A method and system for automatic aggregation and submission of raw data from multi-source subsystems in wind tunnel testing

By introducing data convergence nodes and cross-domain virtual interaction execution agents in wind tunnel tests, the automated aggregation and cross-domain delivery of multi-source data are achieved, solving the problems of low data aggregation efficiency, lack of integrity, and cumbersome operation in existing technologies, and realizing efficient and reliable data transmission and monitoring.

CN123104956APending Publication Date: 2026-11-17AVIC SHENYANG AERODYNAMICS RES INST
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611613479.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, the raw data from multi-source subsystems in wind tunnel tests are difficult to automatically aggregate, verify in completeness, and transfer across domains in an isolated network environment. This results in low operational efficiency, lack of data integrity assurance, cumbersome cross-domain transmission steps, and a lack of real-time monitoring and fault-tolerant recovery mechanisms.

Method used

After the experiment, the multi-source data acquisition subsystem group adds train number and subsystem identifiers and asynchronously transmits them to the data convergence node for two-dimensional index temporary storage and completeness verification, generates a directory to be submitted, and uses cross-domain virtual interaction execution agent to establish interactive session with cross-domain transmission client to realize automated data aggregation and cross-domain submission. A five-state machine is used to manage task status, and a Web-driven engine is combined to control cross-domain transmission to realize real-time monitoring and visual configuration.

Benefits of technology

It automates the main flow of raw data from the multi-source subsystem of wind tunnel testing from acquisition to cross-domain delivery, reduces manual operation, improves data integrity verification capabilities, is compatible with existing network architectures without modification, and reduces operation and maintenance complexity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104956A_ABST
    Figure CN123104956A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for automatically gathering and submitting raw data from multiple subsystems in wind tunnel testing, belonging to the field of wind tunnel testing measurement and control and data management technology. It solves the problem that traditional wind tunnel testing data submission methods and systems in the prior art struggle to automatically gather, verify completeness, and submit raw data generated by multiple acquisition subsystems in an isolated network environment. This invention retrieves ready tasks from the upload ready queue through a cross-domain virtual interactive execution agent, establishes an interactive session with the cross-domain transmission client in the isolated network environment through a Web-driven engine manager, and controls the cross-domain transmission client to complete the operation. The current task status is updated based on the result information returned by the cross-domain data transmission client, completing the automatic gathering and submission of raw data from multiple subsystems. This invention achieves automatic gathering and submission of raw data from multiple subsystems and can be applied to various wind tunnel test sites equipped with multiple heterogeneous data acquisition subsystems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of wind tunnel test measurement and control and data management technology, specifically to a method and system for automatically collecting and submitting raw data from multi-source subsystems in wind tunnel tests. Background Technology

[0002] Wind tunnel testing is one of the most important ground-based testing methods in the aerospace field, used to measure the aerodynamic characteristics of aircraft models under different airflow conditions. In a typical wind tunnel test, the test is organized and executed in units of "runs," each run representing a specific set of test conditions (such as Mach number, angle of attack, sideslip angle, etc.). Under these conditions, the test system collects various data on the model, including aerodynamic loads, surface pressure distribution, and model attitude. Modern wind tunnel facilities are typically equipped with multiple parallel acquisition subsystems, including but not limited to: a steady-state acquisition system for acquiring strain signals from a force balance; a dynamic pressure acquisition system for capturing high-frequency pressure signals from the model; a scanning valve pressure acquisition system for obtaining pressure distribution on the model surface; and a video measurement acquisition system for acquiring data through optical vision and using it for subsequent calculations of the model's actual attitude angles in the airflow. These acquisition subsystems are typically composed of different specialized hardware and software, running on their own independent acquisition terminals, resulting in raw data that varies in format, quantity, and timing, forming a "multi-source heterogeneous" data characteristic.

[0003] In the wind tunnel test management process, the raw data from each test run needs to be submitted to a designated recipient on the office network after the test for subsequent data processing, analysis, and archiving. However, for information security reasons, the wind tunnel test network and the office network are usually physically or logically isolated. Cross-domain data transmission needs to be completed through a dedicated cross-domain data transmission client. The cross-domain data transmission client mainly provides a web-based human-computer interaction mode and does not provide a programmatic calling interface suitable for batch submission scenarios of multiple wind tunnel tests. Its typical operation process includes: user authentication login, managing historical files, uploading the data to be transmitted to the transmission queue, waiting for the data transmission to complete, configuring the recipient information, and executing the sending operation. Currently, the entire data flow process mainly relies on manual operation. Specifically, after each test run, test personnel need to log in to the terminals of each acquisition subsystem, manually copy the raw data files to a dedicated data convergence node, and then start the cross-domain data transmission client on the data convergence node, sequentially executing multiple interactive steps such as login, file upload, recipient configuration, and sending confirmation.

[0004] Therefore, the manual operation method has the following technical defects: First, low operational efficiency: Wind tunnel tests are conducted intensively, with dozens or even hundreds of tests per day. Each test requires manual data collection and cross-domain transfer, consuming manpower and time. Second, lack of data integrity guarantee: When manually collecting data from multiple acquisition subsystems, it is easy to miss data files from a certain subsystem or confuse data from different tests, lacking a systematic means of verifying the completeness of multi-source data. Third, numerous cross-domain transmission operation steps: The web interaction interface of the cross-domain data transmission client involves multiple operation steps, including identity authentication, file management, and recipient configuration, making manual operation cumbersome. Fourth, lack of real-time monitoring of the transfer link: Under the manual operation mode, it is impossible to keep track of the arrival status of data from each subsystem and the execution progress of cross-domain transmission in real time, and anomalies are not detected and handled in a timely manner. Fifth, lack of automated fault tolerance and recovery mechanism: In the event of network fluctuations, client anomalies, or expired identity credentials, manual operation requires re-executing all steps from the beginning, and it is impossible to resume transmission from the breakpoint or automatically retry.

[0005] In summary, a method and system for automatically collecting and submitting raw data from multi-source subsystems in wind tunnel testing are needed. Summary of the Invention

[0006] A brief overview of the invention is given below to provide a basic understanding of certain aspects of it. It should be understood that this overview is not an exhaustive summary of the invention. It is not intended to identify key or essential parts of the invention, nor is it intended to limit the scope of the invention. Its purpose is merely to present certain concepts in a simplified form as a prelude to the more detailed description that follows.

[0007] In view of this, in order to solve the problem that traditional wind tunnel test data submission methods and systems in the prior art are unable to automatically aggregate, verify completeness and submit cross-domain raw data generated by multiple acquisition subsystems in an isolated network environment, the present invention provides a method and system for automatic aggregation and submission of raw data from multiple source subsystems in wind tunnel tests.

[0008] Technical Solution 1 is as follows: A method for automatic aggregation and submission of raw data from multi-source subsystems in wind tunnel testing, comprising the following steps:

[0009] S1. Collect test data through a multi-source data acquisition subsystem group. After each test run, attach the test run identifier and subsystem identifier to the raw data and transmit it asynchronously to the data convergence node via the test local area network in the form of a data stream.

[0010] S2. Receive asynchronously transmitted data through the data convergence node, perform two-dimensional indexing and temporary storage according to train number identifier and subsystem identifier, and perform multi-source data completeness verification for each train number. When the multi-source data completeness verification passes, generate the pending submission directory for the current train number and write the corresponding train number task into the upload ready queue.

[0011] S3. Obtain ready tasks from the upload ready queue through the cross-domain virtual interaction execution agent, establish an interactive session with the cross-domain transmission client in the isolated network environment through the Web drive engine manager, and control the cross-domain transmission client to complete the upload of the directory to be submitted, the transmission status monitoring, and the sending operation to the pre-configured recipient;

[0012] S4. Update the current task status based on the result information returned by the cross-domain data transmission client, and complete the automatic aggregation and delivery of raw data from multiple subsystems.

[0013] Furthermore, in step S2, the two-dimensional index is constructed and temporarily stored through the data temporary storage management module;

[0014] The multi-source data completeness verification process is as follows: Based on the preset subsystem registry, the set of subsystems that the current train should arrive at is determined. For each subsystem, according to the preset completeness rules, a set of expected indexes is generated through the completeness verification module. The set of received index numbers is compared with the set of expected indexes. Only when the set of received index numbers of all registered subsystems covers their respective set of expected indexes, the completeness verification module determines that the multi-source data completeness verification of the current train has passed.

[0015] The task queue management module uses a five-state machine to manage train tasks and schedules train tasks in the data ready state into the upload ready queue. When a voucher error is detected or a pause submission instruction is received, the task currently in the submission state is rolled back to the data ready state, and subsequent tasks are prevented from leaving the queue.

[0016] The process of generating the pending submission directory for the current train number simultaneously handles filename conflict, as follows:

[0017] When flattening the data files of each subsystem from the hierarchical temporary storage directory to a unified upload working directory, a subsystem identifier and index number prefix are added to the beginning of each filename to ensure that the flattened filenames are unique within the working directory; the states of the five-state machine are as follows:

[0018] The "Waiting for Data" status indicates that the multi-source data for the current train has not yet all arrived.

[0019] The data is ready status, which indicates that the multi-source data completeness verification has passed;

[0020] The "Submission in Progress" status indicates that the cross-domain virtual interaction execution agent is performing the submission operation for the current train.

[0021] The "Completed" status indicates that the current train data has been successfully submitted and verified.

[0022] A failure status indicates that the submission operation was terminated abnormally.

[0023] Among them, train tasks in the data ready state enter the upload ready queue to wait for submission, and a five-state machine is used to complete the pause and resume submission, single task forced termination, and failed task re-entry into the queue.

[0024] When data from a subsystem arrives after the corresponding train task has entered the data ready state, if the current train task has not yet started submission, the temporary data is updated and the completeness check is re-executed; if the current train task is already in the submission or completed state, the current data is named late data, marked as ignored and logged.

[0025] When a data transmission request carrying a manual upload flag is detected, and the current request contains a data transmission end flag for the current train, the data convergence node directly marks the corresponding task as data ready without waiting for data to arrive from other subsystems, and does not perform standard multi-source data completeness verification.

[0026] The procedures for pausing and resuming submissions, and forcibly terminating single tasks are as follows:

[0027] When the data convergence node receives a stop signal, it stops retrieving new tasks from the upload ready queue, rolls back the tasks currently in the submission state to the data ready state, sends a stop notification to all registered acquisition subsystems, shuts down the Web drive engine, and persists the current queue state and all task states to the state file.

[0028] When the data convergence node restarts, it restores the task queue and task status from the status file and continues to process unfinished submission tasks.

