A computer motherboard automated integrated testing and intelligent diagnosis system and method

Abstract

The present invention discloses a computer motherboard automated integrated test and intelligent diagnosis system and method, which relates to the field of computer motherboard test diagnosis technology, including a test platform building module, multi-interface adaptation and automatic motherboard identification, including a high-precision signal source and acquisition card, a modular architecture and high-speed bus interconnection; a test project execution module, which carries out multiple functional tests and simulates multiple scenarios to adjust the load; a data acquisition and storage module, which synchronously acquires multiple types of data through multiple channels, and has distributed storage and redundant backup; an intelligent diagnosis and analysis module, big data and machine learning algorithm diagnosis, a report generation and output module, a template engine generates reports and optimizes layout; a system self-check and calibration module, which starts each self-check module. The present invention realizes efficient and accurate testing of computer motherboards, reliable data acquisition and storage, modular design facilitates upgrading, and convenient remote functions, improves test and diagnosis efficiency and quality, helps motherboard production, maintenance and research and development, and enhances industry competitiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of computer motherboard testing and intelligent diagnosis, and in particular to a computer motherboard automated integrated testing and intelligent diagnosis system and method. Background Art

[0002] With the rapid development of computer technology, the functions of computer motherboards are becoming increasingly complex, and their performance and quality play a vital role in the stability and reliability of the entire computer system. In the production, maintenance and research and development of computer motherboards, comprehensive, efficient and accurate testing and diagnosis have become key links.

[0003] Traditional computer motherboard testing methods mainly rely on manual operation and simple testing tools. Manual testing is not only inefficient and consumes a lot of manpower and time, but is also easily affected by the tester's technical level, experience, and subjective factors, making it difficult to ensure the accuracy and consistency of the test results. For example, when conducting complex functional testing and troubleshooting, manual testing may not be able to accurately simulate various actual usage scenarios and boundary conditions, thus missing some potential problems.

[0004] Simple testing tools can often only test one or a few functions of the motherboard, and lack the ability to comprehensively evaluate the overall performance of the motherboard. Traditional testing tools are difficult to quickly adapt to the various new interfaces and technologies brought about by the continuous emergence of new models of motherboards, and test equipment needs to be frequently replaced and upgraded, which further increases the testing cost and cycle.

[0005] In terms of fault diagnosis, traditional methods mainly rely on the experience and judgment of maintenance personnel. For some complex faults, it often takes a lot of time to check and analyze them one by one, and the diagnostic efficiency is extremely low. Moreover, due to the lack of effective data recording and analysis methods, it is difficult to count and summarize the fault types and occurrence frequencies, and it is impossible to provide strong data support for product quality improvement and research and development.

[0006] With the continuous development of automation technology and artificial intelligence technology, there is an urgent need for a system and method that can realize automated integrated testing and intelligent diagnosis of computer motherboards. The system can quickly and accurately perform comprehensive tests on computer motherboards of different models, simulate various real scenarios, automatically collect and analyze test data, and use intelligent algorithms to perform fault diagnosis and prediction, thereby improving the efficiency, accuracy and reliability of testing and diagnosis, reducing costs, and providing strong technical support for the development of the computer motherboard industry to meet the growing market demand and technological innovation requirements. Summary of the invention

[0007] The present invention provides a computer motherboard automated integrated testing and intelligent diagnosis system and method to solve the problems mentioned in the above-mentioned prior art.

[0008] In order to achieve the above object, the present invention adopts the following technical solution: a computer motherboard automated integrated testing and intelligent diagnosis system, comprising:

[0009] Test platform building module: build a physical test platform, equipped with a variety of interface adapters, and have the function of automatically identifying the motherboard model, and automatically load the corresponding test program according to the identification result. It establishes a motherboard model database and adopts a recognition algorithm based on feature matching. The motherboard model feature vector is set as M = (m1, m2, ..., m n ), m1, m2···m n is the eigenvalue of the motherboard of this model in the 1st-nth characteristic dimension. The test platform collects the characteristic vector N=(n1, n2, ..., n n ), n1, n2···n n is the feature value of the motherboard to be tested in the 1st-nth feature dimension, and the similarity is calculated Among them, S represents the calculated similarity value, which is m i The characteristic value of this model motherboard in the i-th characteristic dimension, n i It is the characteristic value of the motherboard to be tested in the i-th characteristic dimension. When S≥0.98, it is determined to be a successful match. The platform includes a high-precision signal generator and a high-speed data acquisition card to collect the feedback signal of the motherboard during the test process;

