Embedded mainboard accelerated aging test method and system based on intelligent algorithm

Abstract

The invention relates to the technical field of mainboard aging test, in particular to an embedded mainboard accelerated aging test method and system based on an intelligent algorithm. The method comprises the following steps: under a preset temperature cycle working condition, scanning a plurality of monitoring areas by using an infrared thermal imager, recording voltage and current values, and generating a real-time parameter data stream; the parameters are compared with a reference database, and an abnormal degree judgment result is obtained through time sequence feature calculation and normalization processing; inputting the result into a pre-trained LSTM deep neural network to generate fault early warning information; temperature circulation and voltage fluctuation parameters are automatically calculated and updated; and finally generating a diagnosis report to evaluate the reliability of the mainboard. Adaptive adjustment of test parameters is realized through real-time monitoring and an intelligent early warning mechanism, and the efficiency and accuracy of the accelerated aging test are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mainboard aging test, in particular to an embedded mainboard accelerated aging test method and system based on intelligent algorithm. BACKGROUND

[0002] As the core component of electronic equipment, the reliability of embedded mainboard directly affects the performance and service life of the whole machine. Accelerated aging test is an important means to evaluate and improve the reliability of mainboard. By testing under severe conditions, potential design and manufacturing defects can be exposed in a short period of time, thereby predicting the reliability performance of the product in actual use.

[0003] The current embedded mainboard accelerated aging test mainly adopts the fixed working condition cycle method, and tests the mainboard by temperature cycle, temperature shock and other environmental stress incentives. The temperature, voltage, current and other parameters of the mainboard are collected in real time during the test, and the performance of the mainboard is judged according to the preset test specification.

[0004] In the prior art, the test condition parameters are pre-set and fixed, and the test conditions cannot be dynamically adjusted according to the actual performance state of the mainboard during the test, which may cause excessive or insufficient stress, affecting the test efficiency and accuracy. This situation needs to be further improved. SUMMARY

[0005] In order to solve the problem that the existing mainboard aging test cannot dynamically adjust the test conditions according to the actual performance state of the mainboard during the test, which may cause excessive or insufficient stress, affecting the test efficiency and accuracy, the present application provides an embedded mainboard accelerated aging test method and system based on intelligent algorithm, which adopts the following technical scheme: In the first aspect, the present application provides an embedded mainboard accelerated aging test method based on intelligent algorithm, comprising the following steps: Under the preset temperature cycle condition, measure the temperature distribution data, working voltage value and current value of the embedded mainboard, and generate real-time parameter data stream; Compare the real-time parameter data stream with the preset performance benchmark database to determine the abnormality degree judgment result; Input the abnormality degree judgment result into the pre-trained machine learning model to generate fault warning information; Adaptively adjust the temperature cycle condition and voltage fluctuation amplitude according to the fault warning information; Generate a fault diagnosis report based on the adjusted temperature cycle condition and voltage fluctuation amplitude, and evaluate the reliability state of the mainboard.

[0006] By adopting the technical scheme, the conventional accelerated aging test adopts a fixed temperature cycle working condition, cannot dynamically adjust stress parameters according to actual performance degradation of the mainboard, leads to low test efficiency, and is prone to missing key fault characteristics; firstly, the infrared thermal imager is used to scan the divided multiple monitoring areas under the preset temperature cycle working condition, and working voltage values and current values are recorded to generate real-time parameter data streams; then, the parameters are compared with a performance benchmark database, time sequence characteristic values are calculated, difference operation and normalization processing are performed, and standardized abnormality degree determination results are obtained; the results are input into an LSTM deep neural network model trained by a large amount of historical data to generate fault early warning information containing an abnormal type and degree; based on the early warning information, the system automatically calculates adjustment amounts of a temperature cycle range and a voltage fluctuation amplitude, and updates test parameters in real time; finally, a diagnosis report is generated according to the adjusted parameters to evaluate reliability of the mainboard; through introduction of real-time monitoring and intelligent early warning mechanisms, test parameters can be adjusted in time according to performance changes of the mainboard, and efficiency and accuracy of the accelerated aging test are significantly improved.

[0007] Optionally, the temperature distribution data of the embedded mainboard is measured, and specifically includes the following steps: The embedded mainboard is divided into multiple temperature monitoring areas, and distribution of monitoring points in each area is obtained; An infrared thermal imager is used to periodically scan the monitoring points in each area to generate temperature field data; Temperature gradients of each temperature monitoring area are calculated according to the temperature field data; Temperature stress concentration areas are identified based on the temperature gradients, and temperature distribution characteristics are generated.

[0008] By adopting the technical scheme, the conventional temperature distribution measurement method often uses single-point or a small number of fixed points for temperature monitoring, and is difficult to comprehensively reflect temperature change characteristics of each area of the mainboard, and is prone to missing temperature stress concentration areas; the mainboard is first divided into multiple temperature monitoring areas; then, an infrared thermal imager is used to periodically scan the monitoring points to obtain complete temperature field data; by calculating temperature gradients of each monitoring area, the non-uniformity of the temperature distribution can be quantitatively characterized; finally, temperature stress concentration areas are identified based on the temperature gradients, and temperature characteristics containing spatial distribution information are generated; accurate measurement of the temperature distribution of the mainboard and accurate identification of the stress concentration areas are realized.