[0029] The state data of the five-state machine adopts an atomic persistence strategy, as follows:

[0030] Write the current state to a temporary file;

[0031] Rename the original status file to a backup file;

[0032] Rename the temporary file to a formal status file;

[0033] When a corrupted official status file is detected, it will be automatically restored from a backup file.

[0034] Furthermore, the preset completeness rules include a static determination mode and a dynamic determination mode, and the specific process is as follows:

[0035] In the static determination mode, the data convergence node pre-configures a list of required index numbers for each acquisition subsystem, and the expected index set is the list of required index numbers. When the received index number set covers all the index numbers in the list of required index numbers, the data of the current subsystem is determined to be complete.

[0036] In the dynamic judgment mode, when the data acquisition subsystem transmits data, it carries the actual total amount of data for the current train number N. The data convergence node generates a set of expected index numbers from 1 to N based on the actual total amount of data N. When the set of received index numbers after deduplication covers the set of expected index numbers, it is determined that the data for the current train in the current subsystem is complete.

[0037] When a train needs to be resubmitted, the data convergence node creates a new task version for the current train, and the version number is incremented. The status of the original version task remains unchanged, and the new version task independently enters the waiting data state and performs completeness verification and submission according to the standard process, thus completing the resubmission version management.

[0038] The data convergence node performs duplicate data detection when receiving data, as follows:

[0039] When a subsystem repeatedly submits data with the same index number for the same train, it calculates the hash value between the newly arrived data and the temporarily stored data. If the hashes match, the duplicate data is ignored; if the hashes do not match, it is recorded as a data anomaly event and an alarm notification is sent to the monitoring terminal.

[0040] When the data convergence node starts the data receiving service, it sends a start notification to all registered acquisition subsystems. When the service stops or the system is reset, it sends a stop or reset notification to all registered acquisition subsystems. The linkage notification has an independent timeout control to complete the service linkage control.

[0041] Furthermore, in S3, the operation of the cross-domain virtual interaction execution agent is decomposed into multiple atomic execution steps, each with an independent retry mechanism, as follows:

[0042] If a certain execution step fails, the current step will be automatically retried within a preset number of retries;

[0043] When an interactive session failure is detected, the Web Drive Engine Manager is automatically reinitialized and execution is resumed through authentication.

[0044] Set segmented waiting periods between adjacent execution steps, and detect a termination event flag in each waiting segment;

[0045] When the termination event flag is detected to be set, the current task is aborted and the task status is updated to failure;

[0046] The Web Drive Engine Manager employs a reuse strategy, as detailed below:

[0047] Lazily initialize the Web Drive Engine Manager when the first task arrives;

[0048] Subsequent tasks reuse existing driver engine instances, using isolation mechanisms to distinguish the operational contexts of different tasks;

[0049] Each time a driver engine instance is obtained, the driver process is checked to see if it is alive and can respond to preset test operations. If it is alive and responds, the test is considered normal; otherwise, the test is considered abnormal, the existing instance is destroyed and the driver engine instance is rebuilt.

[0050] When the cross-domain virtual interaction execution agent detects a password expiration prompt during the process of controlling the cross-domain data transmission client, it selects the next set of credentials from the preset credential list to complete the password update, and persistently stores the current valid credentials after the update is successful.

[0051] Before uploading the directory to be submitted, the cross-domain virtual interaction execution agent performs environment setup operations, as follows:

[0052] The Web-driven engine controls the client to clean up residual data from past submissions and reset the submission session context, ensuring that the current submission operation is not affected by historical data.

[0053] Technical Solution 2 is as follows: A system for automatic aggregation and submission of raw data from multiple subsystems in wind tunnel testing, used to execute the method for automatic aggregation and submission of raw data from multiple subsystems in wind tunnel testing as described in Technical Solution 1, including a group of multi-source data acquisition subsystems, a data convergence node, a cross-domain virtual interaction execution agent, a cross-domain transmission client, and a real-time monitoring terminal;

[0054] The multi-source data acquisition subsystem group is connected to the data convergence node, the data convergence node is connected to the cross-domain virtual interaction execution agent, the data convergence node is connected to the real-time monitoring terminal, and the cross-domain virtual interaction execution agent is connected to the cross-domain transmission client.

[0055] Furthermore, the multi-source data acquisition subsystem group includes a steady-state acquisition terminal, a dynamic pressure acquisition terminal, a scanning valve pressure acquisition terminal, and a video measurement acquisition terminal;

[0056] The data convergence node includes a secure access control module, a data temporary storage management module, a completeness verification module, and a task queue management module connected in sequence.

[0057] The cross-domain virtual interaction execution agent includes a Web drive engine manager, a step orchestrator, and a user interface interaction executor connected in sequence.

[0058] The steady-state acquisition terminal, dynamic pressure acquisition terminal, scanning valve pressure acquisition terminal, and video measurement acquisition terminal are all connected to the secure access control module, the task queue management module is connected to the step arranger, the interface interaction actuator is connected to the cross-domain transmission client, and the secure access control module and the task queue management module are connected to the real-time monitoring terminal.

[0059] The beneficial effects of the present invention are as follows: (1) The present invention realizes the automation of the main flow links of the original data of the multi-source subsystem of wind tunnel test from acquisition to cross-domain delivery, which greatly reduces the links of manual data collection and manual operation of cross-domain data transmission client, and reduces the need for continuous manual monitoring and repetitive operation by test personnel; (2) The multi-source data integrity verification mechanism based on index number improves the multi-source data integrity verification capability and reduces the risk of data omission during manual collection; (3) Without modifying the existing cross-domain data transmission client and isolated network architecture, the existing client is controlled to complete the controlled automatic delivery through the Web drive engine, which is compatible with the dual requirements of security compliance and automation efficiency, and does not require additional modification or special interface development of the existing cross-domain transmission infrastructure; (4) The real-time monitoring and visualization configuration module based on WebSocket enables the operation and maintenance personnel to grasp the system status at any time and adjust the system parameters online, reducing the complexity of system operation and maintenance. Attached Figure Description

[0060] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention. In the drawings:

[0061] Figure 1 A flowchart illustrating the method for automatically collecting and submitting raw data from a multi-source subsystem in a wind tunnel test.

[0062] Figure 2 A schematic diagram of the system structure for the automatic aggregation and submission of raw data from multi-source subsystems in wind tunnel testing;

[0063] Figure 3 This is a schematic diagram of the overall process of the automatic aggregation and submission method for raw data of a multi-source subsystem in wind tunnel testing provided in an embodiment of the present invention;

[0064] Figure 4 A schematic diagram of the multi-source data aggregation and completeness verification process provided in an embodiment of the present invention;

[0065] Figure 5 This is a schematic diagram of the execution flow of a cross-domain virtual interaction execution agent provided in an embodiment of the present invention;

[0066] Figure 6 This is a schematic diagram of task queue state transitions provided in an embodiment of the present invention;

[0067] Figure 7 A schematic diagram of an embodiment of a system for automatically collecting and submitting raw data from multiple subsystems in wind tunnel testing.

[0068] Figure Descriptions: 100. Multi-source data acquisition subsystem group; 300. Data convergence node; 400. Cross-domain virtual interaction execution agent; 500. Cross-domain transmission client; 600. Real-time monitoring terminal; 101. Steady-state acquisition terminal; 102. Dynamic pressure acquisition terminal; 103. Scan valve pressure acquisition terminal; 104. Video measurement acquisition terminal; 301. Secure access control module; 302. Data temporary storage management module; 303. Completeness verification module; 304. Task queue management module; 401. Web-driven engine manager; 402. Step orchestrator; 403. Interface interaction executor. Detailed Implementation

[0069] To make the technical solutions and advantages of the embodiments of the present invention clearer, the exemplary embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0070] Example 1: Reference Figures 1-7 This embodiment details a method for automatically collecting and submitting raw data from a multi-source subsystem in a wind tunnel test, specifically including the following steps:

[0071] S1. Collect test data through a multi-source data acquisition subsystem group. After each test run, attach the test run identifier and subsystem identifier to the raw data and transmit it asynchronously to the data convergence node via the test local area network in the form of a data stream.

[0072] Specifically: the train number identifier is used to uniquely identify a test train number, and the subsystem identifier is used to identify the data acquisition subsystem from which the data originates;

[0073] S2. Receive asynchronously transmitted data through the data convergence node, perform two-dimensional indexing and temporary storage according to train number identifier and subsystem identifier, and perform multi-source data completeness verification for each train number. When the multi-source data completeness verification passes, generate the pending submission directory for the current train number and write the corresponding train number task into the upload ready queue.

[0074] S3. Without modifying the existing cross-domain data transmission client and isolated network architecture, the cross-domain virtual interaction execution agent retrieves ready tasks from the upload ready queue, establishes an interactive session with the cross-domain transmission client in the isolated network environment through the Web drive engine manager, and controls the cross-domain transmission client to complete the upload of the directory to be submitted, the transmission status monitoring, and the sending operation to the pre-configured recipient.

[0075] S4. Update the current task status based on the result information returned by the cross-domain data transmission client, and complete the automatic aggregation and delivery of raw data from multiple subsystems.

[0076] Furthermore, in step S2, the two-dimensional index is constructed and temporarily stored through the data temporary storage management module;

[0077] The multi-source data completeness verification process is as follows: Based on a preset subsystem registry, the set of subsystems that the current train should reach is determined. For each subsystem, according to preset completeness rules, a set of expected indexes is generated through the completeness verification module. The set of received index numbers is compared with the set of expected indexes. The index numbers are used to uniquely identify data files of the same subsystem within the same train. Only when the set of received index numbers of all registered subsystems covers their respective set of expected indexes is the multi-source data completeness verification of the current train determined to be passed by the completeness verification module.

[0078] The task queue management module uses a five-state machine to manage train tasks and schedules train tasks in the data ready state into the upload ready queue. When a voucher error is detected or a pause submission instruction is received, the task currently in the submission state is rolled back to the data ready state, and subsequent tasks are prevented from leaving the queue.

[0079] The process of generating the pending submission directory for the current train number simultaneously handles filename conflict, as follows:

[0080] When flattening the data files of each subsystem from the hierarchical temporary storage directory to a unified upload working directory, a subsystem identifier and index number prefix are added to the beginning of each filename to ensure that the flattened filenames are unique within the working directory; the states of the five-state machine are as follows:

[0081] The "Waiting for Data" status indicates that the multi-source data for the current train has not yet all arrived.

[0082] The data is ready status, which indicates that the multi-source data completeness verification has passed;

[0083] The "Submission in Progress" status indicates that the cross-domain virtual interaction execution agent is performing the submission operation for the current train.

