A test control method and system for electronic products

Abstract

This invention discloses a test control method and system for electronic products, applicable to the field of product testing. The electronic product test control method includes the following steps: constructing a modular test environment chamber, setting and controlling specific environmental parameters for each test module; optimizing test sequences and resource configuration; based on automatically executed working states and performance tests, and according to the characteristics of the modular test environment chamber, simultaneously applying varying environmental conditions to each test module and executing multiple tests; collecting key data generated during multiple tests, and using a distributed reinforcement learning algorithm to analyze the key data and identify design defects in the electronic product; and continuously monitoring abnormal conditions during the testing process. This invention can accurately simulate changing environmental conditions according to the actual usage scenarios of electronic products, ensuring efficient test operation and accurate test data, and reducing resource waste through optimized test sequences and resource configuration.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of product testing, in particular, to a test control method and system for electronic products. BACKGROUND

[0002] Electronic products are devices that primarily achieve their functions through electronic technology; these products contain electronic components and circuits, such as semiconductor devices, resistors, capacitors, and integrated circuits, and are widely used in communication, computing, entertainment, and other fields; the scope of electronic products is very broad, ranging from daily household appliances (such as televisions and refrigerators) and consumer electronics (such as mobile phones and laptops) to professional equipment (such as medical instruments and avionics).

[0003] The implementation of electronic product testing usually requires specialized testing equipment and environments, such as temperature and humidity controlled test rooms, electromagnetic compatibility (EMC) testing facilities, etc.; the testing process includes initial design verification, prototype testing, and final confirmation before production to ensure that the product meets the predetermined standards at each stage from design to production; in addition, feedback from test results is crucial for product design iteration and improvement, helping to optimize design, reduce defects, and improve user satisfaction.

[0004] Traditional testing methods may not accurately simulate the actual use environment of the product, such as temperature, humidity, vibration, etc., resulting in test results that do not fully reflect the performance of the product in real-world environments; old testing equipment is often not energy efficient in energy use, causing energy waste, and may have delays in detecting and responding to equipment failures, affecting the continuity of testing and the accuracy of data; due to the lack of comprehensive testing in complex and changing environments, the adaptability and durability of the product may not be fully verified, and the efficiency of testing sequences and resource allocation is low, resulting in resource waste and increased costs.

[0005] To address the problems in the related art, no effective solutions have been proposed so far. SUMMARY

[0006] To overcome the above problems, the present application aims to provide a test control method and system for electronic products, which aims to solve the problem that due to the lack of comprehensive testing in complex and changing environments, the adaptability and durability of the product may not be fully verified, and the efficiency of testing sequences and resource allocation is low, resulting in resource waste and increased costs.

[0007] To this end, the specific technical solutions adopted by the present application are as follows:

[0008] According to one aspect of the present application, a test control method for electronic products is provided, which includes the following steps:

[0009] S1. Set test conditions according to the expected use environment of electronic products, construct a modular test environment chamber, and set and adjust specific environmental parameters for each test module in the test environment chamber;

[0010] S2. In the modular test environment, pre-configured test scripts are used to automatically execute working status and performance tests, and optimize test sequences and resource configurations.

[0011] S3. Based on the automatic execution of working status and performance tests, according to the characteristics of the modular test environment, different environmental conditions are applied to each test module at the same time, and multiple tests are performed.

[0012] S4. In multiple tests of electronic products, collect test data on performance, durability, and other key indicators; configure Spark cluster resources according to the scale and complexity of the collected test data; use the ReliefF algorithm to evaluate the importance of the collected test data, and create multiple targeted feature subspaces based on the evaluation results; train multiple classification and regression decision trees in multiple feature subspaces using bootstrap sampling methods; integrate all classification and regression decision trees to obtain the final random forest model; use the final random forest model to analyze the test data, predict potential design defects, identify design defects related to defects, and optimize for the identified design defects.

[0013] S5. Implement a real-time anomaly detection mechanism to continuously monitor abnormal situations during the testing process and adjust test parameters immediately when abnormal situations occur.