[0010] Test project execution module: executes multiple functional tests, including CPU performance test, memory read and write test, graphics card output test, network connection test, USB interface transmission test. When performing memory read and write test, the read and write speed verification formula V=D / T is used, where V is the read and write speed, D is the read and write data volume, and T is the read and write time. The memory performance is judged by comparing with the standard read and write speed. When performing CPU performance test, the stress test tool is used to make the CPU core load run continuously between 80% and 100%. By monitoring the fluctuation range of CPU clock frequency, the fluctuation threshold is set to △f. If the fluctuation range △f is exceeded, it is determined that there is a problem with CPU stability. At the same time, key indicators such as instruction execution speed and cache hit rate under different loads are recorded;

[0011] Data acquisition and storage module: collects various data during the test process, including voltage, current, temperature data of each component of the mainboard, timing of signal transmission, and bit error rate data. It adopts multi-channel synchronous acquisition technology, and the number of acquisition channels is K. The data acquisition clock deviation of each channel does not exceed 1 / (C×100)ms. It uses the data cache mechanism. When the acquisition data rate exceeds the storage write rate, the data is temporarily stored in the cache area. The cache area manages data using the first-in-first-out principle.

[0012] Intelligent diagnosis and analysis module: Based on big data analysis and machine learning algorithms, the stored data is intelligently diagnosed and analyzed. The number of algorithm model training samples is no less than 100,000 groups, including motherboard test data under various normal and fault conditions. The neural network algorithm is used, and the neuron connection weight is set to ΔW. ij , adjust the weights through the back-propagation algorithm, the formula is Where η is the learning rate and E is the error function. When analyzing the cause of the fault, a fault tree model is constructed to gradually troubleshoot possible fault points from the hardware architecture level of the motherboard, including from the chipset, circuit module to specific electronic components.

[0013] Report generation and output module: Generates detailed test reports based on test results and diagnostic analysis. The report content includes basic information of the motherboard, test items and results, fault diagnosis information, and maintenance suggestions. The report is generated using template engine technology. The template variable is set to X. The variable is filled according to the test data to generate a personalized report. The report is optimized for typesetting using the document typesetting library.

[0014] System self-check and calibration module: When the system starts, it automatically performs self-check to check the status of each hardware module. By sending a specific test signal and detecting the feedback, the self-check signal is set to S t , the feedback signal is F t , when ∣S t -F t When ∣≤ε, the module is judged to be normal, ε is the allowable error, the self-check coverage rate reaches 100%, and the cyclic redundancy check CRC algorithm is used to detect the data transmission line, generate a check code and compare it with the standard check code;

[0015] The test platform is calibrated regularly. The calibration cycle is set according to the frequency of use. A high-precision calibration source is used to calibrate the signal generator and data acquisition card. The signal generator output signal calibration adjustment formula is A'=A×(Ca / Cm), where A' is the amplitude after calibration, A is the amplitude before calibration, Ca is the calibration source standard value, and Cm is the measured value. During the calibration process, the calibration data is measured multiple times to obtain the average value;

[0016] Remote monitoring and interaction module: supports remote monitoring function. Users can remotely connect to the test system through the network to view the test progress, data and results in real time. Streaming media technology is used to transmit test images. Video encoding compression technology, including H.264 encoding standard, is used to reduce the amount of video data. Identity authentication and authorization mechanism is adopted. Users need to enter user name and password. The password is stored in hash encryption. After verification, the corresponding operation authority is granted according to the user's authority.

[0017] Furthermore, the interface adapter device in the test platform building module adopts a modular design, which can be quickly replaced and expanded according to the mainboard interface type. The interface adapter device adopts a plug-in connection structure, and the connection interface is provided with a fool-proof design. Its internal circuit uses a programmable logic device PLD to achieve flexible conversion of interface protocols, and the new mainboard interface standard can be quickly adapted by programming the PLD.

[0018] Furthermore, when the test project execution module performs CPU performance testing, it uses multi-threaded concurrent testing technology to simulate 10-100 different computing tasks at the same time, fully tapping the multi-core performance of the CPU, and using thread pool technology to manage concurrent threads. The initial size of the thread pool is N0, and the maximum number of threads N is dynamically adjusted according to the number of CPU cores. max , the calculation formula is N max =k×C, where k is the adjustment coefficient and C is the number of CPU cores. During the test, the power consumption changes of the CPU are also monitored, and the power consumption curve is drawn using a power sensor to analyze the power consumption characteristics of the CPU under different loads.