[0084] The "Completed" status indicates that the current train data has been successfully submitted and verified.

[0085] A failure status indicates that the submission operation was terminated abnormally.

[0086] Among them, train tasks in the data ready state enter the upload ready queue to wait for submission, and a five-state machine is used to complete the pause and resume submission, single task forced termination, and failed task re-entry into the queue.

[0087] When data from a subsystem arrives after the corresponding train task has entered the data ready state, if the current train task has not yet started submission, the temporary data is updated and the completeness check is re-executed; if the current train task is already in the submission or completed state, the current data is named late data, marked as ignored and logged.

[0088] When a data transmission request carrying a manual upload flag is detected, and the current request contains a data transmission end flag for the current train, the data convergence node directly marks the corresponding task as data ready without waiting for data to arrive from other subsystems, and does not perform standard multi-source data completeness verification.

[0089] The procedures for pausing and resuming submissions, and forcibly terminating single tasks are as follows:

[0090] When the data convergence node receives a stop signal, it stops retrieving new tasks from the upload ready queue, rolls back the tasks currently in the submission state to the data ready state, sends a stop notification to all registered acquisition subsystems, shuts down the Web drive engine, and persists the current queue state and all task states to the state file.

[0091] When the data convergence node restarts, it restores the task queue and task status from the status file and continues to process unfinished submission tasks.

[0092] The state data of the five-state machine adopts an atomic persistence strategy, as follows:

[0093] Write the current state to a temporary file;

[0094] Rename the original status file to a backup file;

[0095] Rename the temporary file to a formal status file;

[0096] When a corrupted official status file is detected, it will be automatically restored from a backup file.

[0097] Furthermore, the preset completeness rules include a static determination mode and a dynamic determination mode, and the specific process is as follows:

[0098] In the static determination mode, the data convergence node pre-configures a list of required index numbers for each acquisition subsystem (steady-state acquisition terminal 101, dynamic pressure acquisition terminal 102, scanning valve pressure acquisition terminal 103, and video measurement acquisition terminal 104). The expected index set is the list of required index numbers. When the received index number set covers all the index numbers in the list of required index numbers, the data of the current subsystem is determined to be complete.

[0099] In the dynamic judgment mode, when the data acquisition subsystem transmits data, it carries the actual total amount of data for the current train number N. The data convergence node generates a set of expected index numbers from 1 to N based on the actual total amount of data N. When the set of received index numbers after deduplication covers the set of expected index numbers, it is determined that the data for the current train in the current subsystem is complete.

[0100] When a train needs to be resubmitted, the data convergence node creates a new task version for the current train, and the version number is incremented. The status of the original version task remains unchanged, and the new version task independently enters the waiting data state and performs completeness verification and submission according to the standard process, thus completing the resubmission version management.

[0101] The data convergence node performs duplicate data detection when receiving data, as follows:

[0102] When a subsystem repeatedly submits data with the same index number for the same train, it calculates the hash value between the newly arrived data and the temporarily stored data. If the hashes match, the duplicate data is ignored; if the hashes do not match, it is recorded as a data anomaly event and an alarm notification is sent to the monitoring terminal.

[0103] When the data convergence node starts the data receiving service, it sends a start notification to all registered acquisition subsystems. When the service stops or the system is reset, it sends a stop or reset notification to all registered acquisition subsystems. The linkage notification has an independent timeout control to complete the service linkage control.

[0104] Furthermore, in S3, the operation of the cross-domain virtual interaction execution agent is decomposed into multiple atomic execution steps, each with an independent retry mechanism, as follows:

[0105] If a certain execution step fails, the current step will be automatically retried within a preset number of retries;

[0106] When an interactive session failure is detected, the Web Drive Engine Manager is automatically reinitialized and execution is resumed through authentication.

[0107] Set segmented waiting periods between adjacent execution steps, and detect a termination event flag in each waiting segment;

[0108] When the termination event flag is detected to be set, the current task is aborted and the task status is updated to failure;

[0109] The Web Drive Engine Manager employs a reuse strategy, as detailed below:

[0110] Lazily initialize the Web Drive Engine Manager when the first task arrives;

[0111] Subsequent tasks reuse existing driver engine instances, using isolation mechanisms to distinguish the operational contexts of different tasks;

[0112] Each time a driver engine instance is obtained, the driver process is checked to see if it is alive and can respond to preset test operations. If it is alive and responds, the test is considered normal; otherwise, the test is considered abnormal, the existing instance is destroyed and the driver engine instance is rebuilt.

[0113] When the cross-domain virtual interaction execution agent detects a password expiration prompt during the process of controlling the cross-domain data transmission client, it selects the next set of credentials from the preset credential list to complete the password update, and persistently stores the current valid credentials after the update is successful.

[0114] Before uploading the directory to be submitted, the cross-domain virtual interaction execution agent performs environment setup operations, as follows:

[0115] The Web-driven engine controls the client to clean up residual data from past submissions and reset the submission session context, ensuring that the current submission operation is not affected by historical data.

[0116] Specifically, the process in this embodiment is as follows:

[0117] I. Train number data output process of the data acquisition subsystem

[0118] After the data output function of the acquisition terminal 101 is activated, it enters a standby state, waiting for the data output of the acquisition program to be triggered. When the steady-state acquisition program finishes a test run (e.g., run number 2601010001a), the acquisition program triggers the local data output function through a predefined interface (e.g., an HTTP POST request). This request contains the following information: run number identifier (run_id, e.g., "run_2601010001a"), run number sequence number (run_no, e.g., 2601010001a), the list of data to be output and its corresponding index number;

[0119] After receiving a trigger, the data output function of the data acquisition terminal 101 performs the following operations:

[0120] Step A1: Verify the completeness of the request parameters and confirm that necessary information such as train number identifier and data list has been provided.

[0121] Step A2: Group and label the original data by train number, and create an independent grouping directory according to the output task identifier (composed of the client task identifier plus a unique random suffix) to ensure that the data of different train numbers are not mixed up.

[0122] Step A3: Create a data output task and add it to the local output queue. The initial status of the task is "PENDING";

[0123] Step A4: The background transmission thread retrieves tasks from the queue and pushes the grouped data as a data stream to the data receiving interface of the data convergence node 300 via the experimental local area network. The transmission request carries metadata such as train number identifier, subsystem identifier, and index number list. The subsystem identifier is used to uniquely identify the acquisition subsystem and can be a name identifier, network identifier, or a combination thereof; in this embodiment, the pre-configured name identifier of the acquisition terminal is used as the subsystem identifier.

[0124] The transmission timeout is dynamically calculated based on the total data size and the preset minimum transmission rate. For example, when the total data size is 50MB and the minimum transmission rate is set to 2MB / s, the calculated timeout is 25 seconds, which is then added to the base timeout margin to obtain the final value.

[0125] If a network timeout or connection error occurs during transmission, the data output function automatically puts the task back into the queue for retry. The maximum number of retries and the retry interval are set through a configuration file. In this embodiment, the maximum number of retries is configured to 3, and the retry interval is 5 seconds.

[0126] Step A5: After successful transmission, the data output function will update the task status to "COMPLETED" and automatically clean up the grouped data after the configured expiration time;

[0127] The other three data acquisition terminals 102-104 operate independently with the same data output logic, processing the train data output by their respective acquisition programs. There is no direct communication or dependency between the data output functions of each acquisition terminal; each independently transmits data to the data convergence node 300.

[0128] II. Data Convergence Node Reception and Completeness Verification

[0129] like Figure 4 As shown, the data receiving service of data convergence node 300 continuously listens for data transmission requests from each acquisition subsystem;

[0130] Step B1: Secure Access Authentication: Data convergence node 300 maintains a subsystem registry, which records the identity information (such as name identifier, network identifier, or a combination thereof) and a list of required index numbers for all legitimate data collection subsystems. When a data transmission request is received, the requester is first authenticated to verify whether the subsystem identifier it carries is in the registry. Requests that fail authentication will be rejected. In this embodiment, referring to Table 1, four subsystems are configured in the registry;

[0131] Table 1

[0132] Subsystem Identifier name Required Index Number List Subsystem A steady-state acquisition [1, 2, ..., M] Subsystem B Dynamic pressure acquisition [1, 2, ..., P] Subsystem C Scan valve pressure acquisition [1, 2, ..., Q] Subsystem D Video measurement and acquisition [1, ..., R]

[0133] Step B2: Data Temporary Storage. After authentication, the received data files are temporarily stored in local storage in a hierarchical directory structure based on train number identifier and subsystem identifier. Simultaneously, a list of received index numbers for each subsystem under each train number is maintained in memory.

[0134] Step B3: Completeness Check. A completeness check for the current train is triggered each time a new data file is received. The check logic is as follows: Figure 4 As shown, the specific process is as follows:

[0135] (B3a) Traverse all registered subsystems in the subsystem registry;

[0136] (B3b) For each subsystem, check whether data has arrived for the current train; if a registered subsystem has no data yet, it is considered incomplete and waits for subsequent data to arrive.

[0137] (B3c) For a subsystem that has received data, compare its set of received index numbers with the list of required index numbers configured in the registry; if there are any missing index numbers, it is determined to be incomplete.

[0138] (B3d) The multi-source data completeness check for the current train is deemed to have passed only when all the required index numbers for all registered subsystems have been obtained.

[0139] Taking a specific train as an example: the completeness check passes when all M files (indexes 1 to M) of subsystem A, all P files (indexes 1 to P) of subsystem B, all Q files (indexes 1 to Q) of subsystem C, and all R files (indexes 1 to R) of subsystem D have arrived. If subsystem C has only arrived with some files and some indexes are still missing, it is considered incomplete and the system continues to wait.

[0140] Step B3': Exception boundary handling rules

[0141] During the completeness verification and task management processes described above, the data convergence node also handles various boundary conditions according to the following rules:

[0142] (B3'-1) Duplicate data processing: When a subsystem submits data with the same index number repeatedly for the same train, the data convergence node calculates the hash value of the newly arrived data and the temporarily stored data; if the hashes match, the duplicate data is ignored and a success response is returned; if the hashes do not match, it is recorded as a data anomaly event and an alarm notification is sent to the monitoring terminal, while retaining the data version that arrived first.

[0143] (B3'-2) Late data processing: When data from a subsystem arrives after the corresponding train task has entered the data ready state, if the train task has not yet started submission (still in the data ready state), then the temporary data is updated and the completeness check is re-executed; if the train task is already in the submission or completed state, then the late data is marked as ignored and recorded in the log.