[0009] Optionally, the preset temperature cycle working condition includes: A preset constant temperature stress is applied to the mainboard, and initial performance parameters are collected; A temperature is cyclically switched in a temperature cycle range at a preset change rate; Changes of electrical parameters of the mainboard are recorded during the temperature cycle process; Adjust the temperature cycle duration according to the change of the electrical parameter.

[0010] By adopting the technical scheme, the traditional temperature cycle test often adopts fixed temperature range and cycle time, and cannot adjust the test working condition according to the actual performance change of the mainboard, thereby reducing the pertinence of the test. The application first applies a preset constant temperature stress to the mainboard and collects initial performance parameters as reference values. Then, temperature switching is performed in the temperature cycle range according to a preset change rate, so as to ensure the controllability of temperature change. Meanwhile, the change of electrical parameters of the mainboard is recorded during the temperature cycle process. Finally, the temperature cycle duration is dynamically adjusted according to the change of the electrical parameters, so as to realize self-adaptive adjustment of the temperature cycle working condition and improve the efficiency and reliability of the accelerated aging test.

[0011] Optionally, the real-time parameter data stream is compared with a preset performance reference database to determine an abnormality degree determination result, including the following steps: Calculate the time sequence characteristic value of the real-time parameter data stream to obtain a parameter fluctuation trend; Differentially operate the parameter fluctuation trend and a standard fluctuation range in the preset performance reference database to obtain deviation data; Normalize the deviation data to generate a standardized abnormality index; Set a weight coefficient according to the standardized abnormality index, calculate a comprehensive abnormality degree according to the weight coefficient, grade the comprehensive abnormality degree according to a preset threshold, and output the abnormality degree determination result.

[0012] By adopting the technical scheme, the application first calculates the time sequence characteristic value of the real-time parameter data stream to obtain the dynamic change trend of the parameter fluctuation. Then, the fluctuation trend is differentially operated with the standard range in the performance reference database to obtain quantitative deviation data. Then, the deviation data is normalized to make the abnormality degrees of different types of parameters comparable. Finally, the comprehensive abnormality degree is calculated by setting the weight coefficient, and is graded according to the preset threshold. The multi-dimensional evaluation and accurate quantification of the parameter abnormality are realized, and a reliable decision basis is provided for subsequent fault early warning.

[0013] Optionally, the establishment process of the pre-trained machine learning model includes the following steps: Collect parameter data streams of multiple groups of embedded mainboards in a temperature cycle test process to generate an original training data set; Calculate the abnormality degree of the parameter data stream, and perform outlier detection and data cleaning on the original training data set; Construct a labeled data set according to the abnormality degree and corresponding fault records; training an LSTM deep neural network based on the labeled dataset to establish a mapping relationship from abnormality degree to failure warning.

[0014] By adopting the technical solution, the application first collects a large amount of parameter data flow in the temperature cycle test process to form an original training dataset; then performs abnormality degree calculation and cleaning processing on the data to ensure the quality of the training data; then associates the abnormality degree with the actual failure record to build a labeled dataset; finally, an LSTM deep neural network is trained to establish an intelligent mapping relationship from parameter abnormality to failure warning; the warning model is established in a data-driven manner, improving the accuracy and reliability of failure warning.

[0015] Optionally, the temperature cycle condition and the voltage fluctuation amplitude are adaptively adjusted according to the failure warning information, specifically including the following steps: extracting the abnormal parameter type and abnormality degree determination result in the failure warning information; determining the stress type that needs to be strengthened based on the abnormal parameter type; calculating the adjustment amount of the temperature cycle range and the voltage fluctuation amplitude based on the abnormality degree determination result, wherein the adjustment amount of the temperature cycle range is the product of the abnormality degree determination result and a temperature coefficient, and the adjustment amount of the voltage fluctuation amplitude is the product of the abnormality degree determination result and a voltage coefficient, and the temperature coefficient and the voltage coefficient are obtained according to historical test data statistics; updating the temperature cycle condition and the voltage fluctuation parameter in real time according to the adjustment amount.

[0016] By adopting the technical solution, the traditional accelerated aging test uses fixed stress parameters, which cannot dynamically adjust the test intensity according to the failure warning result, resulting in low test efficiency or missing key failure characteristics; the application first extracts the type and degree information of the abnormal parameter from the failure warning information; then determines the stress type that needs to be strengthened according to the abnormal parameter type; then calculates the adjustment amount of the temperature cycle range and the voltage fluctuation amplitude by multiplying the abnormality degree with the temperature coefficient and the voltage coefficient obtained from historical data statistics; finally, the temperature cycle condition and the voltage fluctuation parameter are updated in real time; a quantitative correlation mechanism between the warning information and the test parameters is established, the accurate regulation of the test stress is realized, and the pertinence and efficiency of the accelerated aging test are improved.

