all files is recorded, an email including information related to the upload status of each file is generated; The upload status related information includes file name, upload time, and upload result; If the file is a failed file, the content of the email also includes the failed file name, the specific error cause and all corresponding solutions.

3. The satellite optical fiber transmission data stream automatic drawing method according to claim 1, characterized in that: The step S2 sends the generated email to the user via an SMTP server or a third-party mail server.

4. The satellite optical fiber transmission data stream automatic mapping method according to claim 1, characterized in that: The nodes in the graph structure represent various data items, including device name, device code, and signal content; The edge represents the relationship between the data items, including a signal transmission path or a logical association; The interdependencies and interactions between data items are further captured based on the content in the graph.

5. The method for automatically drawing satellite optical fiber transmission data stream according to claim 1, characterized in that: After building the graph structure, GNN embeds features of nodes in the graph through a neural network, including: Step S4.1: Represent the attributes of each node as a multidimensional vector to capture the characteristic information of the node; Step S4.2: Through the multi-layer message passing mechanism, GNN will propagate and aggregate information between the nodes and edges of the graph; Step S4.3: Each node receives information from its neighbor nodes at each layer, and the information includes the features of the neighbor nodes and the relationship weights between them.

6. The method for automatically drawing satellite optical fiber transmission data stream according to claim 5, characterized in that: The attention mechanism is introduced in the message transmission process. When each node receives neighbor information, it adjusts the amount of information received according to the importance weight of the neighbor node. Add noise or randomly discard some data during the GNN training process to simulate data missing and abnormal situations.

7. A satellite optical fiber transmission data stream automatic mapping system, characterized in that: include: Module M1: Add email and check the upload status of interface data sheet; Uploaded files are saved in a specified directory on the server, and the user's email address and notification preferences are recorded; Module M2: Send the detection status results to the user via email; Module M3: After the email is successfully sent, record and display the past email notifications in another designated area; Module M4: After extracting the content on the interface data sheet, each data point in the interface data sheet and the relationship between the data points are represented as a graph structure, and a recognition model is constructed using GNN and graph embedding; Module M5: Determine whether there is an error on the current interface data sheet. If so, immediately notify the user in the form of a log and provide a prediction plan to correct the error or supplement the missing data; if not, execute module M6; Module M6: standardize the device names and signal contents through which the optical fiber transmission data flow statistics flow, establish a digital model and form a mapping relationship with the constituent elements, and draw the optical fiber transmission data flow statistics according to the rules.

8. The satellite optical fiber transmission data stream automatic mapping system according to claim 7, characterized in that: The nodes in the graph structure represent various data items, including device name, device code, and signal content; The edge represents the relationship between the data items, including a signal transmission path or a logical association; The interdependencies and interactions between data items are further captured based on the content in the graph.

9. The satellite optical fiber transmission data stream automatic mapping system according to claim 7, characterized in that: After building the graph structure, GNN embeds features of nodes in the graph through a neural network, including: Module M4.1: Represent the attributes of each node as a multidimensional vector to capture the characteristic information of the node; Module M4.2: Through a multi-layer message passing mechanism, GNN will propagate and aggregate information between nodes and edges of the graph; Module M4.3: Each node receives information from its neighbor nodes at each layer, which contains the characteristics of the neighbor nodes and the relationship weights between them.

10. The satellite optical fiber transmission data stream automatic mapping system according to claim 9, characterized in that: The attention mechanism is introduced in the message transmission process. When each node receives neighbor information, it adjusts the amount of information received according to the importance weight of the neighbor node. Add noise or randomly discard some data during the GNN training process to simulate data missing and abnormal situations.

Images