[0144] (B3'-3) File flattening conflict handling: When flattening data files from each subsystem from the hierarchical temporary storage directory to the unified upload working directory, in order to avoid files with the same name generated by different subsystems from overwriting each other, the flattening operation adds the prefix "subsystem identifier_index number_" to the beginning of each file name to ensure that the file name after flattening is unique in the working directory;

[0145] (B3'-4) Queue full handling: When the upload ready queue has reached the configured maximum depth limit, newly passed train task is still marked as data ready, but will not enter the upload ready queue for the time being, and its temporary data will not be deleted; when a task is taken out of the queue, ready tasks waiting to be enqueued will be enqueued in the order of first-in first-out.

[0146] (B3'-5) Resubmission and Version Management: When a train needs to be resubmitted due to data correction, the data convergence node creates a new task version for the train, and the version number is incremented; the status of the original version task remains unchanged (if it has been completed, it will still be completed), and the new version task independently enters the waiting data state and performs completeness verification and submission according to the standard process.

[0147] Step B4: Task Enqueueing. After the completeness check passes, the data convergence node updates the task status of the train from "PENDING" to "READY" and adds it to the upload ready queue. The ready queue adopts a first-in, first-out strategy and is configured with a maximum queue depth limit (100 in this embodiment).

[0148] IV. The Working Process of Cross-Domain Virtual Interaction Execution Agent

[0149] like Figure 5As shown, the cross-domain virtual interaction execution agent 400 runs on the data convergence node 300 as a background worker thread, continuously retrieving ready tasks from the upload ready queue and performing delivery operations.

[0150] Taking train number 2601010001a as an example, after the task for this train is retrieved from the queue, the execution agent operates according to the following steps:

[0151] Step C1: Prepare the upload directory.

[0152] The data convergence node flattens all subsystem data files temporarily stored in the corresponding train's temporary storage directory (removing subsystem subdirectory levels) and copies them uniformly to a dedicated upload working directory. This working directory is cleared before each new task begins to ensure that it contains only the complete data file for the current train.

[0153] Step C2: Web drive engine initialization.

[0154] The step orchestrator 402 first obtains the driver engine instance through the Web Driver Engine Manager 401. The Web Driver Engine Manager 401 adopts a lazy initialization strategy: the driver engine instance is created only when the first task arrives, and subsequent tasks reuse the existing instance. The specific logic is as follows:

[0155] (C2a) Check if a driver engine instance already exists; if not, create a new browser driver process and configure parameters such as browser window size (e.g., 1400×900 pixels), page load timeout (e.g., 20 seconds), and operation execution timeout (e.g., 20 seconds); if configured as headless mode, the browser runs in a graphical interface-less environment.

[0156] (C2b) If a driver engine instance already exists, perform a health check: determine if the instance is healthy by detecting whether the driver process is still running; if the process has terminated or is unresponsive, destroy the old instance and recreate it.

[0157] (C2c) After obtaining a healthy driver engine instance, it creates a new operation context for the current task and clears residual session data to ensure operation isolation between tasks.

[0158] Step C3: Identity Authentication.

[0159] The interface interaction executor navigates to the login page of the cross-domain data transfer client via a 403 error. Based on the page state, it locates interactive elements such as the username input field, password input field, and login button. The configured username and password are then entered into the corresponding input fields, triggering the login operation.

[0160] After logging in, check the page response: If a login success indicator is detected (e.g., the page redirects to the expected login page), then authentication is complete; if an incorrect password message is detected, proceed to the credential error handling process (see step C3a); if an expired password message is detected, proceed to the credential update process (see step C3b).

[0161] Step C3a: Credential error handling. When a password error is detected, the current task is returned to the "data ready" state (without consuming retry attempts), the "password pending" flag is set to prevent subsequent tasks from leaving the queue, and a credential error alarm is sent to the monitoring terminal via WebSocket, waiting for the operation and maintenance personnel to update the correct credentials through the monitoring terminal.

[0162] Step C3b: Adaptive credential update. The system pre-configures a list containing multiple sets of credentials. When an expired password is detected, the interface interaction executor 403 automatically sets the old password to the current password and the new password to the next credential in the list in the password modification interface. After the modification is completed, the new credential is used to log in again. After successful modification, the currently valid credential is persistently written to the configuration file to ensure that the latest credential is used subsequently.

[0163] Step C4: Pre-submission environment setup – Cleaning up the user file management area;

[0164] The interface interaction executor navigates to the user file management area of ​​the cross-domain data transfer client at 403. If there are historical data folders left over from previous tasks in this area, the executor searches for the target folder by scrolling through the page list, selects it, and performs a deletion operation. This cleanup operation is to prevent storage overflow from causing subsequent upload failures;

[0165] The search process employs a scrolling search strategy: after each scroll to the bottom of the page list, the search checks whether the target element appears and records the number of consecutive stable cycles (when the page content no longer changes). If the number of stable cycles exceeds a preset threshold (e.g., 3 cycles), the target is determined to not exist, and the search ends.

[0166] Step C5: Pre-submission environment preparation – Cleaning up the historical document recycling area;

[0167] Navigate to the historical file recycling area, and use the same search and operation logic as in step C4 to completely delete the historical folders there, freeing up storage space;

[0168] Step C6: Pre-submission environment preparation – transmission queue cleanup;

[0169] Navigate to the transmission status display area. If historical transmission records exist, perform a clearing operation to ensure the submission environment is clean.

[0170] Step C7: Enter the path of the directory to be submitted into the file input control;

[0171] The interface interaction executor 403 navigates to the user file management area and triggers the directory upload function. Using the directory selection control mechanism supported by the Web-driven engine, the directory to be submitted is associated with the upload control, and the upload is confirmed after setting preset transmission attribute parameters on the upload confirmation screen.

[0172] After the association is complete, set the preset transmission attribute parameters on the upload confirmation screen, and then confirm the upload;

[0173] Step C8: Transmission status monitoring;

[0174] After the path is written, the cross-domain data transfer client begins transmitting data from the test network side to the office network side. The step orchestrator 402 enters a transmission waiting state, periodically checking the status information in the transmission list at a configured polling interval (e.g., 3 seconds).

[0175] Transmission status monitoring logic: Obtain the status text of the current task's corresponding entry in the transmission list. If the status shows "transmission completed," proceed to the next step; if the status shows "transmission failed," trigger a step-level retry. The entire transmission wait has a maximum timeout (180 seconds in this embodiment). Failure to complete within the timeout period is considered a failure.

[0176] During the waiting process, an interruptible sleep mechanism is employed: each sleep session is executed in segments with short intervals (e.g., 0.2 seconds), and an external termination command is checked between each segment to ensure that the task termination request can be responded to in a timely manner.

[0177] Step C9: Perform a sending operation to the pre-configured recipient;

[0178] After the transmission is complete, the interface interaction executor 403 enters the uploaded data folder, performs a select all operation, and then triggers the send function. In the send operation interface, the send title (named with the train number identifier) ​​is automatically filled in, the pre-configured recipient name is searched and selected and added to the send list, and finally the send is confirmed;

[0179] Step C10: Send result verification;

[0180] After the sending operation is executed, the interface interaction executor 403 checks the operation result feedback information on the page. If the sending is confirmed to be successful, the step orchestrator 402 updates the task status to "COMPLETED" and sends a completion notification to the monitoring terminal via WebSocket. If a sending failure is detected, a retry is determined according to the retry policy.

[0181] V. Step-level fault tolerance and retry mechanism

[0182] During the execution of steps C2 to C10 above, the step orchestrator 402 implements independent fault-tolerant control for each step, with the specific mechanism as follows:

[0183] (1) Step-level retries: Each step is configured with an independent maximum number of retries (3 times in this embodiment), and the step will automatically retry if it fails. The retry counts between steps are independent, and the retry quota of one step does not affect the retry quota of other steps;

[0184] (2) Session failure recovery: If the browser session is detected to be invalid (such as page error or driver connection failure) during any step, the step orchestrator 402 automatically calls the Web driver engine manager 401 to reinitialize the driver engine and restarts execution from step C3 (identity authentication);

[0185] (3) Interruptible execution: Each step periodically checks the termination event flag during execution. When an external command to terminate the task is issued (such as when an operations and maintenance personnel terminate a task through a monitoring terminal), the termination event is set, the current step stops execution at the next checkpoint, and the task status is updated to "FAILED".

[0186] (4) Inter-step delay: A configurable delay time is set between adjacent steps (0.1 seconds in this embodiment) to ensure that the page state is stable before executing the next step;

[0187] VI. Task Queue State Machine Management

[0188] like Figure 6 As shown, the task queue management module in data convergence node 300 maintains a five-state machine to manage the lifecycle of all train tasks:

[0189] State S1 – PENDING: The initial state after task creation, indicating that the multi-source data for the current train has not yet all arrived. In this state, the data convergence node continuously receives data from each subsystem and updates the received index numbers.

[0190] State S2 – Data Ready: This state is entered after the multi-source data integrity check passes. The task is placed in the first-in-first-out ready queue to await cross-domain virtual interaction execution agent processing.

[0191] State S3 – Uploading: This state is entered after the execution agent retrieves a task from the ready queue and begins the upload operation. Only one task can be in the upload state at a time.

[0192] Status S4 - COMPLETED: This status is entered after the submission operation is successfully completed and the sending result is verified.

[0193] Status S5 - Failed: This status is entered after the submission operation is terminated abnormally or the retry limit is exceeded.

[0194] State transition rules: S1→S2 (Completion check passed); S2→S3 (Retrieved by the agent); S3→S4 (Submission successful); S3→S5 (Submission failed); S3→S2 (Pause upload or rollback due to credential error); S5→S2 (Retry triggered by operations personnel).

[0195] Queue management also supports the following operations: Pause submission – rolls back the task currently in S3 state to S2 state, and new tasks will no longer be dequeued until submission resumes; Single task termination – forcibly stops the currently executing submission task and marks it as S5 state; System reset – clears all task states and temporary data, restoring to the initial state.

[0196] The persistence of state data employs a three-file round-robin strategy: first, the latest state is written to a temporary file (.tmp), then the current official file is renamed to a backup file (.bak), and finally, the temporary file is renamed to the official state file. This strategy ensures that if the writing process is unexpectedly interrupted at any time (such as a process crash or power outage), at least one complete and usable state file (either the official file or the backup file) exists, thus guaranteeing the atomicity and reliability of the state data.