[0017] Optionally, a failure diagnosis report is generated, specifically including the following steps: extracting abnormal parameter features and corresponding time series data from the failure warning information to establish a spatiotemporal correlation mapping between the abnormal parameter features and the temperature cycle parameters, wherein the abnormal parameter features include abnormal parameter types and abnormality degree determination results; Determine a fault type and a fault occurrence area based on the spatio-temporal correlation mapping, and generate a fault diagnosis report.

[0018] By adopting the technical solution, the application first extracts abnormal parameter features and time series data from the fault early warning information, establishes a spatio-temporal correlation mapping with temperature cycle parameters, then determines a fault type and a specific occurrence area based on the mapping relationship, and generates a complete diagnosis report, thereby realizing dynamic tracking and accurate positioning of a fault occurrence process, and providing technical support for product reliability improvement.

[0019] In a second aspect, the application provides an embedded mainboard accelerated aging test system based on an intelligent algorithm, comprising: A parameter acquisition module is configured to measure temperature distribution data, working voltage values and current values of the embedded mainboard under a preset temperature cycle condition, and generate real-time parameter data streams; An abnormality detection module is configured to compare the real-time parameter data streams with a preset performance benchmark database, and determine an abnormality degree determination result; A fault early warning module is configured to input the abnormality degree determination result into a pre-trained machine learning model, and generate fault early warning information; A test parameter adjustment module is configured to adaptively adjust a temperature cycle condition and a voltage fluctuation amplitude according to the fault early warning information; A mainboard reliability evaluation module is configured to generate a fault diagnosis report based on the adjusted temperature cycle condition and the voltage fluctuation amplitude, and evaluate a mainboard reliability state.

[0020] In a third aspect, the application provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned embedded mainboard accelerated aging test method based on an intelligent algorithm when executing the computer program.

[0021] In a fourth aspect, the application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above-mentioned embedded mainboard accelerated aging test method based on an intelligent algorithm when executed by a processor.

[0022] In summary, the application has at least one of the following beneficial technical effects: The application generates real-time parameter data stream by scanning multiple monitoring areas and recording voltage and current values using an infrared thermal imager under preset temperature cycle conditions; compares the parameters with the benchmark database, obtains abnormality degree determination results through time series feature calculation and normalization processing; inputs the results into a pre-trained LSTM deep neural network to generate fault warning information; automatically calculates and updates temperature cycle and voltage fluctuation parameters accordingly; finally generates a diagnostic report to evaluate the reliability of the motherboard; realizes adaptive adjustment of test parameters through real-time monitoring and intelligent warning mechanism, improving the efficiency and accuracy of accelerated aging test; Traditional temperature cycle test often uses fixed temperature range and cycle time, which cannot adjust the test conditions according to the actual performance change of the motherboard, reducing the pertinence of the test; the application first applies a preset constant temperature stress to the motherboard and collects initial performance parameters as benchmark values; then switches the temperature within the temperature cycle range according to the preset change rate to ensure the controllability of temperature change; at the same time, the electrical parameter changes of the motherboard are recorded continuously during the temperature cycle; finally, the temperature cycle duration is dynamically adjusted according to the change of electrical parameters; realizing the adaptive adjustment of temperature cycle conditions, improving the efficiency and reliability of accelerated aging test; The application first collects a large amount of parameter data stream in the temperature cycle test process to form the original training data set; then performs abnormality calculation and cleaning processing on the data to ensure the quality of the training data; then associates the abnormality degree with the actual fault record to build a labeled data set; finally, the LSTM deep neural network is trained to establish an intelligent mapping relationship from parameter abnormality to fault warning; the warning model is established through data-driven method, improving the accuracy and reliability of fault warning. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart of the embedded motherboard accelerated aging test method based on intelligent algorithm of the embodiment of the application; Figure 2 is a flowchart of step S100 in the embedded motherboard accelerated aging test method based on intelligent algorithm of the embodiment of the application; Figure 3 is a flowchart of step S200 in the embedded motherboard accelerated aging test method based on intelligent algorithm of the embodiment of the application; Figure 4 is a flowchart of the establishment of machine learning model in the embedded motherboard accelerated aging test method based on intelligent algorithm of the embodiment of the application; Figure 5 is a flowchart of step S400 in the embedded motherboard accelerated aging test method based on intelligent algorithm of the embodiment of the application; Figure 6is a flowchart of step S500 in the method for accelerating the aging test of the embedded mainboard based on the intelligent algorithm according to the embodiment of the present application. Figure 7 is a module schematic diagram of the system for accelerating the aging test of the embedded mainboard based on the intelligent algorithm according to the embodiment of the present application. Figure 8 is an internal structure diagram of the electronic device according to the embodiment of the present application. DETAILED DESCRIPTION

[0024] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to be limiting of the present application. As used in the specification and the appended claims of the present application, the singular forms "a," "an" and "the" are intended to include both singular and plural forms, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or" as used herein refers to any or all possible combinations of one or more of the associated listed items.

[0025] Hereinafter, the terms "first" and "second" are only for the purpose of description, and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0026] The embodiments of the present application will be further described in detail below with reference to the accompanying drawings of the specification.

[0027] In a first aspect, the present application provides a method for accelerating the aging test of an embedded mainboard based on an intelligent algorithm, with reference to Figure 1 , comprising the following steps: S100, under a preset temperature cycle condition, measuring temperature distribution data, working voltage value and current value of the embedded mainboard, and generating real-time parameter data stream.