[0019] Furthermore, when collecting the mainboard temperature data, the data acquisition and storage module adopts a combination of a non-contact infrared temperature sensor and a contact thermocouple sensor. The installation position of the non-contact infrared temperature sensor is optimized through optical simulation; the contact thermocouple sensor uses thermal conductive glue to tightly fit the thermocouple to the surface of the heating element to reduce thermal resistance. At the same time, the collected temperature data is filtered in real time, using a sliding average filtering algorithm with a filtering window size of W to remove noise interference in the temperature data.

[0020] Furthermore, the machine learning algorithm in the intelligent diagnosis and analysis module adopts an architecture that combines a deep neural network with a decision tree algorithm. The deep neural network is constructed using multiple convolutional layers and fully connected layers. The convolution kernel size and step size of the convolution layer use a 3x3 convolution kernel with a step size of 1. When constructing a decision tree, the decision tree algorithm uses the information gain ratio as the feature selection criterion and selects the split attribute. During the model training process, the cross-validation technology is used to divide the training data into K parts, where K is 5 or 10. K-1 parts are used in turn to train the model, and the remaining part is used to verify the model performance.

[0021] Furthermore, when generating a test report, the report generation and output module customizes the report content and format according to user needs. The user operates through a graphical interface, and the interface chooses to display detailed data charts of specific test items and hide certain irrelevant information. During the report generation process, the data is compressed using the ZIP compression algorithm. At the same time, a hyperlink is added to the report.

[0022] Furthermore, a method for applying the computer motherboard automated integrated testing and intelligent diagnosis system comprises the following steps:

[0023] Platform preparation steps: Build the test platform, connect the computer motherboard to be tested to the test platform, start the platform and complete the motherboard model identification and test program loading. The loading time should not exceed 10 seconds. If the test platform cannot automatically identify the motherboard model, the user is allowed to manually enter the motherboard model information. When manually entering information, the system displays a partial matching list of motherboard models. After the user selects, the system performs accurate matching verification again.

[0024] Test execution steps: Execute each test in the preset test item sequence, collect data in real time during the test, and for each test item, if the test result is within the standard indicator range, mark it as normal; if it is out of the range, mark it as abnormal and record detailed abnormal information. During the test execution, if an abnormal situation occurs, immediately suspend the test and save the current test status;

[0025] Diagnosis and analysis steps: The collected data is transmitted to the intelligent diagnosis and analysis module, and the machine learning algorithm is used to perform fault diagnosis and cause analysis to obtain diagnostic results and maintenance suggestions. During the diagnosis and analysis process, if a difficult fault is encountered, the system will automatically search the similar case library. The cases in the library are classified and stored according to the fault type and characteristics, and the most similar case is found through the similarity matching algorithm;

[0026] Report generation steps: Generate a test report based on the test results and diagnostic analysis, and select the output method to output the report according to user needs. The system will automatically check the integrity of the report.

[0027] Furthermore, in the platform preparation step, if the test platform cannot automatically identify the motherboard model, the user is allowed to manually enter the motherboard model information. When manually entering, the system uses a fuzzy matching algorithm. Let the motherboard model string entered by the user be U, and the motherboard model string in the database be D. Calculate the edit distance L. When L≤L max When the motherboard model is listed, the user can choose. max The maximum edit distance threshold is set. After selection, the exact match verification is performed again.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The system can automatically identify a variety of motherboard models and load the corresponding test program. The test items are comprehensive, with a coverage rate of over 95%. It can simulate a variety of scenarios and accurately detect the performance of the motherboard. Compared with traditional manual testing, the test efficiency is increased several times, and the accuracy of the test results is increased to within ±5%, which greatly reduces misjudgment and missed judgment, and effectively guarantees the quality of the motherboard.

[0030] Based on big data and machine learning algorithms, the intelligent diagnosis and analysis module can accurately identify the type of fault with a diagnostic accuracy of more than 90%. It can also accurately locate the fault to the millimeter level and give effective maintenance suggestions, effectively solving the inefficiency and uncertainty problems of traditional empirical diagnosis, reducing maintenance difficulty and cost, and shortening maintenance time.

[0031] The data collection and storage module can accurately collect multiple types of data with an accuracy rate of over 99%. The distributed storage architecture ensures reliable data storage and convenient query and analysis. These data provide a rich basis for product quality improvement, R&D innovation, and help companies optimize motherboard design, improve production processes, and enhance market competitiveness.

[0032] The modular system is easy to upgrade and expand, and the interface is flexible. The report generation and output module can generate reports in multiple formats and output them in multiple ways as needed. The remote monitoring and interaction module can realize remote operation and monitoring, which is convenient for users to use in different scenarios, improves work coordination and convenience, and adapts to modern production and management needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The present invention provides a schematic block diagram of a computer motherboard automated integrated testing and intelligent diagnosis system. DETAILED DESCRIPTION

[0034] 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.

