Performance test method of intelligent railway train fault image recognition system

By building performance test data sets and real-time monitoring modules, the performance of the railway train fault image intelligent recognition system is evaluated, and the accuracy and efficiency of traditional testing methods are solved, and high accuracy and efficient performance evaluation is achieved to ensure system stability and flexibility.

CN118968230APending Publication Date: 2024-11-15HARBIN KEJIA GENERAL MECHANICAL & ELECTRICAL CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202410935442.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

Traditional model performance testing methods are difficult to adapt to complex and changeable environments, resulting in inaccurate test results and cumbersome test process, and it is impossible to effectively evaluate the performance of the intelligent identification system for railway train fault images.

Method used

The railway train passing images of each passing station are collected as test samples, and the performance test data set is constructed. Passing tasks are generated through the train passing simulation module and performance monitoring module, and performance indicator data are monitored in real time, and performance trend charts are displayed through the visual platform. Perform multi-dimensional performance evaluation.

Benefits of technology

Achieve high accuracy and high efficiency performance testing, which can more reliably evaluate the performance of intelligent identification systems, reduce manual intervention, and be flexible and scalable, ensuring the system operates stably under various loads and environments, reducing release risks .

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968230A_ABST
    Figure CN118968230A_ABST
Patent Text Reader

Abstract

The invention discloses a performance test method of a railway train fault image intelligent identification system, and belongs to the technical field of intelligent identification model performance test. The method aims at solving the problem that an existing method cannot evaluate the performance of a railway train fault image intelligent recognition system. Comprising the steps that a train station-crossing simulation module is adopted to generate a train-crossing task according to a set train-crossing time interval, a train-crossing column number and target train basic information, a corresponding test sample in a performance test data set is read according to the train-crossing task, and the test sample is transmitted to a fault image intelligent recognition system; in the fault identification process of the fault image intelligent identification system, a performance monitoring module is adopted to monitor various performance index data of the server and the fault image intelligent identification system; and converting the received performance index data into a performance index trend chart in real time by adopting a visual platform, and displaying the performance index trend chart, wherein the performance index trend chart is used for performance analysis of the image intelligent identification system. The method is used for testing the performance of the intelligent identification system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a performance testing method for a railway train fault image intelligent recognition system, and belongs to the technical field of intelligent recognition model performance testing. Background Art

[0002] With the development of technology and the improvement of user needs, performance testing of model systems becomes more and more important.

[0003] In order to ensure the stability of the railway train fault image intelligent recognition system in actual operation, in addition to testing its functions and indicators, performance testing is also required. Due to its limitations, the traditional model performance testing method is difficult to adapt to complex and changing environments, and there are problems such as inaccurate test results and cumbersome test processes.

[0004] Therefore, it is necessary to provide a performance testing method for the railway train fault image intelligent recognition system. Summary of the invention

[0005] Aiming at the problem that the existing methods are unable to evaluate the performance of a railway train fault image intelligent recognition system, the present invention provides a performance testing method for a railway train fault image intelligent recognition system.

[0006] A performance testing method of a railway train fault image intelligent recognition system of the present invention,

[0007] Collect different railway train passing images of each passing station as test samples to construct a performance test data set; the test samples are configured with a train information table, including the name of the passing station, the passing time, the corresponding train basic information and the passing status;

[0008] The train passing simulation module is used to generate a passing task according to the set passing time interval, the number of passing trains and the basic information of the target train. The corresponding test samples in the performance test data set are read according to the passing task, and the test samples are transmitted to the fault image intelligent recognition system.

[0009] During the process of fault identification by the fault image intelligent identification system, the performance monitoring module is used to monitor various performance indicator data of the server and the fault image intelligent identification system;

[0010] The visualization platform is used to convert the received performance indicator data into a performance indicator trend chart in real time and display it. The performance indicator trend chart is used for performance analysis of the image intelligent recognition system.

[0011] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, the test sample types include normal images, missing images, oversized images, repeated images, overbright images, overdark images, storage failure images, missed images, timed images and program abnormality images.

[0012] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, the train passing task includes setting the performance test data set connection information, the train passing time interval, the test rounds, and the train basic information reading path; the train basic information is stored in the vehicle information table sql file.

[0013] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, the train passing simulation module modifies the passing time of the current target train in the train information table to the current system time according to the minimum passing time data of the target train, modifies the default receiving state of the passing state to the receiving completed state, and after the train passing is completed, modifies the reporting state of the target train in the train information table from the unreported state to the reported state.