[0028] Among them, the preset temperature cycle condition includes applying a preset constant temperature stress to the mainboard, collecting initial performance parameters; switching the temperature in the temperature cycle range at a preset change rate; recording the change of electrical parameters of the mainboard during the temperature cycle; adjusting the temperature cycle duration according to the change of electrical parameters.

[0029] In this embodiment, the preset temperature cycle condition refers to the standardized temperature stress condition applied to the embedded mainboard, including constant temperature stress parameters and temperature change parameters. The constant temperature stress refers to the stable environmental temperature maintained within a certain time. The temperature cycle range refers to the upper and lower limit interval of temperature change. The preset change rate refers to the speed of temperature change. The electrical parameters include working voltage and current value. The real-time parameter data stream refers to the continuously recorded parameter measurement results according to the sampling period.

[0030] Specifically, first, the embedded mainboard is placed in a constant temperature environment, the environmental temperature is set between room temperature and high temperature, the duration is several hours, and the initial voltage and current parameters of the mainboard at this temperature are collected as reference values. Then, the temperature is cyclically switched in the room temperature to high temperature interval at a rate of several degrees Celsius per minute, while recording the working voltage and current value of the mainboard. When the fluctuation amplitude of voltage or current value exceeds the preset percentage of reference value, the duration of the current temperature cycle is correspondingly extended or shortened.

[0031] S200, compare the real-time parameter data stream with the preset performance reference database to determine the abnormality degree determination result.

[0032] In this embodiment, the performance reference database refers to the parameter range data set stored in the normal working state of the mainboard. The abnormality degree determination result refers to the quantitative index of the parameter deviation from the normal range. The performance reference database contains three types of reference data of temperature distribution characteristics, voltage fluctuation range and current change range. Each type of data establishes a corresponding parameter fluctuation upper and lower limit mapping table.

[0033] Specifically, first, the data in the last complete temperature cycle period is extracted from the real-time parameter data stream, and the mean, variance and peak value of the temperature, voltage and current three types of parameters are calculated as feature values. Compare these feature values with the mapping table in the performance reference database. When the parameter exceeds the normal fluctuation range, calculate the deviation degree according to the exceeding amplitude. The deviation degrees of the three types of parameters are respectively assigned weight coefficients for weighted summation to obtain the normalized abnormality degree determination result.

[0034] S300, input the abnormality degree determination result into the pre-trained machine learning model to generate the fault warning information.

[0035] In this embodiment, the pre-trained machine learning model refers to the LSTM deep neural network model trained by historical test data.

[0036] Specifically, the abnormality degree determination result is input into a pre-trained LSTM model, and the model outputs fault warning information according to the time sequence characteristics of the parameter anomaly. The pre-training process of the LSTM model uses historical data of temperature cycle tests, and the data is used for training after cleaning and labeling. When a new abnormality degree determination result is input, the model can predict the potential fault risk based on the learned feature patterns. For example, when the temperature distribution anomaly is detected and the duration exceeds the threshold, the model will output the corresponding type of fault warning.

[0037] S400, adaptively adjusting the temperature cycle condition and the voltage fluctuation amplitude according to the fault warning information.

[0038] In this embodiment, adaptive adjustment refers to dynamically adjusting the test parameters according to the abnormal parameter type and abnormality degree determination result in the fault warning information.

[0039] Specifically, first, the abnormal parameter type is extracted from the fault warning information to determine the stress type that needs to be strengthened. The abnormality degree determination result is multiplied by the temperature coefficient and the voltage coefficient respectively to obtain the adjustment amount of the temperature cycle range and the voltage fluctuation amplitude. The control parameters of the test equipment are updated according to the calculated adjustment amount.

[0040] S500, generating a fault diagnosis report based on the adjusted temperature cycle condition and voltage fluctuation amplitude, and evaluating the mainboard reliability state.

[0041] In this embodiment, the fault diagnosis report refers to a technical document recording the fault information found in the test process and the mainboard reliability evaluation result. The reliability state refers to the performance level of the mainboard under the current test conditions. The test conditions include the adjusted temperature cycle condition and the voltage fluctuation amplitude.

[0042] Specifically, first, record the adjusted temperature cycle parameters and voltage fluctuation parameters, including temperature range, change rate, voltage fluctuation value, etc. Test conditions. Monitor the performance parameters of the mainboard under these conditions, record abnormal phenomena and fault characteristics. According to the test results, write a fault diagnosis report, including test condition record, fault phenomenon description and performance evaluation result.

[0043] In one embodiment, referring to Figure 2 , in step S100, the temperature distribution data of the embedded mainboard is measured, including the following steps: S110, divide the embedded mainboard into multiple temperature monitoring areas, and obtain the distribution of monitoring points in each area.

[0044] In this embodiment, the temperature monitoring area refers to a temperature measurement unit divided according to the function modules and component distribution of the mainboard. The monitoring point distribution refers to the sampling position coordinates for measuring temperature in each area. The area division considers the physical size, device density and heat characteristics of the mainboard.