[0035] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0036] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, and it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0037] Reference Figure 1 :A computer motherboard automated integrated testing and intelligent diagnosis system, comprising:

[0038] Test platform building module: build a physical test platform, equipped with a variety of interface adapters, which can be physically connected to more than 95% of different computer motherboards on the market. The interface connection stability is higher than 99%, and it has the function of automatically identifying motherboard models, with an identification accuracy of more than 98%. It automatically loads the corresponding test program according to the identification results. It establishes a motherboard model database and adopts a recognition algorithm based on feature matching. The motherboard model feature vector is set as M = (m1, m2, ..., m n ), m1, m2···m n is the eigenvalue of the motherboard of this model in the 1st-nth characteristic dimension. The test platform collects the characteristic vector N=(n1, n2, ..., n n ), n1, n2···n n is the feature value of the motherboard to be tested in the 1st-nth feature dimension, and the similarity is calculated Among them, S represents the calculated similarity value, which is m i The characteristic value of this model motherboard in the i-th characteristic dimension, ni It is the characteristic value of the motherboard to be tested in the i-th characteristic dimension. When S≥0.98, it is determined to be a successful match. The platform includes a high-precision signal generator that can generate various test signals with a frequency range of 1Hz-10GHz and an amplitude accuracy of ±0.1%, as well as a high-speed data acquisition card with a sampling frequency of up to 100MS / s and a sampling accuracy of 16 bits, which is used to collect the feedback signal of the motherboard during the test process.

[0039] The platform adopts a modular architecture design. Each functional module can be upgraded and replaced independently. The modules are connected by a high-speed bus with a data transmission rate of no less than 10GB / s, which reduces system maintenance costs and upgrade difficulty and improves system scalability.

[0040] Test project execution module: execute multiple functional tests, including but not limited to CPU performance test, memory read and write test, graphics card output test, network connection test, USB interface transmission test, etc. Each test has a preset standard indicator range, and the test coverage rate is more than 95%. When performing memory read and write tests, the read and write speed verification formula V=D / T is used, where V is the read and write speed, D is the amount of read and write data, and T is the read and write time. Compare with the standard read and write speed to judge the memory performance. When performing CPU performance testing, use the stress test tool to make the CPU core load run continuously between 80%-100% for a period of time. By monitoring the fluctuation range of the CPU clock frequency, the fluctuation threshold is set to △f. If the fluctuation range △f is exceeded, it is determined that there is a problem with the CPU stability. At the same time, key indicators such as instruction execution speed and cache hit rate under different loads are recorded to comprehensively evaluate the CPU performance.

[0041] During the test, a variety of actual usage scenarios can be simulated, such as high-load operation, multi-tasking, long-term continuous operation, etc. The authenticity of the simulated scenarios is not less than 90%, so as to comprehensively test the performance of the motherboard under different working conditions. Different loads are simulated by dynamically adjusting the system resource allocation, and the load adjustment coefficient is set to K, which is between 0.1-1 according to the requirements of different scenarios.

[0042] Data acquisition and storage module: collects various data during the test process, such as voltage, current, temperature data of each component of the mainboard, timing of signal transmission, bit error rate data, etc. The time interval for data acquisition can be flexibly set between 1ms-100ms, and the accuracy of data acquisition is not less than 99%. Adopting multi-channel synchronous acquisition technology, the number of acquisition channels is set to K, and the data acquisition clock deviation of each channel does not exceed 1 / (C×100)ms to ensure data synchronization. Using the data cache mechanism, the cache capacity is B. When the acquisition data rate exceeds the storage write rate, the data is first temporarily stored in the cache area. The cache area uses the first-in-first-out principle to manage data to prevent data loss.

[0043] The collected data is stored in a large-capacity database. The database adopts a distributed storage architecture and has data redundancy backup function. The data storage reliability is not less than 99.9%. It can store data records of at least 1,000 complete tests to facilitate subsequent query and analysis.

[0044] Intelligent diagnosis and analysis module: Based on big data analysis and machine learning algorithms, the stored data is intelligently diagnosed and analyzed. The number of algorithm model training samples is no less than 100,000 groups, covering various motherboard test data under normal and fault conditions, and can automatically identify the fault type of the motherboard. The fault diagnosis accuracy is no less than 90%. Using the neural network algorithm, the neuron connection weight is ΔW ij , adjust the weights through the back-propagation algorithm, the formula is Where η is the learning rate and E is the error function, which improves the diagnostic accuracy. When analyzing the cause of the fault, a fault tree model is constructed to gradually check possible fault points from the hardware architecture level of the motherboard, such as from the chipset, circuit module to specific electronic components, and the root cause of the fault is determined in combination with data characteristics to improve the accuracy of fault location.