[0014] Optionally, in a modular test environment, automated execution of operational and performance tests using pre-configured test scripts, and optimization of test sequences and resource configurations, includes the following steps:

[0015] S21. In the modular test environment chamber, set the required environmental parameters to predefined standard values ​​according to the test requirements;

[0016] S22. Run the pre-configured test script and monitor the script's execution status and performance metrics;

[0017] S23, Using rough sets Q The learning algorithm adjusts the workload and energy consumption of the test equipment, optimizing the utilization of the test equipment and the power supply.

[0018] S24. Conduct preliminary analysis and verification of the test results, collect feedback data, and ensure that each test achieves the preset goals.

[0019] Alternatively, rough sets can be used. QLearning algorithms adjust the workload and energy consumption of test equipment, optimizing equipment utilization and power supply, including the following steps:

[0020] S231. Determine the initial state and action set of the test equipment, and initialize... Q The table and parameters define the state space and action set, and set the initial parameters. Q Values, learning rate, and discount factor;

[0021] S232. Monitor and record the current workload and energy consumption status of the test equipment, and collect data as status input;

[0022] S233, According to the current Q A function that selects an action from the action set, executes the selected action, observes the reward obtained, and the next state;

[0023] S234, According to rough set Q Learning update rules adjustment Q value;

[0024] S235, according to the adjusted Q Adjust the workload and power settings of the test equipment, monitor the performance and energy consumption of the test equipment after adjustment, and evaluate the optimization effect;

[0025] S236. Repeat steps S232-S235 until the predetermined optimization goal is achieved.

[0026] Optionally, based on rough set Q Learning update rules adjustment Q The expression for the value is:

[0027] ;

[0028] In the formula, Indicates that it is in the execution state. e Take action below d Expected returns;

[0029] Indicates the learning rate;

[0030] r Indicates the execution of an action e In state d The reward received directly;

[0031] Indicates the discount factor;

[0032] n Indicates the number of steps;

[0033] Indicates the next execution state e All actions in ' d 'of Q Maximum value;

[0034] v ′ represents the average inclusion value calculated based on rough set theory;

[0035] e Indicates the execution of an action;

[0036] e ′ indicates the next execution state;

[0037] d Indicates an action;

[0038] d ′ indicates all actions.

[0039] Optionally, according to the adjusted Q Adjusting the workload and power settings of the test equipment, monitoring the performance and energy consumption of the test equipment after adjustment, and evaluating the optimization effect include the following steps:

[0040] S2351, from Q The table identifies the optimal action for each state, and the operating parameters of the test equipment are adjusted accordingly to match the recommended optimal action;

[0041] S2352. Implement parameter adjustments for the test equipment and simultaneously initiate real-time monitoring of the workload, energy consumption, and performance indicators of the adjusted test equipment.

[0042] S2353. Collect the adjusted performance data, analyze the performance data, and determine the actual impact of the adjustment on the performance of the test equipment.

[0043] S2354. Compare performance data before and after the adjustment to evaluate improvements in energy efficiency and workload management.

[0044] Optionally, collecting and analyzing the adjusted performance data to determine the actual impact of the adjustment on the performance of the test equipment includes the following steps:

[0045] S23531. Determine key performance indicators and set the frequency and duration of data collection;

[0046] S23532. Real-time collection of performance data during equipment operation;

[0047] S23533. Analyze performance data using time series analysis and configure data charts;

[0048] S23534. Based on the analysis results of performance data, evaluate the effectiveness of each adjustment measure, and compile the analysis results and data charts into a detailed report.

[0049] Optionally, based on automatically executed working status and performance tests, and according to the characteristics of the modular test environment, varying environmental conditions are applied to each test module simultaneously, and multiple tests are performed, including the following steps:

[0050] S31. Based on the actual usage scenarios of electronic products, determine the simulated environmental conditions and set performance evaluation standards;

[0051] S32. Adjust the environmental conditions of the test room according to the requirements of each test module;

[0052] S33. Start the test program for the electronic product in the modular test environment chamber and record the running data of all modules;

[0053] S34. During the test, change the environmental conditions according to the predetermined plan, and monitor the impact of the environmental changes on the performance of each module; at the same time, conduct multiple performance tests to comprehensively evaluate the product performance.