[0014] According to the performance test method of the railway train fault image intelligent recognition system of the present invention, in the vehicle passing task, by querying the vehicle information table sql file, inserting the test sample according to the set vehicle sequence and the time interval of the vehicle passing;

[0015] The train passing simulation module copies the corresponding test samples to the FTP server path according to the train passing task for use by the fault image intelligent recognition system;

[0016] The train station passing simulation module executes the test task in a distributed manner and is implemented in Python language; each train station passing simulation module corresponds to a different server.

[0017] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, the performance monitoring module monitors the fault image intelligent recognition system script in real time and collects various performance indicator data of the server and the fault image intelligent recognition system every 30 seconds;

[0018] Server performance indicator data includes CPU usage, GPU usage, memory usage, GPU frequency, disk throughput, CPU temperature, GPU temperature, and network throughput;

[0019] The performance indicator data of the fault image intelligent recognition system include the number of software program threads and memory usage.

[0020] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, during the process of fault recognition by the fault image intelligent recognition system, a performance monitoring module is used to collect various performance indicator data in real time through shell scripts and python scripts;

[0021] Use the iostat command to obtain CPU usage, disk throughput, and task queue number; use the free command to obtain memory usage and cache usage; use the nvidia-smi command to obtain GPU usage, GPU temperature, and GPU frequency; use the ps command to obtain the memory usage and program thread number of the fault image intelligent recognition system; use the sar command to obtain network throughput;

[0022] Deploy the script of the performance monitoring module to the selected server, set the performance monitoring module to start automatically at boot, obtain performance indicator data at the set sampling time interval, and store the performance indicator data in the MySQL database.

[0023] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, in the performance monitoring module, an alarm threshold is set for each performance indicator data, and when the collected performance indicator data exceeds the alarm threshold, an alarm signal is generated and displayed on a visualization platform;

[0024] The performance monitoring module checks the log script program of the fault image intelligent recognition system according to the set time step. When an error or ret≠0 appears in the log, the current log information is sent to the visualization platform for display as the judgment data of the problem process status of the fault image intelligent recognition system;

[0025] The performance monitoring module monitors the rabbitmq queue message script in real time, uses rabbitmqctl to view the server rabbitmq status and consumption data in real time, and displays it on the visualization platform.

[0026] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, the performance index trend graph is a line drawing with the time axis as the horizontal axis and the performance index data item as the vertical axis;

[0027] Analyze the hardware resource usage trend and server performance of different manufacturers based on the performance indicator data corresponding to the server in the performance indicator trend chart;

[0028] Combined with the number of trains passing simultaneously corresponding to the passing time, and based on the corresponding performance indicator data in the performance indicator trend chart corresponding to the operation of the fault image intelligent recognition system, a multi-dimensional performance evaluation of the fault image intelligent recognition system is carried out.

[0029] According to the performance testing method of the railway train fault image intelligent recognition system of the present invention, the performance analysis of the image intelligent recognition system according to the performance index trend chart includes:

[0030] Evaluate the performance and performance bottlenecks of the fault image intelligent recognition system based on memory usage, GPU usage, disk throughput, and CPU usage;

[0031] Select and set a long-term performance indicator trend chart to evaluate the long-term operation stability of the fault image intelligent recognition system;

[0032] Select performance indicator trend graphs under different test environments and conditions to evaluate the repeatable performance of the fault image intelligent recognition system.

[0033] Beneficial effects of the present invention: The method of the present invention can be used for performance testing of intelligent recognition systems before release, has high test accuracy and test efficiency, and can ensure the stability of the test object, specifically:

[0034] By automatically generating test cases and performing real-time performance analysis, it is possible to more accurately simulate actual usage scenarios and thus more reliably evaluate the performance of the intelligent recognition system with high accuracy and reliability.

[0035] Through distributed testing methods, manual intervention and test execution time can be significantly reduced, saving time and human resource costs.

[0036] The design of the method of the present invention takes into account the development of future technologies and changes in demand, has high flexibility and scalability, and can match the test scenarios and requirements of various forms of railway train fault image intelligent recognition systems.

[0037] The method of the present invention can realize test optimization intelligently, and can adjust test parameters and strategies according to test results to achieve the best test effect, thereby improving test efficiency and accuracy.