[0045] For the diagnosed faults, the cause of the fault can be further analyzed and detailed fault location information can be provided with positioning accuracy at the millimeter level. At the same time, maintenance suggestions can be given, and the effectiveness of the maintenance suggestions is not less than 80%.

[0046] Report generation and output module: Generate detailed test reports based on test results and diagnostic analysis. The report content includes basic information of the motherboard, test items and results, fault diagnosis information, maintenance suggestions, etc. The report format is standardized and can be generated in multiple formats such as PDF and HTML. The generation time does not exceed 1 minute. Use template engine technology to generate reports, set the template variable to X, and fill in the variables according to the test data to generate personalized reports. Use the document typesetting library to optimize the report layout to ensure clear charts and neat text, and improve report readability.

[0047] The test report can be output through network transmission, local printing or storage in a mobile storage device, with at least 3 output methods, which is convenient for users to obtain and use. The encrypted transmission protocol is used to ensure the security of network transmission. The encryption algorithm uses the AES encryption algorithm with a key length of 256 bits to prevent the leakage of report data.

[0048] System self-test and calibration module: When the system starts, it automatically performs self-test to check the status of each hardware module, including signal generator, data acquisition card, etc. By sending a specific test signal and detecting the feedback, the self-test signal is set to S t , the feedback signal is F t , when ∣S t -F tWhen ∣≤ε, the module is judged to be normal, ε is the allowable error, and the self-check coverage rate reaches 100%. The cyclic redundancy check (CRC) algorithm is used to detect the data transmission line, generate a check code and compare it with the standard check code to ensure the accuracy of data transmission.

[0049] The test platform is calibrated regularly. The calibration cycle can be set according to the frequency of use. A high-precision calibration source is used to calibrate the signal generator and data acquisition card. The signal generator output signal calibration adjustment formula is A'=A×(Ca / Cm), where A' is the amplitude after calibration, A is the amplitude before calibration, Ca is the calibration source standard value, and Cm is the measured value to ensure the test accuracy. During the calibration process, the calibration data is measured multiple times to take the average value to reduce the measurement error. The number of measurements is not less than 5 times.

[0050] Remote monitoring and interaction module: Supports remote monitoring function. Users can remotely connect to the test system through the network to view the test progress, data and results in real time. Streaming media technology is used to transmit test images, with a frame rate of no less than 25fps and a delay of no more than 1s. Video encoding compression technology, such as the H.264 encoding standard, is used to reduce the amount of video data and improve transmission efficiency.

[0051] Realize remote interaction, users can remotely set test parameters, start or pause tests, and the response time of remote operation instructions does not exceed 0.5s, which is convenient for remote collaboration and management. Adopt identity authentication and authorization mechanism, users need to enter user name and password, and password is stored in hash encryption. After verification, corresponding operation permissions are granted according to user permissions to ensure system security.

[0052] In the present invention, the interface adapter device in the test platform building module adopts a modular design and can be quickly replaced and expanded according to the mainboard interface type. The interface adapter device adopts a plug-in connection structure, and a fool-proof design is provided at the connection interface to prevent mis-insertion. Its internal circuit adopts a programmable logic device (PLD) to achieve flexible conversion of the interface protocol. By programming the PLD, it can quickly adapt to the new mainboard interface standard, and the replacement time does not exceed 5 minutes. The interface adaptation capacity after expansion is increased by no less than 30%.

[0053] In the present invention, the test project execution module adopts multi-threaded concurrent testing technology when performing CPU performance testing, which can simulate 10-100 different computing tasks at the same time and fully exploit the multi-core performance of the CPU. The concurrent threads are managed by thread pool technology. The initial size of the thread pool is N0, and the maximum number of threads N is dynamically adjusted according to the number of CPU cores. max , the calculation formula is N max= k × C, where k is the adjustment coefficient and C is the number of CPU cores. The accuracy of the test results is within ±5%. During the test, the power consumption changes of the CPU can also be monitored. A power sensor with an accuracy of ±0.1W is used to draw a power consumption curve and analyze the power consumption characteristics of the CPU under different loads.