[0054] S35. Collect performance data of all test modules and evaluate the performance of each module under various environmental conditions through statistical and comparative analysis.

[0055] Optionally, the ReliefF algorithm is used to evaluate the importance of the collected test data, and multiple targeted feature subspaces are created based on the evaluation results. Creating multiple feature subspaces includes the following steps:

[0056] S431. Extract feature data from the collected test data;

[0057] S432. Randomly select a preset number of instances from the test data, and find the nearest neighbors of the same type and the nearest neighbors of different types for each selected instance;

[0058] S434. Update the weights of each feature data based on the feature differences between the target instance and its similar and dissimilar neighbors;

[0059] S435. By repeating steps S433-S434, multiple instances are evaluated, and the average weight of each feature throughout the process is calculated as the importance score.

[0060] S436. Sort all features according to their importance scores, and select the features with the highest importance to create multiple feature subspaces.

[0061] Optionally, multiple classification and regression decision trees are trained using bootstrap sampling in multiple feature subspaces; and all classification and regression decision trees are integrated to obtain the final random forest model, including the following steps:

[0062] Select an appropriate training dataset from the multiple feature subspaces created;

[0063] During the training of each decision tree, a bootstrap sampling method is used to randomly draw samples from the training dataset with replacement.

[0064] The dataset obtained through bootstrap sampling is used to independently train each classification and regression decision tree, allowing each tree to learn the different features of the feature data.

[0065] All trained classification and regression decision trees are merged into a random forest model;

[0066] Based on the training results and preliminary validation feedback, the parameters of the random forest model were adjusted.

[0067] According to another aspect of the present invention, a test control system for electronic products is also provided, the electronic product test control system comprising: a test environment construction module, an automated test execution module, a dynamic environment test module, a data analysis module, and an anomaly detection and response module;

[0068] The test environment construction module, automated test execution module, dynamic environment test module, data analysis module, and anomaly detection and response module are sequentially connected.

[0069] The test environment construction module is used to set test conditions according to the expected use environment of electronic products, build a modular test environment chamber, and set and adjust specific environmental parameters for each test module in the test environment chamber;

[0070] The automated test execution module is used to automatically execute working status and performance tests in a modular test environment using pre-configured test scripts, and optimize test sequences and resource configurations.

[0071] The dynamic environment testing module is used for automatically executed working state and performance testing. Based on the characteristics of the modular test environment, it applies changing environmental conditions to each test module and executes multiple tests.

[0072] The data analysis module is used to collect test data on performance, durability, and other key indicators in multiple tests of electronic products; configure Spark cluster resources according to the scale and complexity of the collected test data; use the ReliefF algorithm to evaluate the importance of the collected test data and create multiple targeted feature subspaces based on the evaluation results; train multiple classification and regression decision trees in multiple feature subspaces using bootstrap sampling methods; integrate all classification and regression decision trees to obtain the final random forest model; use the final random forest model to analyze the test data, predict potential design defects, identify design defects related to defects, and optimize for the identified design defects;

[0073] The anomaly detection and response module is used to implement a real-time anomaly detection mechanism, continuously monitor abnormal conditions during the testing process, and immediately adjust test parameters when an anomaly occurs.

[0074] Compared with the prior art, this application has the following beneficial effects:

[0075] 1. This invention uses an automated control system to precisely set and adjust environmental parameters, ensuring the repeatability and consistency of tests, which helps to accurately evaluate the performance and stability of products under different environments; it utilizes rough sets... Q Learning algorithms optimize the workload and energy consumption of testing equipment, effectively improving energy efficiency and equipment utilization while reducing energy waste. Real-time monitoring and data analysis enable timely identification and response to abnormal states and performance degradation during testing, accelerating fault diagnosis and repair. Detailed performance data and adjustment effect evaluation reports provide feedback for product design and testing processes, supporting continuous improvement and iterative development. The rapid adjustment of automated test scripts and environmental parameters enhances the automation and flexibility of the testing process, adapting to rapidly changing testing needs and complex testing scenarios.