[0045] Specifically, first of all, according to the physical structure of the mainboard, it is divided into four functional areas of central processing area, power management area, interface area and storage area. In each area, the position of the monitoring point is determined according to the device layout, dense monitoring points are arranged around high-power devices, and sparse points are used in low-power areas. By establishing a regional division table to record the spatial coordinate information of each monitoring point, the position reference is provided for subsequent temperature measurement.

[0046] Further, according to the application scenario of the mainboard, a partition strategy library is established. Industrial control type mainboard focuses on power and interface area, sets higher monitoring point density; communication type mainboard strengthens the monitoring of radio frequency device area; calculation intensive mainboard arranges more monitoring points around processor and memory. At the same time, a heat conduction path table is established to record the thermal coupling relationship between key devices, and the distribution of monitoring points is optimized accordingly.

[0047] S120, periodically scan the monitoring points of each area with an infrared thermal imager to generate temperature field data.

[0048] Specifically, the infrared thermal imager is fixed above the test platform, and the lens is directed to the surface of the mainboard. The scanning period is set to several seconds, and the temperature collection of all monitoring points is completed in each period. The collected temperature data is stored in order according to the monitoring point coordinates to form a temperature field data matrix containing temperature values and position information.

[0049] S130, calculate the temperature gradient of each temperature monitoring area according to the temperature field data.

[0050] In this embodiment, the temperature gradient refers to the rate of change of temperature in space, which is used to represent the unevenness of temperature distribution.

[0051] Specifically, the temperature gradient between adjacent monitoring points in each monitoring area is calculated according to the temperature field data. First, a gradient calculation matrix is established, and the temperature difference between adjacent points is divided by the distance between points. The gradient values in each area are counted to obtain the regional temperature gradient distribution characteristics.

[0052] S140, identify the temperature stress concentration area based on the temperature gradient, and generate the temperature distribution characteristics.

[0053] In this embodiment, the temperature stress concentration area refers to the local area with large temperature gradient, which usually indicates abnormal heating phenomenon. The temperature distribution characteristics include the spatial distribution information of the temperature field and the position information of the stress concentration area.

[0054] Specifically, a temperature gradient threshold table is established, and the calculated regional temperature gradient is compared with the threshold. When the temperature gradient of a region exceeds the threshold, the region is marked as a temperature stress concentration region. The location information and temperature gradient value of the region are recorded, and characteristic data describing the temperature distribution state of the entire motherboard are generated.

[0055] Further, a temperature distribution pattern library is established, including typical temperature distribution patterns of each functional region in a normal working state. The measured temperature distribution characteristics are matched with the pattern library, and a similarity score is calculated. Meanwhile, a hotspot correlation matrix is established in combination with the device layout information, the spatial relationship between multiple temperature stress concentration regions is analyzed, and it is judged whether there is a heat conduction chain reaction. When an abnormal temperature distribution pattern or hotspot correlation phenomenon is found, the related region is marked as a key monitoring object, and the temperature gradient threshold is adjusted accordingly.

[0056] In one embodiment, with reference to Figure 3 In step S200, the abnormality degree determination result is determined by comparing the real-time parameter data stream with the preset performance benchmark database, including the following steps: S210, calculate the time series characteristic value of the real-time parameter data stream to obtain the parameter fluctuation trend.

[0057] In this embodiment, the time series characteristic value refers to the parameter statistical characteristics extracted within a specific observation time window. The parameter fluctuation trend refers to the dynamic characteristics of the parameter change over time.

[0058] Specifically, the sliding time window method is used to process the real-time parameter data stream, and the window length is one temperature cycle period. In each window, the mean, standard deviation, maximum value, minimum value, rising edge number and falling edge number of the parameter are calculated as characteristic values. A feature extraction table is established to record the characteristic calculation methods corresponding to different parameters, and automatic extraction of characteristic values is realized.

[0059] S220, difference operation is performed between the parameter fluctuation trend and the standard fluctuation range in the preset performance benchmark database to obtain deviation data.

[0060] In this embodiment, the performance benchmark database stores the parameter standard range in the normal working state of the motherboard. The standard fluctuation range refers to the upper and lower limit interval of the parameter allowed to change.

[0061] Specifically, a parameter standard range table is established to record the fluctuation range of each type of parameter in different working modes. The real-time acquired parameter fluctuation trend is difference calculated with the corresponding standard range to obtain the normalized deviation value. The parameter deviation evaluation is quickly completed through the table lookup method.

[0062] S230, the deviation data is normalized to generate a standardized abnormality index.

[0063] In this embodiment, normalization refers to converting different dimensional deviation data to the same scale. The standardized anomaly index is a dimensionless parameter after normalization, which is used to represent the degree of anomaly.

[0064] Specifically, a normalization processing table is established, and appropriate normalization methods are selected for different parameter types. Temperature deviation adopts maximum and minimum value normalization; voltage deviation uses mean normalization; current deviation adopts standard deviation normalization. The standardized anomaly index is obtained through normalization operation.

[0065] Further, an anomaly index correction rule is established, and the normalization result is adjusted according to the physical meaning of the parameter. Considering the coupling effect between parameters, when multiple parameters appear abnormal at the same time, the anomaly index is corrected to avoid the influence of repeated calculation.