[0038] The method of the present invention can realize comprehensive performance evaluation of the system, which not only covers server performance testing, but also includes testing to identify the performance of the system itself. According to the test results, it can ensure that the system runs stably under various loads and environments.

[0039] The method of the present invention can realize risk prediction and prevention of intelligent recognition systems. It predicts potential performance problems in advance through real-time performance analysis and intelligent optimization, and can guide measures to prevent faults, thereby reducing the risks of intelligent recognition systems after they are released. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a performance test flow chart of the performance test method of the railway train fault image intelligent recognition system of the present invention;

[0041] Figure 2 It is a flow chart of the simulated train passing station test;

[0042] Figure 3 This is the performance indicator monitoring page of the method of the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0045] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0046] Specific implementation method 1. Combination Figure 1 and Figure 2 As shown, the present invention provides a performance testing method for a railway train fault image intelligent recognition system, comprising:

[0047] Collect different railway train passing images of each passing station as test samples to construct a performance test data set; the test samples are configured with a train information table, including the name of the passing station, the passing time, the corresponding train basic information and the passing status;

[0048] The train passing simulation module is used to generate a passing task according to the set passing time interval, the number of passing trains and the basic information of the target train. The corresponding test samples in the performance test data set are read according to the passing task, and the test samples are transmitted to the fault image intelligent recognition system.

[0049] During the process of fault identification by the fault image intelligent identification system, the performance monitoring module is used to monitor various performance indicator data of the server and the fault image intelligent identification system;

[0050] The visualization platform is used to convert the received performance indicator data into a performance indicator trend chart in real time and display it. The performance indicator trend chart is used for performance analysis of the image intelligent recognition system.

[0051] This implementation method is divided into three parts: train passing simulation module, performance monitoring module and visualization platform. The train passing simulation module can flexibly simulate the necessary test data such as the time interval between trains passing each station, the number of trains passing at the same time, basic train information, image information, etc., and integrate the test data to send it to the image intelligent recognition system in the train passing task mode. During the train image recognition process, the performance monitoring module collects the performance indicator data of the server and software program during the performance test process, and according to the configuration, it can collect performance indicator data at regular intervals, send the collected performance data to the data visualization platform, and display the performance indicator trend chart during the task process on the visualization platform, so as to discover potential performance problems of the system.

[0052] Build a performance test dataset:

[0053] The performance test data set consists of image data sets from different manufacturers, different detection stations, and different vehicle models. The project data set has an independent directory, which is distinguished by the project name. Various situations in the simulation train acquisition process, as an example, the test sample types include normal images, missing images, oversized images, repeated images, overbright images, overdark images, storage failure images, missed images, timeout images, and program abnormal images.

[0054] In actual work, the on-site image data can be updated regularly every month and added to the performance test data set to ensure the comprehensiveness of the performance test data set, with a focus on adding abnormal image data sets such as missed reports, timeouts, and program anomalies that occur on site.

[0055] The train passing task is mainly used to simulate the software function of train passing through stations:

[0056] The function of the software for simulating train passing is mainly to read the passing time of different detection stations, basic train information and image information on the FTP server in the database, set the passing time interval, the number of trains passing at the same time, and generate the passing task.

[0057] Further, configuration file settings: the configuration file needs to be set accordingly according to the actual test requirements.

[0058] Combination Figure 2 As shown, the train passing task includes setting the performance test data set connection information, the train passing time interval, the test round, and the train basic information reading path; the train basic information is stored in the vehicle information table sql file.

[0059] As an example, set the database connection information, such as: "hostname = 192.168.0.1database = press_test"; set the time interval for passing the train, passing the train every 15 minutes, such as: "update_rounds_time_interval = 900s"; set the test rounds, representing how many rounds are required for this test, such as: "test_rounds = 100" means 100 rounds of testing and 100 passing of the train; set the vehicle information table sql file reading path to insert vehicle information data for subsequent simulation of train passing through the station.

[0060] Setting of train information table, read:

[0061] The data information of each train consists of different detection stations, passing time, passing status, etc.

[0062] In the process of train operation, this implementation method queries the passing station information and the passing serial number information, and queries the minimum passing time data in the database. The train passing simulation module modifies the passing time (pass_time) of the current target train in the train information table to the current system time according to the minimum passing time data of the target train, modifies the default receiving state of the passing state (pass_state) to the receiving state, and after the train passes, modifies the reporting state of the target train in the train information table from the unreported state to the reported state. Specific stations can be manually modified according to actual test requirements, or multi-station interspersed testing can be performed.