[0054] In the present invention, the data acquisition and storage module uses a combination of a non-contact infrared temperature sensor and a contact thermocouple sensor when collecting the temperature data of the mainboard. The temperature resolution of the non-contact infrared temperature sensor is 0.1°C, and its installation position is optimized through optical simulation to ensure that the temperature of the key heating area of ​​the mainboard can be accurately captured. The measurement accuracy of the contact thermocouple sensor is ±0.5°C, and a special thermally conductive adhesive is used to tightly fit the thermocouple to the surface of the heating element to reduce thermal resistance, improve measurement accuracy, and ensure the comprehensiveness and accuracy of temperature data collection. At the same time, the collected temperature data is filtered in real time, using a sliding average filtering algorithm with a filtering window size of W to remove noise interference in the temperature data and more realistically reflect the temperature changes of the mainboard.

[0055] In the present invention, the machine learning algorithm in the intelligent diagnosis and analysis module adopts an architecture that combines a deep neural network with a decision tree algorithm. The deep neural network is constructed using multiple convolutional layers and fully connected layers. The convolution kernel size and step size of the convolution layer are optimized and selected, for example, a 3x3 convolution kernel with a step size of 1 is used to better extract data features. When constructing a decision tree, the decision tree algorithm uses the information gain ratio as the feature selection criterion, selects split attributes, reduces the depth and complexity of the decision tree, and improves the model training convergence speed by not less than 20%. In the model training process, a cross-validation technique is used to divide the training data into K parts (K is generally 5 or 10), and K-1 parts are used in turn to train the model, and the remaining part is used to verify the model performance, thereby improving the generalization ability of the model.

[0056] In the present invention, when generating a test report, the report generation and output module can customize the report content and format according to user needs. The user can operate through a graphical interface, which provides a variety of options, such as selecting to display detailed data charts of specific test items, hiding certain irrelevant information, etc., and the degree of customization is not less than 80%. During the report generation process, the data is compressed and the ZIP compression algorithm is used to reduce the size of the report file for easy storage and transmission. At the same time, a hyperlink is added to the report to facilitate users to quickly jump to the detailed page of related test data or diagnostic results, thereby improving the interactivity of the report.

[0057] The present invention also discloses a computer motherboard automatic integrated testing and intelligent diagnosis method, comprising the following steps:

[0058] Platform preparation steps: Build the test platform, connect the computer motherboard to be tested to the test platform, start the platform and complete the motherboard model identification and test program loading, the loading time shall not exceed 10 seconds. If the test platform cannot automatically identify the motherboard model, the user is allowed to manually enter the motherboard model information, and the manually entered information verification accuracy rate reaches 100%, ensuring that the test program is correctly loaded. When manually entering information, the system will display a partial matching list of the motherboard model, and the system will perform accurate matching verification again after the user selects.

[0059] Test execution steps: Execute each test in the preset test item sequence, collect data in real time during the test, and for each test item, if the test result is within the standard index range, it is marked as normal, if it is out of range, it is marked as abnormal and detailed abnormal information is recorded. The test execution time is between 10 minutes and 60 minutes depending on the complexity of the motherboard. During the test execution, if an abnormal situation occurs, immediately pause the test and save the current test status, including test data, program running parameters, etc., for subsequent fault analysis.

[0060] Diagnostic analysis steps: The collected data is transmitted to the intelligent diagnostic analysis module, and the machine learning algorithm is used to perform fault diagnosis and cause analysis to obtain diagnostic results and maintenance recommendations. The diagnostic analysis time does not exceed 5 minutes. During the diagnostic analysis process, if a difficult fault is encountered, the system will automatically search the similar case library. The cases in the library are classified and stored according to the fault type and characteristics. The most similar case is found through the similarity matching algorithm, and its diagnostic results and maintenance recommendations are referred to further improve the accuracy of the diagnosis.

[0061] Report generation steps: Generate a test report based on the test results and diagnostic analysis, and select the output method to output the report according to user needs. The report generation and output time does not exceed 1 minute. After the report is generated, the system will automatically check the integrity of the report, including whether the report format is standardized, whether the data is complete, whether it contains necessary diagnostic information and maintenance suggestions, etc., to ensure the quality of the report.

[0062] In the present invention, in the platform preparation step, if the test platform cannot automatically identify the motherboard model, the user is allowed to manually input the motherboard model information. When manually inputting, the system uses a fuzzy matching algorithm, assuming that the motherboard model string input by the user is, the motherboard model string in the database is U, and the edit distance D is calculated. When L≤L max When the motherboard model is listed, the user can choose. max The maximum edit distance threshold is set. After selection, the exact match verification is performed again, and the verification accuracy reaches 100%, ensuring that the test program is loaded correctly. During the manual input of information, the system provides an automatic completion function for the motherboard model. According to the characters entered by the user, possible motherboard models are screened from the database and displayed to improve input efficiency.