[0066] S240, according to the standardized anomaly index, set the weight coefficient, calculate the comprehensive abnormality degree according to the weight coefficient, and classify the comprehensive abnormality degree according to the preset threshold value, and output the abnormality degree judgment result.

[0067] In this embodiment, the weight coefficient reflects the influence degree of different parameters on system anomaly. The comprehensive abnormality degree is the weighted result of multiple parameter anomalies. The preset threshold value is used to divide the abnormality degree into different levels.

[0068] Specifically, a weight distribution table is established, and the weight coefficient is set based on the importance of the parameter. The weight of temperature parameter is biased to the area with large temperature gradient; the weight of voltage parameter pays attention to the core power supply circuit; the weight of current parameter highlights the module with large power consumption. Multiply the standardized anomaly index by the weight coefficient to obtain the comprehensive abnormality degree.

[0069] In one embodiment, referring to Figure 4 , in step S300, the establishment process of the pre-trained machine learning model includes the following steps: S310, collect parameter data streams of multiple groups of embedded mainboards in the temperature cycle test process, and generate original training data set.

[0070] In this embodiment, the original training data set refers to the temperature cycle test data record without processing. The parameter data stream includes three basic parameters of temperature distribution, working voltage and running current. The data collection process covers the running state of the mainboard under different temperature cycle working conditions.

[0071] S320, calculate the abnormality degree of the parameter data stream, and perform outlier detection and data cleaning on the original training data set.

[0072] In this embodiment, outlier detection refers to identifying and processing data points that do not conform to the normal distribution rule. Data cleaning refers to removing noise interference and invalid data.

[0073] S330, construct a labeled dataset according to the abnormal degree and the corresponding fault record.

[0074] In this embodiment, the labeled dataset refers to a training sample set associating the abnormal degree with the actual fault record. The fault record contains three basic attributes: fault type, occurrence time, and impact range.

[0075] Specifically, a fault type mapping table is established to classify fault phenomena into three categories: temperature anomaly, voltage anomaly, and current anomaly. A label association rule table is created to specify the correspondence between abnormal degree and fault type. The abnormal degree data and fault record are paired in chronological order, and cross-checking is performed to ensure the accuracy of the label.

[0076] S340, train an LSTM deep neural network based on the labeled dataset to establish a mapping relationship from abnormal degree to fault warning.

[0077] In this embodiment, the LSTM deep neural network is a recurrent neural network structure suitable for processing time series data. The mapping relationship refers to the conversion rule learned by the network from abnormal degree to fault warning. The training process includes two stages: network structure design and parameter optimization.

[0078] Specifically, a network structure configuration table is constructed to specify the input layer dimension, number of hidden layers, and number of neurons. A training parameter table is created to set the learning rate, batch size, and number of training rounds. The cross-validation method is used to evaluate the model performance, and the optimal network configuration is selected to establish a model evaluation index system.

[0079] In one embodiment, referring to Figure 5 , in step S400, the temperature cycle operating conditions and voltage fluctuation amplitude are adaptively adjusted based on the fault warning information, which includes the following steps: S410, extract the abnormal parameter type and abnormal degree determination result from the fault warning information.

[0080] In this embodiment, the abnormal parameter type refers to the type of parameter that has occurred, including temperature distribution anomaly, voltage fluctuation anomaly, and current change anomaly.

[0081] Specifically, an early warning information analysis table is established to specify the data structure and analysis rules of the fault warning information. The abnormal parameter identifier and abnormal degree value are extracted from the warning information to generate a parameter abnormal feature vector. The abnormal parameter type is quickly identified through table lookup, and the corresponding abnormal degree determination result is obtained.

[0082] S420, determine the stress type that needs to be strengthened based on the abnormal parameter type.

[0083] In this embodiment, the stress type refers to the environmental stress applied to the mainboard, mainly including temperature stress and voltage stress. The enhanced stress refers to increasing the intensity of a specific stress according to abnormal conditions, which is used to verify the reliability of the mainboard under more severe conditions.

[0084] Specifically, a stress mapping table is established to associate abnormal parameter types with corresponding stress types. When temperature distribution is abnormal, the temperature cycle stress is determined to be enhanced; when voltage fluctuation is abnormal, the voltage fluctuation stress is determined to be enhanced. Suitable stress enhancement strategies are selected for different abnormal types.

[0085] S430, calculating the adjustment amount of the temperature cycle range and the voltage fluctuation amplitude based on the abnormality degree determination result.

[0086] The adjustment amount of the temperature cycle range is the product of the abnormality degree determination result and the temperature coefficient, and the adjustment amount of the voltage fluctuation amplitude is the product of the abnormality degree determination result and the voltage coefficient. The temperature coefficient and the voltage coefficient are obtained according to historical test data statistics.

[0087] In this embodiment, the adjustment amount refers to the specific value of the temperature cycle range and the voltage fluctuation amplitude that needs to be changed. The temperature coefficient and the voltage coefficient are standardized adjustment factors obtained from historical test data statistics, which are used to convert the abnormality degree into actual adjustment amount.