[0197] VII. Real-time monitoring and visualization configuration

[0198] The real-time monitoring terminal 600 maintains communication with the data convergence node 300 via a WebSocket long connection, receiving and displaying the following information:

[0199] Status Overview: The system displays the health status of each component in a tabbed interface, including: the running status of the data receiving service (running / stopped); the status of the Web drive engine (started / not started / abnormal); the online status of each acquisition subsystem (online / offline, displayed separately according to the subsystem identifier); the current depth and capacity of the upload ready queue; and the execution progress of the currently submitted task.

[0200] Abnormal Event Push: When the following abnormal events are detected, an alarm notification will be pushed to the monitoring terminal immediately via WebSocket: identity credential error (incorrect or expired password); abnormal interruption of data receiving service; offline of the data acquisition subsystem;

[0201] Task List and Operations: Displays detailed information for all train tasks, including train number, current status, creation time, update time, source subsystem, error message, etc., and provides cancellation, termination, and retry operations;

[0202] Visual configuration orchestration: Operations personnel can configure and modify the following parameters online via the monitoring terminal: identity credentials (username, password) and credential list; data recipient name; registration information of the collection subsystem (identity identifier, name, required index number); timeout parameters (element wait timeout, page load timeout, transmission wait timeout, etc.); automatic cleanup policy (expiration time of completed tasks, expiration time of pending tasks, cleanup scope). Configuration changes take effect immediately, and changes to critical configurations (such as browser-related parameters) will trigger the automatic rebuilding of the Web drive engine.

[0203] Example 2: Reference Figures 1-7 This embodiment details a method for automatically collecting and submitting raw data from a multi-source subsystem in a wind tunnel test, specifically including the following steps:

[0204] S1. Collect test data through a multi-source data acquisition subsystem group. After each test run, attach the test run identifier and subsystem identifier to the raw data and transmit it asynchronously to the data convergence node via the test local area network in the form of a data stream.

[0205] Specifically: the train number identifier is used to uniquely identify a test train number, and the subsystem identifier is used to identify the data acquisition subsystem from which the data originates;

[0206] S2. Receive asynchronously transmitted data through the data convergence node, perform two-dimensional indexing and temporary storage according to train number identifier and subsystem identifier, and perform multi-source data completeness verification for each train number. When the multi-source data completeness verification passes, generate the pending submission directory for the current train number and write the corresponding train number task into the upload ready queue.

[0207] S3. Without modifying the existing cross-domain data transmission client and isolated network architecture, the cross-domain virtual interaction execution agent retrieves ready tasks from the upload ready queue, establishes an interactive session with the cross-domain transmission client in the isolated network environment through the Web drive engine manager, and controls the cross-domain transmission client to complete the upload of the directory to be submitted, the transmission status monitoring, and the sending operation to the pre-configured recipient.

[0208] S4. Update the current task status based on the result information returned by the cross-domain data transmission client, and complete the automatic aggregation and delivery of raw data from multiple subsystems.

[0209] Furthermore, in step S2, the two-dimensional index is constructed and temporarily stored through the data temporary storage management module;

[0210] The multi-source data completeness verification process is as follows: Based on a preset subsystem registry, the set of subsystems that the current train should reach is determined. For each subsystem, according to preset completeness rules, a set of expected indexes is generated through the completeness verification module. The set of received index numbers is compared with the set of expected indexes. The index numbers are used to uniquely identify data files of the same subsystem within the same train. Only when the set of received index numbers of all registered subsystems covers their respective set of expected indexes is the multi-source data completeness verification of the current train determined to be passed by the completeness verification module.

[0211] The task queue management module uses a five-state machine to manage train tasks and schedules train tasks in the data ready state into the upload ready queue. When a voucher error is detected or a pause submission instruction is received, the task currently in the submission state is rolled back to the data ready state, and subsequent tasks are prevented from leaving the queue.

[0212] The process of generating the pending submission directory for the current train number simultaneously handles filename conflict, as follows:

[0213] When flattening the data files of each subsystem from the hierarchical temporary storage directory to a unified upload working directory, a subsystem identifier and index number prefix are added to the beginning of each filename to ensure that the flattened filenames are unique within the working directory; the states of the five-state machine are as follows:

[0214] The "Waiting for Data" status indicates that the multi-source data for the current train has not yet all arrived.

[0215] The data is ready status, which indicates that the multi-source data completeness verification has passed;

[0216] The "Submission in Progress" status indicates that the cross-domain virtual interaction execution agent is performing the submission operation for the current train.

[0217] The "Completed" status indicates that the current train data has been successfully submitted and verified.

[0218] A failure status indicates that the submission operation was terminated abnormally.

[0219] Among them, train tasks in the data ready state enter the upload ready queue to wait for submission, and a five-state machine is used to complete the pause and resume submission, single task forced termination, and failed task re-entry into the queue.

[0220] When data from a subsystem arrives after the corresponding train task has entered the data ready state, if the current train task has not yet started submission, the temporary data is updated and the completeness check is re-executed; if the current train task is already in the submission or completed state, the current data is named late data, marked as ignored and logged.

[0221] When a data transmission request carrying a manual upload flag is detected (see Manual Upload Mode), and the current request contains a data transmission end flag for the current train, the data convergence node directly marks the corresponding task as data ready without waiting for data from other subsystems to arrive, and does not perform standard multi-source data completeness checks.

[0222] The procedures for pausing and resuming submissions, and forcibly terminating single tasks are as follows:

[0223] When the data convergence node receives a stop signal, it stops retrieving new tasks from the upload ready queue, rolls back the tasks currently in the submission state to the data ready state, sends a stop notification to all registered acquisition subsystems, shuts down the Web drive engine, and persists the current queue state and all task states to the state file.

[0224] When the data convergence node restarts, it restores the task queue and task status from the status file and continues to process unfinished submission tasks.

[0225] The state data of the five-state machine adopts an atomic persistence strategy, as follows:

[0226] Write the current state to a temporary file;

[0227] Rename the original status file to a backup file;

[0228] Rename the temporary file to a formal status file;

[0229] When a corrupted official status file is detected, it will be automatically restored from a backup file.

[0230] Furthermore, the preset completeness rules include a static determination mode and a dynamic determination mode, and the specific process is as follows:

[0231] In the static determination mode, the data convergence node pre-configures a list of required index numbers for each acquisition subsystem (steady-state acquisition terminal 101, dynamic pressure acquisition terminal 102, scanning valve pressure acquisition terminal 103, and video measurement acquisition terminal 104). The expected index set is the list of required index numbers. When the received index number set covers all the index numbers in the list of required index numbers, the data of the current subsystem is determined to be complete.

[0232] In the dynamic judgment mode, when the data acquisition subsystem transmits data, it carries the actual total amount of data for the current train number N. The data convergence node generates a set of expected index numbers from 1 to N based on the actual total amount of data N. When the set of received index numbers after deduplication covers the set of expected index numbers, it is determined that the data for the current train in the current subsystem is complete.

[0233] When a train needs to be resubmitted, the data convergence node creates a new task version for the current train, and the version number is incremented. The status of the original version task remains unchanged, and the new version task independently enters the waiting data state and performs completeness verification and submission according to the standard process, thus completing the resubmission version management.

[0234] The data convergence node performs duplicate data detection when receiving data, as follows:

[0235] When a subsystem repeatedly submits data with the same index number for the same train, it calculates the hash value between the newly arrived data and the temporarily stored data. If the hashes match, the duplicate data is ignored; if the hashes do not match, it is recorded as a data anomaly event and an alarm notification is sent to the monitoring terminal.

[0236] When the data convergence node starts the data receiving service, it sends a start notification to all registered acquisition subsystems. When the service stops or the system is reset, it sends a stop or reset notification to all registered acquisition subsystems. The linkage notification has an independent timeout control to complete the service linkage control (see Linkage Control of Acquisition Subsystems for details).

[0237] Furthermore, in S3, the operation of the cross-domain virtual interaction execution agent is decomposed into multiple atomic execution steps, each with an independent retry mechanism, as follows:

[0238] If a certain execution step fails, the current step will be automatically retried within a preset number of retries;

[0239] When an interactive session failure is detected, the Web Drive Engine Manager is automatically reinitialized and execution is resumed through authentication.

[0240] Set segmented waiting periods between adjacent execution steps, and detect a termination event flag in each waiting segment;

[0241] When the termination event flag is detected to be set, the current task is aborted and the task status is updated to failure;

[0242] The Web Drive Engine Manager employs a reuse strategy, as detailed below:

[0243] Lazily initialize the Web Drive Engine Manager when the first task arrives;

[0244] Subsequent tasks reuse existing driver engine instances, using isolation mechanisms to distinguish the operational contexts of different tasks;

[0245] Each time a driver engine instance is obtained, the driver process is checked to see if it is alive and can respond to preset test operations. If it is alive and responds, the test is considered normal; otherwise, the test is considered abnormal, the existing instance is destroyed and the driver engine instance is rebuilt.

[0246] When the cross-domain virtual interaction execution agent detects a password expiration prompt during the process of controlling the cross-domain data transmission client, it selects the next set of credentials from the preset credential list to complete the password update, and persistently stores the current valid credentials after the update is successful.

[0247] Before uploading the directory to be submitted, the cross-domain virtual interaction execution agent performs environment setup operations, as follows:

[0248] The Web-driven engine controls the client to clean up residual data from past submissions and reset the submission session context, ensuring that the current submission operation is not affected by historical data.

[0249] Specifically, based on Example 1, this embodiment describes the implementation of dynamic completeness determination and fault tolerance enhancement for scenarios where the data acquisition subsystem may generate different numbers of data files in different train trips, and scenarios where it is necessary to further enhance the reliability of data delivery.

[0250] I. Dynamic Completeness Determination Model

[0251] In Example 1, the required index numbers for each subsystem are statically configured in the system registry. However, in certain test scenarios, the same subsystem may generate different amounts of data in different train runs. For example, in some special tests, the dynamic acquisition subsystem may collect separate results for each angle in the same train run, resulting in variations in the number of outputs for each train run.

[0252] To adapt to this scenario, when the data acquisition subsystem transmits data to the data convergence node, it can include the actual total data volume N of the current train in the transmission request. After receiving this parameter, the data convergence node generates a set of expected index numbers from 1 to N based on the actual total data volume N. When the deduplicated set of received index numbers covers the expected set of index numbers, it is determined that the data of the current train in the subsystem is complete. This dynamic determination mode no longer uses the list of required index numbers statically configured in the registry.

[0253] For example, if the dynamic pressure acquisition system reports total_indices_dynamic = 6 in train 2601010001a, the data convergence node expects the system to transmit 6 data files (indices 1 to 6) in this train. Once all 6 files arrive, the data from this subsystem in this train is considered complete. However, in train 2601010002a, the system reports total_indices_dynamic = 4, so only 4 files (indices 1 to 4) are needed to satisfy the completeness condition.