[0063] Going further, the basic information of the train vehicle table is inserted:

[0064] According to the passing station information and passing serial number information queried in the database, the vehicle information table sql file is queried in the passing task, such as: "000493.sql", and inserted into the vehicle table (vehicle) according to the set vehicle sequence and the passing time interval, and then the test samples are arranged according to the vehicle table; the time interval for inserting vehicle table data can be modified according to the specific actual test requirements, such as: "Each data is inserted at an interval of 1 second".

[0065] According to the passing station information and serial number information, the train passing simulation module copies the corresponding test samples from the performance data test set source data path to the FTP server path according to the passing task, so that the fault image intelligent recognition system can download images and identify faults. The image copying interval for each car can be modified according to the specific actual test requirements, such as "the image data copying interval for each car is 1 second".

[0066] Modification of the reporting status of the train information table: After the train passes, the train reporting status is changed from unreported to reported according to the passing station information, passing time, and serial number information, proving that the passing identification of this test train has been completed and is waiting for the next task cycle.

[0067] Distributed test execution:

[0068] The train transit simulation module executes test tasks in a distributed manner and uses Python language to implement a one-click train transit program. It can efficiently distribute performance test scenarios to multiple servers for execution, ensuring that the test load is evenly distributed and improving test efficiency. Each train transit simulation module corresponds to a different server.

[0069] Read the path and configuration of the train transit simulation module, and start n train transit simulation modules at the same time. When concurrency is not required, single-point start can be performed.

[0070] Each server executes the vehicle passing operation in sequence according to the station passing time, simulates the vehicle passing concurrently, and the recognition program performs image recognition. After the recognition is completed, the fault message is uploaded to the platform.

[0071] Real-time performance monitoring and display:

[0072] As an example, the performance monitoring module monitors the fault image intelligent recognition system script in real time, and collects various performance indicator data of the server and the fault image intelligent recognition system during the performance test every 30 seconds;

[0073] Server performance indicator data includes CPU usage, GPU usage, memory usage, GPU frequency, disk throughput, CPU temperature, GPU temperature, and network throughput;

[0074] The performance indicator data of the fault image intelligent recognition system include the number of software program threads and memory usage.

[0075] Real-time performance indicator data collection:

[0076] As an example, in the process of train passing fault identification by the fault image intelligent recognition system, the performance monitoring module is used to collect various performance indicator data in real time through shell scripts and python scripts;

[0077] Use the iostat command to obtain CPU usage, disk throughput, task queue number, IOPS and other performance indicators; use the free command to obtain memory usage and cache usage; use the nvidia-smi command to obtain GPU usage, GPU temperature and GPU frequency; use the ps command to obtain the memory usage and program thread number of the fault image intelligent recognition system and framework program; use the sar command to obtain network throughput;

[0078] Real-time performance monitoring system script deployment:

[0079] Deploy the script of the performance monitoring module to the selected server, set the performance monitoring module to start automatically at boot time, obtain performance indicator data at the set sampling time interval, for example, once every 30 seconds, and store the performance indicator data in the MySQL database. The interval for obtaining performance indicator data can be adjusted according to the actual situation.

[0080] Real-time performance indicator data alarm:

[0081] In the performance monitoring module of this embodiment, an alarm threshold is set for each performance indicator data. When the collected performance indicator data exceeds the alarm threshold, an alarm signal is generated and displayed on the visualization platform; for example, when the CPU usage rate exceeds 75%, the system can automatically alarm;

[0082] The performance monitoring module checks the log script of the fault image intelligent recognition system at a set time step, for example, every 30 seconds. When an error or ret≠0 appears in the log, the current log information is sent to the visualization platform for display as judgment data for the fault image intelligent recognition system problem process status, so as to timely discover the recognition process problems and ensure the system stability.

[0083] Real-time monitoring of message queue status: The performance monitoring module monitors the rabbitmq queue message script in real time, uses rabbitmqctl to view the server rabbitmq status and consumption data in real time, and displays it on the visualization platform to ensure the normal operation of the system.