[0063] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A computer motherboard automated integrated testing and intelligent diagnosis system, characterized in that: include: Test platform building module: build a physical test platform, equipped with a variety of interface adapters, and have the function of automatically identifying the motherboard model, and automatically load the corresponding test program according to the identification result. It establishes a motherboard model database and adopts a recognition algorithm based on feature matching. The motherboard model feature vector is set as M = (m1, m2, ..., m n ), m1, m2···m n is the eigenvalue of the motherboard of this model in the 1st-nth characteristic dimension. The test platform collects the characteristic vector N=(n1, n2, ..., n n ), n1, n2···n n is the feature value of the motherboard to be tested in the 1st-nth feature dimension, and the similarity is calculated Among them, S represents the calculated similarity value, which is m i The characteristic value of this model motherboard in the i-th characteristic dimension, n i It is the characteristic value of the motherboard to be tested in the i-th characteristic dimension. When S≥0.98, it is determined to be a successful match. The platform includes a high-precision signal generator and a high-speed data acquisition card to collect the feedback signal of the motherboard during the test process; Test project execution module: executes multiple functional tests, including CPU performance test, memory read and write test, graphics card output test, network connection test, USB interface transmission test. When performing memory read and write test, the read and write speed verification formula V=D / T is used, where V is the read and write speed, D is the read and write data volume, and T is the read and write time. The memory performance is judged by comparing with the standard read and write speed. When performing CPU performance test, the stress test tool is used to make the CPU core load run continuously between 80% and 100%. By monitoring the fluctuation range of CPU clock frequency, the fluctuation threshold is set to △f. If the fluctuation range △f is exceeded, it is determined that there is a problem with CPU stability. At the same time, key indicators such as instruction execution speed and cache hit rate under different loads are recorded; Data acquisition and storage module: collects various data during the test process, including voltage, current, temperature data of each component of the mainboard, timing of signal transmission, and bit error rate data. It adopts multi-channel synchronous acquisition technology, and the number of acquisition channels is K. The data acquisition clock deviation of each channel does not exceed 1 / (C×100)ms. It uses the data cache mechanism. When the acquisition data rate exceeds the storage write rate, the data is temporarily stored in the cache area. The cache area manages data using the first-in-first-out principle. Intelligent diagnosis and analysis module: Based on big data analysis and machine learning algorithms, the stored data is intelligently diagnosed and analyzed. The number of algorithm model training samples is no less than 100,000 groups, including motherboard test data under various normal and fault conditions. The neural network algorithm is used, and the neuron connection weight is set to ΔW. ij , adjust the weights through the back-propagation algorithm, the formula is Where η is the learning rate and E is the error function. When analyzing the cause of the fault, a fault tree model is constructed to gradually troubleshoot possible fault points from the hardware architecture level of the motherboard, including from the chipset, circuit module to specific electronic components. Report generation and output module: Generates detailed test reports based on test results and diagnostic analysis. The report content includes basic information of the motherboard, test items and results, fault diagnosis information, and maintenance suggestions. The report is generated using template engine technology. The template variable is set to X. The variable is filled according to the test data to generate a personalized report. The report is optimized for typesetting using the document typesetting library. System self-check and calibration module: When the system starts, it automatically performs self-check to check the status of each hardware module. By sending a specific test signal and detecting the feedback, the self-check signal is set to S t , the feedback signal is F t , when ∣S t -F t When ∣≤ε, the module is judged to be normal, ε is the allowable error, the self-check coverage rate reaches 100%, and the cyclic redundancy check CRC algorithm is used to detect the data transmission line, generate a check code and compare it with the standard check code; The test platform is calibrated regularly. The calibration cycle is set according to the frequency of use. A high-precision calibration source is used to calibrate the signal generator and data acquisition card. The signal generator output signal calibration adjustment formula is A'=A×(Ca / Cm), where A' is the amplitude after calibration, A is the amplitude before calibration, Ca is the calibration source standard value, and Cm is the measured value. During the calibration process, the calibration data is measured multiple times to obtain the average value; Remote monitoring and interaction module: supports remote monitoring function. Users can remotely connect to the test system through the network to view the test progress, data and results in real time. Streaming media technology is used to transmit test images. Video encoding compression technology, including H.264 encoding standard, is used to reduce the amount of video data. Identity authentication and authorization mechanism is adopted. Users need to enter user name and password. The password is stored in hash encryption. After verification, the corresponding operation authority is granted according to the user's authority.