[0254] Dynamic and static decision-making modes can be used in combination: static configuration is used for subsystems with a fixed number of data outputs, while dynamic reporting is used for subsystems with a variable number of outputs. During verification, data convergence nodes prioritize dynamic reporting; if no dynamic reporting is received, they revert to static configuration.

[0255] II. Manual Upload Mode

[0256] In certain special cases (such as system debugging and retransmission of historical data), it is necessary to support direct data submission by ignoring standard multi-source completeness checks. To this end, the data acquisition subsystem can include a "manual upload" flag in its transmission request. When the data convergence node detects the manual upload flag, upon receiving a request from the acquisition subsystem carrying a data transmission completion flag for the current train, it directly marks the corresponding task as "data ready" without waiting for data from other subsystems. Tasks in manual upload mode will be displayed in the monitoring terminal with a special type identifier ("manual") to distinguish them from tasks in automatic mode.

[0257] III. Adaptive Calculation of Transmission Timeout

[0258] When the data acquisition subsystem transmits data to the data convergence node, the transmission timeout is not a fixed value, but is dynamically calculated based on the total size of the data transmitted in that session. The calculation formula is as follows:

[0259]

[0260] in, Set the base timeout (e.g., 30 seconds). This represents the total size (in bytes) of the data transmitted in this transaction. The minimum expected transfer rate (bytes / second, configurable, such as 2MB / s) is set, and the margin is an additional margin of time (such as 10 seconds). The adaptive calculation of the transfer timeout ensures that large data transfers will not be mistakenly judged as failures due to fixed timeouts, while small data transfers will not have excessively long allowed timeouts.

[0261] IV. Linkage Control of the Data Acquisition Subsystem

[0262] When the data convergence node starts the data receiving service, it sends a "start" notification to all registered acquisition subsystems; when it stops the service or performs a system reset, it sends a "stop" or "reset" notification. Upon receiving the notification, the acquisition subsystem starts or stops its local data output accordingly. This linkage mechanism ensures the consistency of the entire data flow chain and prevents the acquisition subsystem from still transmitting data when the convergence node is not ready.

[0263] The linkage notification has independent timeout control: notification timeout for a single collection subsystem (e.g., 8 seconds) and total notification timeout for all collection subsystems (e.g., 15 seconds). After the timeout, the start and stop operation of the main service will not be blocked, but the timeout information will be recorded and the monitoring terminal will be notified via WebSocket.

[0264] Example 3: Reference Figures 1-7 This embodiment details a method for automatically collecting and submitting raw data from a multi-source subsystem in a wind tunnel test, specifically including the following steps:

[0265] S1. Collect test data through a multi-source data acquisition subsystem group. After each test run, attach the test run identifier and subsystem identifier to the raw data and transmit it asynchronously to the data convergence node via the test local area network in the form of a data stream.

[0266] Specifically: the train number identifier is used to uniquely identify a test train number, and the subsystem identifier is used to identify the data acquisition subsystem from which the data originates;

[0267] S2. Receive asynchronously transmitted data through the data convergence node, perform two-dimensional indexing and temporary storage according to train number identifier and subsystem identifier, and perform multi-source data completeness verification for each train number. When the multi-source data completeness verification passes, generate the pending submission directory for the current train number and write the corresponding train number task into the upload ready queue.

[0268] S3. Without modifying the existing cross-domain data transmission client and isolated network architecture, the cross-domain virtual interaction execution agent retrieves ready tasks from the upload ready queue, establishes an interactive session with the cross-domain transmission client in the isolated network environment through the Web drive engine manager, and controls the cross-domain transmission client to complete the upload of the directory to be submitted, the transmission status monitoring, and the sending operation to the pre-configured recipient.

[0269] S4. Update the current task status based on the result information returned by the cross-domain data transmission client, and complete the automatic aggregation and delivery of raw data from multiple subsystems.

[0270] Furthermore, in step S2, the two-dimensional index is constructed and temporarily stored through the data temporary storage management module;

[0271] The multi-source data completeness verification process is as follows: Based on a preset subsystem registry, the set of subsystems that the current train should reach is determined. For each subsystem, according to preset completeness rules, a set of expected indexes is generated through the completeness verification module. The set of received index numbers is compared with the set of expected indexes. The index numbers are used to uniquely identify data files of the same subsystem within the same train. Only when the set of received index numbers of all registered subsystems covers their respective set of expected indexes is the multi-source data completeness verification of the current train determined to be passed by the completeness verification module.

[0272] The task queue management module uses a five-state machine to manage train tasks and schedules train tasks in the data ready state into the upload ready queue. When a voucher error is detected or a pause submission instruction is received, the task currently in the submission state is rolled back to the data ready state, and subsequent tasks are prevented from leaving the queue.

[0273] The process of generating the pending submission directory for the current train number simultaneously handles filename conflict, as follows:

[0274] When flattening the data files of each subsystem from the hierarchical temporary storage directory to a unified upload working directory, a subsystem identifier and index number prefix are added to the beginning of each filename to ensure that the flattened filenames are unique within the working directory; the states of the five-state machine are as follows:

[0275] The "Waiting for Data" status indicates that the multi-source data for the current train has not yet all arrived.

[0276] The data is ready status, which indicates that the multi-source data completeness verification has passed;

[0277] The "Submission in Progress" status indicates that the cross-domain virtual interaction execution agent is performing the submission operation for the current train.

[0278] The "Completed" status indicates that the current train data has been successfully submitted and verified.

[0279] A failure status indicates that the submission operation was terminated abnormally.

[0280] Among them, train tasks in the data ready state enter the upload ready queue to wait for submission, and a five-state machine is used to complete the pause and resume submission, single task forced termination, and failed task re-entry into the queue.

[0281] When data from a subsystem arrives after the corresponding train task has entered the data ready state, if the current train task has not yet started submission, the temporary data is updated and the completeness check is re-executed; if the current train task is already in the submission or completed state, the current data is named late data, marked as ignored and logged.

[0282] When a data transmission request carrying a manual upload flag is detected, and the current request contains a data transmission end flag for the current train, the data convergence node directly marks the corresponding task as data ready without waiting for data to arrive from other subsystems, and does not perform standard multi-source data completeness verification.

[0283] The procedures for pausing and resuming submissions, and forcibly terminating single tasks are as follows:

[0284] When the data convergence node receives a stop signal, it stops retrieving new tasks from the upload ready queue, rolls back the tasks currently in the submission state to the data ready state, sends a stop notification to all registered acquisition subsystems, shuts down the Web drive engine, and persists the current queue state and all task states to the state file.

[0285] When the data convergence node restarts, it restores the task queue and task status from the status file and continues to process unfinished submission tasks.

[0286] The state data of the five-state machine adopts an atomic persistence strategy, as follows:

[0287] Write the current state to a temporary file;

[0288] Rename the original status file to a backup file;

[0289] Rename the temporary file to a formal status file;

[0290] When a corrupted official status file is detected, it will be automatically restored from a backup file.

[0291] Furthermore, the preset completeness rules include a static determination mode and a dynamic determination mode, and the specific process is as follows:

[0292] In the static determination mode, the data convergence node pre-configures a list of required index numbers for each acquisition subsystem (steady-state acquisition terminal 101, dynamic pressure acquisition terminal 102, scanning valve pressure acquisition terminal 103, and video measurement acquisition terminal 104). The expected index set is the list of required index numbers. When the received index number set covers all the index numbers in the list of required index numbers, the data of the current subsystem is determined to be complete.

[0293] In the dynamic judgment mode, when the data acquisition subsystem transmits data, it carries the actual total amount of data for the current train number N. The data convergence node generates a set of expected index numbers from 1 to N based on the actual total amount of data N. When the set of received index numbers after deduplication covers the set of expected index numbers, it is determined that the data for the current train in the current subsystem is complete.

[0294] When a train needs to be resubmitted, the data convergence node creates a new task version for the current train, and the version number is incremented. The status of the original version task remains unchanged, and the new version task independently enters the waiting data state and performs completeness verification and submission according to the standard process, thus completing the resubmission version management.

[0295] The data convergence node performs duplicate data detection when receiving data, as follows:

[0296] When a subsystem repeatedly submits data with the same index number for the same train, it calculates the hash value between the newly arrived data and the temporarily stored data. If the hashes match, the duplicate data is ignored; if the hashes do not match, it is recorded as a data anomaly event and an alarm notification is sent to the monitoring terminal.

[0297] When the data convergence node starts the data receiving service, it sends a start notification to all registered acquisition subsystems. When the service stops or the system is reset, it sends a stop or reset notification to all registered acquisition subsystems. The linkage notification has an independent timeout control to complete the service linkage control.

[0298] Furthermore, in S3, the operation of the cross-domain virtual interaction execution agent is decomposed into multiple atomic execution steps, each with an independent retry mechanism, as follows:

[0299] If a certain execution step fails, the current step will be automatically retried within a preset number of retries;

[0300] When an interactive session failure is detected, the Web Drive Engine Manager is automatically reinitialized and execution is resumed through authentication.

[0301] Set segmented waiting periods between adjacent execution steps, and detect a termination event flag in each waiting segment;

[0302] When the termination event flag is detected to be set, the current task is aborted and the task status is updated to failure;

[0303] The Web Drive Engine Manager employs a reuse strategy, as detailed below:

[0304] Lazily initialize the Web Drive Engine Manager when the first task arrives;

[0305] Subsequent tasks reuse existing driver engine instances, using isolation mechanisms to distinguish the operational contexts of different tasks;

[0306] Each time a driver engine instance is obtained, the driver process is checked to see if it is alive and can respond to preset test operations. If it is alive and responds, the test is considered normal; otherwise, the test is considered abnormal, the existing instance is destroyed and the driver engine instance is rebuilt.

[0307] When the cross-domain virtual interaction execution agent detects a password expiration prompt during the process of controlling the cross-domain data transmission client, it selects the next set of credentials from the preset credential list to complete the password update, and persistently stores the current valid credentials after the update is successful.

[0308] Before uploading the directory to be submitted, the cross-domain virtual interaction execution agent performs environment setup operations, as follows:

[0309] The Web-driven engine controls the client to clean up residual data from past submissions and reset the submission session context, ensuring that the current submission operation is not affected by historical data.

[0310] Specifically, based on Examples 1 and 2, Example 3 describes an optimized implementation of the system under continuous train delivery for the scenario of high-intensity continuous operation of wind tunnel tests.