[0088] Specifically, a coefficient query table is established to record the temperature coefficient and the voltage coefficient under different working conditions. The abnormality degree determination result is multiplied by the temperature coefficient and the voltage coefficient respectively to calculate the adjustment amount of the temperature cycle range and the voltage fluctuation amplitude.

[0089] Further, the upper limit of the adjustment amount is set according to the mainboard type and the application scenario, and the adjustment amount is corrected. An adjustment amount cumulative record table is constructed to track the cumulative effect of continuous adjustment, and the adjustment limit is triggered when the cumulative adjustment amount reaches the upper limit. At the same time, parameter linkage rules are established, and when the temperature adjustment amount is large, the voltage adjustment amount is correspondingly reduced to avoid excessive stress causing damage to the mainboard.

[0090] S440, real-time updating the temperature cycle working condition and the voltage fluctuation parameter according to the adjustment amount.

[0091] In this embodiment, real-time updating refers to dynamically adjusting the test parameters during the test process. The temperature cycle working condition includes the temperature range and the change rate, and the voltage fluctuation parameter includes the fluctuation center value and the fluctuation amplitude.

[0092] Specifically, a parameter update rule table is established to specify the adjustment step and the update period of the test parameters. According to the calculated adjustment amount, the temperature cycle working condition and the voltage fluctuation parameter are adjusted step by step according to the rule table. The step-by-step adjustment method ensures the stability and controllability of the test process.

[0093] In one embodiment, referring to Figure 6 In step S500, a fault diagnosis report is generated, specifically including the following steps: S510, extracting abnormal parameter features and corresponding time series data from the fault warning information, and establishing a spatio-temporal correlation mapping of the abnormal parameter features and the temperature cycle parameters.

[0094] The abnormal parameter features include abnormal parameter types and abnormal degree determination results.

[0095] In this embodiment, the spatio-temporal correlation mapping refers to the association between the abnormal parameter features and the temperature cycle parameters in the time and space dimensions. The abnormal parameter features include two parts of information: abnormal parameter types and abnormal degree determination results.

[0096] Specifically, a parameter feature extraction table is established to specify the method of extracting abnormal parameter types and abnormal degrees from the fault warning information. A spatio-temporal correlation analysis table is created to record the distribution characteristics of abnormal parameter features at different temperature cycle stages, and to build a mapping relationship between the parameter features and the temperature cycle parameters.

[0097] S520, determining the fault type and the fault occurrence area based on the spatio-temporal correlation mapping, and generating a fault diagnosis report.

[0098] In this embodiment, the fault diagnosis report refers to a technical document describing the fault characteristics, the occurrence location, and the influence range. The fault type is determined according to the spatio-temporal correlation mapping result, and the fault occurrence area is identified through the spatial distribution characteristics of the parameter abnormalities.

[0099] Specifically, the spatio-temporal correlation mapping result is matched with a preset fault type mode to establish a fault feature identification table. A standardized fault diagnosis report is generated according to the matching result, including fault nature determination, area positioning result, and reliability evaluation conclusion.

[0100] It should be understood that the size of the serial number of each step in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0101] In a second aspect, the present application provides an embedded mainboard accelerated aging test system based on an intelligent algorithm. The embedded mainboard accelerated aging test system based on an intelligent algorithm of the present application will be described below in combination with the above-mentioned embedded mainboard accelerated aging test method based on an intelligent algorithm.

[0102] Referring to Figure 7 An embedded mainboard accelerated aging test system based on an intelligent algorithm, comprising: The parameter acquisition module is configured to measure temperature distribution data, working voltage value and current value of the embedded mainboard under a preset temperature cycle condition, and generate a real-time parameter data stream. The anomaly detection module is configured to compare the real-time parameter data stream with a preset performance benchmark database, and determine an anomaly degree determination result. The fault early warning module is configured to input the anomaly degree determination result into a pre-trained machine learning model, and generate fault early warning information. The test parameter adjustment module is configured to adaptively adjust the temperature cycle condition and voltage fluctuation amplitude according to the fault early warning information. The mainboard reliability evaluation module is configured to generate a fault diagnosis report based on the adjusted temperature cycle condition and voltage fluctuation amplitude, and evaluate a mainboard reliability state.

[0103] In one embodiment, the present application provides an electronic device, which can be a server, and an internal structure diagram thereof can be as shown in Figure 8 The electronic device includes a processor, a memory and a network interface connected through a system bus. The processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the electronic device is configured to store data. The network interface of the electronic device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement an embedded mainboard accelerated aging test method based on an intelligent algorithm.

[0104] Those skilled in the art can understand that Figure 8 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0105] In one embodiment, an electronic device is also provided, which includes a memory and a processor. The memory stores a computer program, and the processor implements the steps in each of the above method embodiments when executing the computer program.