[0084] Going a step further, real-time performance indicator data visualization:

[0085] Build a performance data visualization environment, configure the server host group and IP to be tested, and the performance indicator data display style. For example, the performance indicator trend chart is a line chart with the time axis as the horizontal axis and the performance indicator data item as the vertical axis.

[0086] Analyze the hardware resource usage trend and server performance of different manufacturers based on the performance indicator data corresponding to the server in the performance indicator trend chart;

[0087] Combined with the number of trains passing simultaneously corresponding to the passing time, and based on the corresponding performance indicator data in the performance indicator trend chart corresponding to the operation of the fault image intelligent recognition system, a multi-dimensional performance evaluation of the fault image intelligent recognition system is carried out.

[0088] As an example, the CPU usage display style can be a line chart with the time axis as the horizontal axis and the CPU usage as the vertical axis. Real-time display of performance indicator data obtained by the real-time performance monitoring system script, real-time analysis of performance data, can timely discover potential performance bottlenecks and problems.

[0089] Finally, the performance test results are analyzed:

[0090] The performance analysis of the image intelligent recognition system based on the performance indicator trend chart includes:

[0091] 1) Analyze the performance indicator data collected during the test, such as evaluating the performance of the fault image intelligent recognition system based on memory usage, GPU utilization, disk throughput, and CPU utilization; identify the performance bottlenecks of the software system through performance indicator data analysis, such as excessive CPU utilization (over 80%), insufficient memory, etc.

[0092] 2) Select and set a long-term performance indicator trend chart to evaluate the long-term operation stability of the fault image intelligent recognition system; for example, analyze the performance stability of the software system under long-term operation (7*24 hours) to ensure that the software can run stably under various loads and environments.

[0093] 3) Select performance indicator trend charts under different test environments and conditions to evaluate the repeatability of the fault image intelligent recognition system test results and ensure the consistency of the test results under different test environments and conditions.

[0094] This implementation can realize server performance testing, including hardware resource testing, network performance testing, concurrent connection testing, etc. At the same time, it can realize performance testing automation, reduce manual intervention, and improve testing efficiency.

[0095] 1) Develop corresponding performance indicator data acquisition scripts for different hardware resources such as CPU, GPU, memory, network card, etc., obtain corresponding performance indicator data, and display the data on Grafana to view the usage trend of each hardware resource.

[0096] 2) Combination Figure 3 As shown, you can design multiple server performance test indicator comparison pages, compare the performance indicator data trends of servers from different manufacturers, and clarify the performance of servers from each manufacturer.

[0097] Software performance testing module:

[0098] It can realize multi-dimensional performance evaluation of the railway train fault image intelligent recognition system to ensure that the recognition system can operate stably under various loads and environments. For example, it can identify the differences in performance indicators such as the memory usage trend of the recognition process program, the GPU usage trend, the GPU temperature trend, and the number of recognition process threads when two and four trains pass at the same time.

[0099] Performance test report output:

[0100] Write a performance test report based on the analysis results. It mainly includes test process records, test analysis results, and system adjustment suggestions, etc., evaluate the quality and completeness of the test report, and ensure that the test report can accurately reflect the test results and performance.

[0101] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that the features described in conjunction with a single embodiment may be used in other described embodiments.

Claims

1. A performance test method for a railway train fault image intelligent recognition system. It is characterized in that Collect different railway train passing images of each passing station as test samples to construct a performance test data set; the test samples are configured with a train information table, including the name of the passing station, the passing time, the corresponding train basic information and the passing status; The train passing simulation module is used to generate a passing task according to the set passing time interval, the number of passing trains and the basic information of the target train. The corresponding test samples in the performance test data set are read according to the passing task, and the test samples are transmitted to the fault image intelligent recognition system. During the process of fault identification by the fault image intelligent identification system, the performance monitoring module is used to monitor various performance indicator data of the server and the fault image intelligent identification system; The visualization platform is used to convert the received performance indicator data into a performance indicator trend chart in real time and display it. The performance indicator trend chart is used for performance analysis of the image intelligent recognition system.

2. The performance testing method of the railway train fault image intelligent recognition system according to claim 1, It is characterized in that The test sample types include normal images, missing images, oversized images, repeated images, overbright images, overdark images, storage failure images, missed images, timed-out images, and program abnormal images.

3. The performance testing method of the railway train fault image intelligent recognition system according to claim 2, It is characterized in that The train passing task includes setting the performance test data set connection information, train passing time interval, test rounds, and train basic information reading path; the train basic information is stored in the vehicle information table sql file.