[0311] I. Cross-task reuse optimization of Web-driven engines

[0312] In scenarios involving large-volume, continuous submissions, restarting and closing the browser process for each submission would incur significant time overhead (a single browser startup typically takes 3-5 seconds). Web Drive Engine Manager 401 employs the following reuse optimization strategy:

[0313] (1) Persistent driver instance: The browser driver process continues to run after its initial creation until it receives an explicit shutdown command or detects an anomaly.

[0314] (2) Operation Context Isolation: Each new task isolates the interaction state of different tasks by creating an independent operation context, and releases the context after the task is completed. The manager periodically cleans up redundant context resources.

[0315] (3) Session data cleanup: Before each new task begins, clear the residual session data in the browser to prevent the login session of the previous task from affecting the identity authentication of the current task.

[0316] (4) Hot configuration update: When the operation and maintenance personnel modify the browser-related configuration parameters (such as window size, headless mode switching, etc.) through the monitoring terminal, the manager automatically destroys the current driver instance and recreates it, and the new configuration takes effect immediately.

[0317] By employing the aforementioned reuse strategy, in scenarios where multiple train requests are submitted consecutively, the time overhead caused by repeatedly launching the browser can be significantly reduced compared to creating a new browser instance each time.

[0318] II. Queue Depth Control and Back Pressure Mechanism

[0319] The upload ready queue is configured with a maximum depth limit (100 in this embodiment). When the queue depth reaches the upper limit, newly completed train tasks will be queued and wait until there is an empty space in the queue. This mechanism prevents memory resources from being consumed indefinitely when the submission speed is lower than the data arrival speed.

[0320] III. Automatic Cleanup of Expired Tasks

[0321] To prevent the temporary storage directory and task records from growing indefinitely after long-term operation, the data convergence node performs automatic cleanup operations periodically. The cleanup strategy is configured via a configuration file and includes:

[0322] Cleanup timeout for completed tasks (e.g., 3 hours): Tasks with a status of "completed" that have exceeded 3 hours will have their status records and temporary files automatically deleted.

[0323] Task cleanup timeout (e.g., 3 hours): Tasks with a status of "waiting for data" that have not met the completeness requirements after 3 hours are considered abnormal data and will be automatically cleaned up.

[0324] The cleanup scope can be configured to clean up only the status records or clean up both the status records and physical files.

[0325] IV. Shutdown and Status Maintenance

[0326] When the data convergence node receives a shutdown signal (such as when maintenance personnel manually stop the service or when system maintenance is performed), the following shutdown procedure is executed:

[0327] (1) Stop retrieving new tasks from the ready queue;

[0328] (2) If there is a submission task currently being executed, notify the cross-domain virtual interaction execution agent to stop the current operation and roll back the task status to "data ready" (READY) to ensure that it can be re-executed on the next startup;

[0329] (3) Send a "stop" notification to all registered data collection subsystems;

[0330] (4) Disable the Web drive engine (set a timeout limit, and force the browser process to terminate after the timeout).

[0331] (5) Persist the current queue state and all task states to the state file.

[0332] On the next startup, if persistent recovery is enabled, the previous task queue and state will be restored from the state file to continue processing unfinished tasks.

[0333] V. Page Element Location Strategy for Cross-Domain Transmission Client Interaction

[0334] During the interaction between the cross-domain virtual interaction execution agent and the cross-domain data transmission client, the interface interaction executor 403 uses pre-configured element location rules for interaction control. The location rules for all target interactive elements are centrally maintained in a unified location configuration, covering various interactive controls such as identity authentication, file management, transmission status, directory upload and send operations;

[0335] Element location employs a waiting and polling mechanism: within the configured maximum waiting time (e.g., 12 seconds), it repeatedly attempts to locate the target element with short polling intervals until the element appears and meets the interactive conditions. If the target element is not found after the maximum waiting time, the retry logic for this step is triggered.

[0336] VI. Deployment Plan for Data Output Function of Acquisition Terminal

[0337] The data output function of each acquisition terminal is designed as an independently configurable, lightweight functional module, which can be deployed as an extension component of the acquisition software. The data output functional module and its runtime dependencies are packaged into a single executable program using a packaging tool. Combined with a separate configuration file, this allows for rapid deployment on each acquisition terminal. The configuration file includes the following key parameters: network identifier and port of the data convergence node, local service listening port, data marshalling directory path, transmission timeout parameters, retry policy parameters, state persistence switch, and automatic cleanup policy.

[0338] The data output module provides a status query interface, which supports remote querying of operational metrics such as the current task queue depth, the number of completed tasks, and the number of failed tasks, making it easier for maintenance personnel to monitor the data output status of each acquisition terminal.

[0339] Examples 1-3 fully describe the technical solution of the present invention from three aspects: basic functions, enhanced flexibility, and high throughput optimization. Those skilled in the art can select appropriate implementation methods or combine the features of multiple examples according to the specific wind tunnel test facility configuration and operation requirements.

[0340] Example 4: Reference Figures 1-7 This embodiment describes a system for automatically collecting and submitting raw data from a multi-source subsystem of a wind tunnel test. This system is used to execute the method for automatically collecting and submitting raw data from a multi-source subsystem of a wind tunnel test as described in any one of embodiments 1-3. It includes a multi-source data acquisition subsystem group 100, a data convergence node 300, a cross-domain virtual interaction execution agent 400, a cross-domain transmission client 500, and a real-time monitoring terminal 600.

[0341] The multi-source data acquisition subsystem group 100 is connected to the data convergence node 300, the data convergence node 300 is connected to the cross-domain virtual interaction execution agent 400, the data convergence node 300 is connected to the real-time monitoring terminal 600, and the cross-domain virtual interaction execution agent 400 is connected to the cross-domain transmission client 500.

[0342] Furthermore, the multi-source data acquisition subsystem group 100 includes a steady-state acquisition terminal 101, a dynamic pressure acquisition terminal 102, a scanning valve pressure acquisition terminal 103, and a video measurement acquisition terminal 104;

[0343] The data convergence node 300 includes a security access control module 301, a data temporary storage management module 302, a completeness verification module 303, and a task queue management module 304 connected in sequence.

[0344] The cross-domain virtual interaction execution agent 400 includes a Web drive engine manager 401, a step arranger 402, and a user interface interaction executor 403 connected in sequence.

[0345] The steady-state acquisition terminal 101, dynamic pressure acquisition terminal 102, scanning valve pressure acquisition terminal 103, and video measurement acquisition terminal 104 are all connected to the security access control module 301. The task queue management module 304 is connected to the step arranger 402. The interface interaction actuator 403 is connected to the cross-domain transmission client 500. The security access control module 301 and the task queue management module 304 are connected to the real-time monitoring terminal 600 through node control and status aggregation operations, respectively.

[0346] Specifically, the multi-source data acquisition subsystem group 100 includes four acquisition terminals deployed at the wind tunnel test site: steady-state acquisition terminal 101, dynamic pressure acquisition terminal 102, scanning valve pressure acquisition terminal 103, and video measurement acquisition terminal 104. Each acquisition terminal is connected to the corresponding measurement equipment, runs its own dedicated data acquisition software, and has the function of asynchronously transmitting the raw data with train number and subsystem identifier attached to it as a data stream to the data convergence node after the train is completed. This data output function runs as an independent process on the corresponding acquisition terminal, without interfering with the normal operation of the acquisition software. It can be packaged into an independent executable program for easy deployment on acquisition terminals in different operating system environments. Among them, the steady-state acquisition terminal 101 is connected to a strain balance to acquire strain signals of the model under stress; the dynamic pressure acquisition terminal 102 is connected to a high-frequency dynamic pressure sensor to acquire unsteady dynamic pressure data of the model; the scanning valve pressure acquisition terminal 103 is connected to a multi-channel electronic pressure scanning valve to synchronously acquire the pressure at each pressure measurement point on the model surface; the video measurement acquisition terminal 104 is connected to an industrial camera measurement device to record images of the model in the airflow through optical video means for subsequent calculation of the actual attitude angle; the acquisition terminals are connected to different types of measurement devices, including at least one of a balance force measuring device, a pressure scanning device (scanning valve measurement device or dynamic pressure measurement device), and a video measurement device;

[0347] refer to Figure 7 To facilitate the transmission of data from the acquisition subsystem terminal to the convergence node, this application also includes a local data receiving module, a data marking module, and an asynchronous transmission module, as detailed below:

[0348] The local data receiving module is used to monitor the data output interface of the acquisition terminal and receive the raw train data.

[0349] The data tagging module is used to add train number identifier, subsystem identifier and index number information to the original train number data;

[0350] The asynchronous transmission module is used to transmit the marked data in batches to the data convergence node in a non-blocking manner, and has the ability to adaptively calculate the transmission timeout and retransmit in case of failure.

[0351] The data convergence node 300 is a dedicated computer device that connects to four acquisition terminals via the experimental local area network. The data convergence node 300 runs a data receiving service, which is responsible for receiving data transmitted from the four acquisition terminals, performing multi-source data integrity verification, and managing the upload task queue.

[0352] The secure access control module 301 is used to verify the access legitimacy of the collection subsystem based on the subsystem registry. It maintains a registry of allowed subsystems, which records the identity information and a list of required index numbers for each subsystem. Only the subsystems recorded in the registry are allowed to transmit data to the data convergence node.

[0353] The data temporary storage management module 302 is used to establish a two-dimensional index structure based on train number identifier and subsystem identifier to temporarily store the received data;

[0354] The completeness verification module 303 is used to generate a set of expected indexes according to preset completeness rules and to determine whether the train data of each subsystem is complete.

[0355] The task queue management module 304 is used to maintain train tasks with five states, and to schedule tasks in the data ready state to enter the upload ready queue.

[0356] The cross-domain virtual interaction execution agent 400, deployed on the data convergence node 300, includes a Web driver engine manager 401, a step orchestrator 402, and a user interface interaction executor 403. The Web driver engine manager 401 is responsible for managing the lifecycle of the browser driver, adopting lazy initialization and cross-task reuse strategies. It distinguishes the operation context of different tasks through an isolation mechanism and checks whether the driver process is alive and can respond to preset test operations each time an instance is obtained. It automatically rebuilds the process when an anomaly is detected. The step orchestrator 402 is responsible for orchestrating and controlling the execution flow of the submission operation, decomposing the submission operation into atomic steps such as identity authentication, pre-submission environment preparation, uploading the directory to be submitted, transmission waiting, sending to the pre-configured recipient, and result verification. It independently performs retry control for each step. The user interface interaction executor 403 is responsible for performing specific interaction operations with the web interface of the cross-domain data transmission client, that is, locating the target control according to the page state and interactive element identifier and executing the interface interaction control operation.