2. The computer motherboard automated integrated testing and intelligent diagnosis system according to claim 1, characterized in that: The interface adapter device in the test platform building module adopts a modular design and can be quickly replaced and expanded according to the mainboard interface type. The interface adapter device adopts a plug-in connection structure and a fool-proof design is provided at the connection interface. Its internal circuit uses a programmable logic device PLD to achieve flexible conversion of interface protocols. The new mainboard interface standard can be quickly adapted by programming the PLD.

3. The computer motherboard automated integrated testing and intelligent diagnosis system according to claim 1, characterized in that: When the test project execution module performs CPU performance testing, it uses multi-threaded concurrent testing technology to simulate 10-100 different computing tasks at the same time, fully tapping the multi-core performance of the CPU, and using thread pool technology to manage concurrent threads. The initial size of the thread pool is N0, and the maximum number of threads N is dynamically adjusted according to the number of CPU cores. max , the calculation formula is N max =k×C, where k is the adjustment coefficient and C is the number of CPU cores. During the test, the power consumption changes of the CPU are also monitored, and the power consumption curve is drawn using a power sensor to analyze the power consumption characteristics of the CPU under different loads.

4. The computer motherboard automated integrated testing and intelligent diagnosis system according to claim 1, characterized in that: When collecting the temperature data of the mainboard, the data acquisition and storage module adopts a combination of a non-contact infrared temperature sensor and a contact thermocouple sensor. The installation position of the non-contact infrared temperature sensor is optimized through optical simulation; the contact thermocouple sensor uses thermal conductive glue to fit the thermocouple tightly to the surface of the heating element to reduce thermal resistance. At the same time, the collected temperature data is filtered in real time, using a sliding average filtering algorithm with a filtering window size of W to remove noise interference in the temperature data.

5. The computer motherboard automated integrated testing and intelligent diagnosis system according to claim 1, characterized in that: The machine learning algorithm in the intelligent diagnosis and analysis module adopts an architecture that combines a deep neural network with a decision tree algorithm. The deep neural network is constructed using multiple convolutional layers and fully connected layers. The convolution kernel size and step size of the convolution layer use a 3x3 convolution kernel with a step size of 1. When constructing a decision tree, the decision tree algorithm uses the information gain ratio as the feature selection criterion and selects the split attribute. During the model training process, the cross-validation technology is used to divide the training data into K parts, where K is 5 or 10. K-1 parts are used in turn to train the model, and the remaining part is used to verify the model performance.

6. The computer motherboard automated integrated testing and intelligent diagnosis system according to claim 1, characterized in that: When generating a test report, the report generation and output module customizes the report content and format according to user needs. The user operates through a graphical interface, which chooses to display detailed data charts of specific test items and hide certain irrelevant information. During the report generation process, the data is compressed using the ZIP compression algorithm. At the same time, a hyperlink is added to the report.

7. A method for applying the computer motherboard automated integrated testing and intelligent diagnosis system according to any one of claims 1 to 6, characterized in that: The following steps are involved: Platform preparation step: construct the test platform as claimed in claim 1, connect the motherboard of the computer to be tested to the test platform, start the platform and complete the motherboard model identification and test program loading, the loading time shall not exceed 10 seconds, if the test platform cannot automatically identify the motherboard model, allow the user to manually enter the motherboard model information, when manually entering the information, the system displays a partial matching list of the motherboard model, after the user selects, the system performs accurate matching verification again; Test execution steps: Execute each test in the preset test item sequence, collect data in real time during the test, and for each test item, if the test result is within the standard indicator range, mark it as normal; if it is out of the range, mark it as abnormal and record detailed abnormal information. During the test execution, if an abnormal situation occurs, immediately suspend the test and save the current test status; Diagnosis and analysis steps: The collected data is transmitted to the intelligent diagnosis and analysis module, and the machine learning algorithm is used to perform fault diagnosis and cause analysis to obtain diagnostic results and maintenance suggestions. During the diagnosis and analysis process, if a difficult fault is encountered, the system will automatically search the similar case library. The cases in the library are classified and stored according to the fault type and characteristics, and the most similar case is found through the similarity matching algorithm; Report generation steps: Generate a test report based on the test results and diagnostic analysis, and select the output method to output the report according to user needs. The system will automatically check the integrity of the report.

8. The computer motherboard automated integrated testing and intelligent diagnosis method according to claim 7, characterized in that: In the platform preparation step, if the test platform cannot automatically identify the motherboard model, the user is allowed to manually enter the motherboard model information. When manually entering, the system uses a fuzzy matching algorithm. Let the motherboard model string entered by the user be U, and the motherboard model string in the database be D. Calculate the edit distance L. When L≤L max When the motherboard model is listed, the user can choose. max The maximum edit distance threshold is set. After selection, the exact match verification is performed again.

Images