4. The performance testing method of the railway train fault image intelligent recognition system according to claim 3, It is characterized in that The train passing simulation module modifies the passing time of the current target train in the train information table to the current system time according to the minimum passing time data of the target train, modifies the default receiving state of the passing state to the receiving completed state, and after the train passing is completed, modifies the reporting state of the target train in the train information table from the unreported state to the reported state.

5. The performance testing method of the railway train fault image intelligent recognition system according to claim 4, It is characterized in that In the vehicle passing task, by querying the vehicle information table sql file, insert test samples according to the set vehicle sequence and the time interval between vehicles passing; The train passing simulation module copies the corresponding test samples to the FTP server path according to the train passing task for use by the fault image intelligent recognition system; The train station passing simulation module executes the test task in a distributed manner and is implemented in Python language; each train station passing simulation module corresponds to a different server.

6. The performance testing method of the railway train fault image intelligent recognition system according to claim 5, It is characterized in that The performance monitoring module monitors the fault image intelligent recognition system script in real time and collects various performance indicator data of the server and the fault image intelligent recognition system every 30 seconds; Server performance indicator data includes CPU usage, GPU usage, memory usage, GPU frequency, disk throughput, CPU temperature, GPU temperature, and network throughput; The performance indicator data of the fault image intelligent recognition system include the number of software program threads and memory usage.

7. The performance testing method of the railway train fault image intelligent recognition system according to claim 6, It is characterized in that In the process of fault identification by the fault image intelligent identification system, the performance monitoring module is used to collect various performance indicator data in real time through shell scripts and python scripts; Use the iostat command to obtain CPU usage, disk throughput, and task queue number; use the free command to obtain memory usage and cache usage; use the nvidia-smi command to obtain GPU usage, GPU temperature, and GPU frequency; use the ps command to obtain the memory usage and program thread number of the fault image intelligent recognition system; use the sar command to obtain network throughput; Deploy the script of the performance monitoring module to the selected server, set the performance monitoring module to start automatically at boot, obtain performance indicator data at the set sampling time interval, and store the performance indicator data in the MySQL database.

8. The performance testing method of the railway train fault image intelligent recognition system according to claim 7, It is characterized in that In the performance monitoring module, alarm thresholds are set for each performance indicator data. When the collected performance indicator data exceeds the alarm threshold, an alarm signal is generated and displayed on the visualization platform; The performance monitoring module checks the log script program of the fault image intelligent recognition system according to the set time step. When an error or ret≠0 appears in the log, the current log information is sent to the visualization platform for display as the judgment data of the problem process status of the fault image intelligent recognition system; The performance monitoring module monitors the rabbitmq queue message script in real time, uses rabbitmqctl to view the server rabbitmq status and consumption data in real time, and displays it on the visualization platform.

9. The performance testing method of the railway train fault image intelligent recognition system according to claim 8, It is characterized in that The performance index trend chart is a line chart with the time axis as the horizontal axis and the performance index data item as the vertical axis; Analyze the hardware resource usage trend and server performance of different manufacturers based on the performance indicator data corresponding to the server in the performance indicator trend chart; Combined with the number of trains passing simultaneously corresponding to the passing time, and based on the corresponding performance indicator data in the performance indicator trend chart corresponding to the operation of the fault image intelligent recognition system, a multi-dimensional performance evaluation of the fault image intelligent recognition system is carried out.

10. The performance testing method of the railway train fault image intelligent recognition system according to claim 9, It is characterized in that The performance analysis of the image intelligent recognition system based on the performance indicator trend chart includes: Evaluate the performance and performance bottlenecks of the fault image intelligent recognition system based on memory usage, GPU usage, disk throughput, and CPU usage; Select and set a long-term performance indicator trend chart to evaluate the long-term operation stability of the fault image intelligent recognition system; Select performance indicator trend graphs under different test environments and conditions to evaluate the repeatable performance of the fault image intelligent recognition system.

Citation Information

Patent Citations

  • Method and device for optimizing performance of GPU server

    CN110032449A

  • Method and device for testing train intelligent detection system based on image recognition

    CN114253842A

  • Test system based on railway train fault image recognition

    CN117539668A

  • Train image acquisition device test system and method based on train operation state signal simulation

    CN117969033A

  • System and method for measuring and monitoring performance in a computer network

    US20050027858A1