[0357] Cross-domain transmission client 500, also known as cross-domain data transmission client in isolated network environment, is deployed at the network isolation boundary and provides cross-domain data transmission service based on web interface. Cross-domain virtual interaction execution agent 400 interacts with cross-domain transmission client 500 through the browser on data convergence node 300.

[0358] The real-time monitoring terminal 600 can be a web browser client deployed on any computer within the experimental local area network. It communicates with the data convergence node via WebSocket and is used to receive and display status information and abnormal alarms. Specifically, it communicates with the data convergence node 300 through a long WebSocket connection to receive and display the system's operating status in real time. The real-time monitoring terminal 600 includes a status overview display module, a configuration orchestration module, and a task management module. The status overview display module is used to display the service operating status, the driving engine status, the online status of each acquisition subsystem, the queue depth, and the current submission progress. The configuration orchestration module is used to configure identity credentials, data recipients, subsystem registration information, timeout parameters, and automatic cleanup strategies online. The task management module is used to display a list of all submitted tasks and their detailed status, and supports canceling, terminating, or retrying tasks.

[0359] Although the invention has been described with reference to a limited number of embodiments, those skilled in the art will understand from the foregoing description that other embodiments are conceivable within the scope of the invention described herein. Furthermore, it should be noted that the language used in this specification has been chosen primarily for readability and instructional purposes, and not for the purpose of interpreting or limiting the subject matter of the invention. Therefore, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure of the invention is illustrative and not restrictive, and the scope of the invention is defined by the appended claims.

Claims

1. A method for automatically collecting and submitting raw data from a multi-source subsystem in a wind tunnel test, characterized in that, Includes the following steps: S1. Collect test data through a multi-source data acquisition subsystem group. After each test run, attach the test run identifier and subsystem identifier to the raw data and transmit it asynchronously to the data convergence node via the test local area network in the form of a data stream. S2. Receive asynchronously transmitted data through the data convergence node, perform two-dimensional indexing and temporary storage according to train number identifier and subsystem identifier, and perform multi-source data completeness verification for each train number. When the multi-source data completeness verification passes, generate the pending submission directory for the current train number and write the corresponding train number task into the upload ready queue. S3. Obtain ready tasks from the upload ready queue through the cross-domain virtual interaction execution agent, establish an interactive session with the cross-domain transmission client in the isolated network environment through the Web drive engine manager, and control the cross-domain transmission client to complete the upload of the directory to be submitted, the transmission status monitoring, and the sending operation to the pre-configured recipient; S4. Update the current task status based on the result information returned by the cross-domain data transmission client, and complete the automatic aggregation and delivery of raw data from multiple subsystems.

2. The method for automatic aggregation and submission of raw data from a multi-source subsystem in a wind tunnel test according to claim 1, characterized in that, In step S2, the two-dimensional index is constructed and temporarily stored through the data temporary storage management module. The multi-source data completeness verification process is as follows: Based on the preset subsystem registry, the set of subsystems that the current train should arrive at is determined. For each subsystem, according to the preset completeness rules, a set of expected indexes is generated through the completeness verification module. The set of received index numbers is compared with the set of expected indexes. Only when the set of received index numbers of all registered subsystems covers their respective set of expected indexes, the completeness verification module determines that the multi-source data completeness verification of the current train has passed. The task queue management module uses a five-state machine to manage train tasks and schedules train tasks in the data ready state into the upload ready queue. When a voucher error is detected or a pause submission instruction is received, the task currently in the submission state is rolled back to the data ready state, and subsequent tasks are prevented from leaving the queue. The process of generating the pending submission directory for the current train number simultaneously handles filename conflict, as follows: When flattening the data files of each subsystem from the hierarchical temporary storage directory to a unified upload working directory, a subsystem identifier and index number prefix are added to the beginning of each filename to ensure that the flattened filenames are unique within the working directory; the states of the five-state machine are as follows: The "Waiting for Data" status indicates that the multi-source data for the current train has not yet all arrived. The data is ready status, which indicates that the multi-source data completeness verification has passed; The "Submission in Progress" status indicates that the cross-domain virtual interaction execution agent is performing the submission operation for the current train. The "Completed" status indicates that the current train data has been successfully submitted and verified. A failure status indicates that the submission operation was terminated abnormally. Among them, train tasks in the data ready state enter the upload ready queue to wait for submission, and a five-state machine is used to complete the pause and resume submission, single task forced termination, and failed task re-entry into the queue. When data from a subsystem arrives after the corresponding train task has entered the data ready state, if the current train task has not yet started submission, the temporary data is updated and the completeness check is re-executed; if the current train task is already in the submission or completed state, the current data is named late data, marked as ignored and logged. When a data transmission request carrying a manual upload flag is detected, and the current request contains a data transmission end flag for the current train, the data convergence node directly marks the corresponding task as data ready without waiting for data to arrive from other subsystems, and does not perform standard multi-source data completeness verification. The procedures for pausing and resuming submissions, and forcibly terminating single tasks are as follows: When the data convergence node receives a stop signal, it stops retrieving new tasks from the upload ready queue, rolls back the tasks currently in the submission state to the data ready state, sends a stop notification to all registered acquisition subsystems, shuts down the Web drive engine, and persists the current queue state and all task states to the state file. When the data convergence node restarts, it restores the task queue and task status from the status file and continues to process unfinished submission tasks. The state data of the five-state machine adopts an atomic persistence strategy, as follows: Write the current state to a temporary file; Rename the original status file to a backup file; Rename the temporary file to a formal status file; When a corrupted official status file is detected, it will be automatically restored from a backup file.

3. The method for automatic aggregation and submission of raw data from a multi-source subsystem in a wind tunnel test according to claim 2, characterized in that, The preset completeness rules include a static judgment mode and a dynamic judgment mode, and the specific process is as follows: In the static determination mode, the data convergence node pre-configures a list of required index numbers for each acquisition subsystem, and the expected index set is the list of required index numbers. When the received index number set covers all the index numbers in the list of required index numbers, the data of the current subsystem is determined to be complete. In the dynamic judgment mode, when the data acquisition subsystem transmits data, it carries the actual total amount of data for the current train number N. The data convergence node generates a set of expected index numbers from 1 to N based on the actual total amount of data N. When the set of received index numbers after deduplication covers the set of expected index numbers, it is determined that the data for the current train in the current subsystem is complete. When a train needs to be resubmitted, the data convergence node creates a new task version for the current train, and the version number is incremented. The status of the original version task remains unchanged, and the new version task independently enters the waiting data state and performs completeness verification and submission according to the standard process, thus completing the resubmission version management. The data convergence node performs duplicate data detection when receiving data, as follows: When a subsystem repeatedly submits data with the same index number for the same train, it calculates the hash value between the newly arrived data and the temporarily stored data. If the hashes match, the duplicate data is ignored; if the hashes do not match, it is recorded as a data anomaly event and an alarm notification is sent to the monitoring terminal. When the data convergence node starts the data receiving service, it sends a start notification to all registered acquisition subsystems. When the service stops or the system is reset, it sends a stop or reset notification to all registered acquisition subsystems. The linkage notification has an independent timeout control to complete the service linkage control.

4. The method for automatic aggregation and submission of raw data from a multi-source subsystem in a wind tunnel test according to claim 2, characterized in that, In S3, the operation of the cross-domain virtual interaction execution agent is decomposed into multiple atomic execution steps, each with an independent retry mechanism, as follows: If a certain execution step fails, the current step will be automatically retried within a preset number of retries; When an interactive session failure is detected, the Web Drive Engine Manager is automatically reinitialized and execution is resumed through authentication. Set segmented waiting periods between adjacent execution steps, and detect a termination event flag in each waiting segment; When the termination event flag is detected to be set, the current task is aborted and the task status is updated to failure; The Web Drive Engine Manager employs a reuse strategy, as detailed below: Lazily initialize the Web Drive Engine Manager when the first task arrives; Subsequent tasks reuse existing driver engine instances, using isolation mechanisms to distinguish the operational contexts of different tasks; Each time a driver engine instance is obtained, the driver process is checked to see if it is alive and can respond to preset test operations. If it is alive and responds, the test is considered normal; otherwise, the test is considered abnormal, the existing instance is destroyed and the driver engine instance is rebuilt. When the cross-domain virtual interaction execution agent detects a password expiration prompt during the process of controlling the cross-domain data transmission client, it selects the next set of credentials from the preset credential list to complete the password update, and persistently stores the current valid credentials after the update is successful. Before uploading the directory to be submitted, the cross-domain virtual interaction execution agent performs environment setup operations, as follows: The Web-driven engine controls the client to clean up residual data from past submissions and reset the submission session context, ensuring that the current submission operation is not affected by historical data.

5. A system for automatically collecting and submitting raw data from a multi-source subsystem in a wind tunnel test, characterized in that: The method for automatically collecting and submitting raw data from a multi-source subsystem of a wind tunnel test as described in any one of claims 1-4 includes a multi-source data acquisition subsystem group (100), a data convergence node (300), a cross-domain virtual interaction execution agent (400), a cross-domain transmission client (500), and a real-time monitoring terminal (600). The multi-source data acquisition subsystem group (100) is connected to the data convergence node (300), the data convergence node (300) is connected to the cross-domain virtual interaction execution agent (400), the data convergence node (300) is connected to the real-time monitoring terminal (600), and the cross-domain virtual interaction execution agent (400) is connected to the cross-domain transmission client (500).

6. The system for automatic aggregation and submission of raw data from a multi-source subsystem in wind tunnel testing according to claim 5, characterized in that, The multi-source data acquisition subsystem group (100) includes a steady-state acquisition terminal (101), a dynamic pressure acquisition terminal (102), a scanning valve pressure acquisition terminal (103), and a video measurement acquisition terminal (104). The data convergence node (300) includes a security access control module (301), a data temporary storage management module (302), a completeness verification module (303), and a task queue management module (304) connected in sequence. The cross-domain virtual interaction execution agent (400) includes a Web drive engine manager (401), a step orchestrator (402), and a user interface interaction executor (403) connected in sequence. The steady-state acquisition terminal (101), dynamic pressure acquisition terminal (102), scanning valve pressure acquisition terminal (103), and video measurement acquisition terminal (104) are all connected to the security access control module (301), the task queue management module (304) is connected to the step arranger (402), the interface interaction actuator (403) is connected to the cross-domain transmission client (500), and the security access control module (301) and the task queue management module (304) are connected to the real-time monitoring terminal (600).