[0106] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The above-mentioned computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the computer program can include the processes of the above-mentioned embodiments of each method. Wherein, any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include at least one of non-volatile and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. The volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not as a limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0107] The above are preferred embodiments of the present application, not to limit the protection scope of the present application, therefore: all equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method for accelerated aging testing of embedded motherboards based on intelligent algorithms, characterized in that, Includes the following steps: Under preset temperature cycling conditions, measure the temperature distribution data, operating voltage value and current value of the embedded motherboard, and generate a real-time parameter data stream; The anomaly level is determined by comparing the real-time parameter data stream with a preset performance benchmark database. The anomaly determination result is input into a pre-trained machine learning model to generate fault warning information; Based on the fault warning information, the temperature cycle conditions and voltage fluctuation amplitude are adaptively adjusted. A fault diagnosis report is generated based on the adjusted temperature cycle conditions and voltage fluctuation amplitude, and the motherboard reliability status is assessed.

2. The embedded motherboard accelerated aging test method based on intelligent algorithm according to claim 1, characterized in that, Measuring the temperature distribution data of an embedded motherboard involves the following steps: The embedded motherboard is divided into multiple temperature monitoring zones, and the distribution of monitoring points in each zone is obtained. Infrared thermal imagers are used to periodically scan monitoring points in each area to generate temperature field data; Calculate the temperature gradient of each of the temperature monitoring areas based on the temperature field data; Based on the temperature gradient, regions of concentrated temperature stress are identified, and temperature distribution characteristics are generated.

3. The embedded motherboard accelerated aging test method based on intelligent algorithm according to claim 2, characterized in that, The preset temperature cycle conditions include: Apply a preset constant temperature stress to the motherboard and collect initial performance parameters; The temperature is cyclically switched within a temperature cycle range at a preset rate of change. Record the changes in the motherboard's electrical parameters during temperature cycling; The duration of the temperature cycle is adjusted according to the changes in the electrical parameters.

4. The embedded motherboard accelerated aging test method based on intelligent algorithm according to claim 1, characterized in that, The anomaly level is determined by comparing the real-time parameter data stream with a preset performance benchmark database, including the following steps: Calculate the time-series feature values ​​of the real-time parameter data stream to obtain the parameter fluctuation trend; The difference between the parameter fluctuation trend and the standard fluctuation range in the preset performance benchmark database is calculated to obtain the deviation data. The deviation data is normalized to generate a standardized anomaly index; Weighting coefficients are set according to the standardized anomaly index, the comprehensive anomaly degree is calculated according to the weighting coefficients, and the comprehensive anomaly degree is classified according to a preset threshold, and the anomaly degree judgment result is output.

5. The embedded motherboard accelerated aging test method based on intelligent algorithm according to claim 1, characterized in that, The process of building the pre-trained machine learning model includes the following steps: Collect parameter data streams from multiple sets of embedded motherboards during temperature cycling tests to generate the original training dataset; Calculate the degree of anomaly in the parameter data stream, and perform outlier detection and data cleaning on the original training dataset; A labeled dataset is constructed based on the degree of anomaly and the corresponding fault records; Based on the labeled dataset, an LSTM deep neural network is trained to establish a mapping relationship between anomaly level and fault warning.

6. The embedded motherboard accelerated aging test method based on intelligent algorithm according to claim 1, characterized in that, Based on the fault warning information, the temperature cycle conditions and voltage fluctuation amplitude are adaptively adjusted, specifically including the following steps: Extract the abnormal parameter types and abnormality determination results from the fault warning information; The type of stress that needs to be strengthened is determined based on the type of abnormal parameters. The adjustment amounts for the temperature cycle range and voltage fluctuation amplitude are calculated based on the anomaly determination result. The adjustment amount for the temperature cycle range is the product of the anomaly determination result and the temperature coefficient, and the adjustment amount for the voltage fluctuation amplitude is the product of the anomaly determination result and the voltage coefficient. The temperature coefficient and voltage coefficient are obtained statistically from historical test data. The temperature cycle conditions and voltage fluctuation parameters are updated in real time according to the adjustment amount.

7. The embedded motherboard accelerated aging test method based on intelligent algorithm according to claim 1, characterized in that, Generating a fault diagnosis report includes the following steps: Extract abnormal parameter features and corresponding time series data from the fault warning information, and establish a spatiotemporal correlation mapping between the abnormal parameter features and temperature cycle parameters. The abnormal parameter features include the abnormal parameter type and the abnormality degree determination result. Based on the spatiotemporal correlation mapping, the fault type and fault occurrence area are determined, and a fault diagnosis report is generated.

8. An embedded motherboard accelerated aging test system based on intelligent algorithms, characterized in that, include: The parameter acquisition module is used to measure the temperature distribution data, operating voltage value and current value of the embedded motherboard under preset temperature cycling conditions, and generate a real-time parameter data stream. An anomaly detection module is used to compare the real-time parameter data stream with a preset performance benchmark database to determine the degree of anomaly. The fault warning module is used to input the anomaly degree determination result into a pre-trained machine learning model to generate fault warning information; The test parameter adjustment module is used to adaptively adjust the temperature cycling conditions and voltage fluctuation amplitude based on the fault warning information. The motherboard reliability assessment module is used to generate a fault diagnosis report based on the adjusted temperature cycling conditions and the voltage fluctuation amplitude, and to assess the motherboard reliability status.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the embedded motherboard accelerated aging test method based on intelligent algorithms as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the embedded motherboard accelerated aging test method based on intelligent algorithms as described in any one of claims 1-7.

Images