[0076] 2. This invention can accurately simulate varying environmental conditions, such as temperature, humidity, and vibration, based on the actual usage scenarios of electronic products, thereby improving the practicality and reliability of test results. Simultaneously, by applying varying environmental conditions to each test module and performing multiple tests, it allows for a comprehensive evaluation of the product's performance under different environmental conditions, enhancing the product's adaptability and durability. By collecting key data and utilizing distributed reinforcement learning algorithms, it can not only identify design defects but also perform targeted optimizations to promote continuous improvement in product design. The implementation of a real-time anomaly detection mechanism enables timely identification and adjustment of abnormal situations during testing, ensuring efficient test operation and accurate test data. By optimizing test sequences and resource allocation, it reduces resource waste and improves the cost-effectiveness of the testing process. Attached Figure Description

[0077] The above-mentioned features, characteristics, and advantages of the present invention, as well as their implementation methods, will become clearer and more readily understood in conjunction with the following description of the embodiments, which are illustrated in detail with reference to the accompanying drawings. Schematic diagrams are shown here:

[0078] Figure 1 This is a flowchart of a test control method for electronic products according to an embodiment of the present invention;

[0079] Figure 2 This is a schematic block diagram of a test control system for electronic products according to an embodiment of the present invention.

[0080] In the picture:

[0081] 1. Test environment construction module; 2. Automated test execution module; 3. Dynamic environment testing module; 4. Data analysis module; 5. Anomaly detection and response module. Detailed Implementation

[0082] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0083] According to an embodiment of the present invention, a test control method and system for electronic products are provided.

[0084] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, according to an embodiment of the present invention, a test control method for electronic products is provided, the electronic product test control method comprising the following steps:

[0085] S1. Set test conditions according to the expected usage environment of electronic products, construct a modular test environment chamber, and set and adjust specific environmental parameters for each test module in the test environment chamber.

[0086] It needs to be explained that key environmental parameters are identified based on the expected usage environment of the electronic product; a detailed analysis of the product's usage scenarios is conducted to understand the various conditions that the product may encounter in actual use. Based on these analyses, specific parameters to be simulated in the test environment are set, such as temperature range of -10°C to 50°C, humidity of 40% to 90%, etc.

[0087] Before the test begins, environmental parameters are initialized according to the set test conditions. Using an automated control system can improve the accuracy and repeatability of the settings. During the test, environmental parameters are monitored in real time and adjusted as needed. For example, if the temperature deviates from the set range, the heating or cooling system is automatically or manually adjusted to restore the target state. All changes in environmental parameters are recorded for subsequent analysis.

[0088] S2. In the modular test environment, pre-configured test scripts are used to automatically execute working status and performance tests, and optimize test sequences and resource configurations.

[0089] Preferably, in a modular testing environment, automatically executing operational status and performance tests using pre-configured test scripts, and optimizing test sequences and resource configurations includes the following steps:

[0090] S21. In the modular test environment chamber, set the required environmental parameters to predefined standard values ​​according to the test requirements.

[0091] S22. Run the pre-configured test script and monitor the script's execution status and performance metrics.

[0092] It is important to explain that before testing begins, the standard values ​​for the required environmental parameters must be determined. These standard values ​​are typically based on the product's design specifications and the intended operating environment. An automated control system is used to precisely set and adjust the environmental parameters in the test chamber. Rapid adjustment of these parameters is achieved through programming, ensuring the consistency and repeatability of the test. The control system usually includes sensors and actuators. Sensors monitor the actual environmental parameters, while actuators adjust the equipment according to a preset program to achieve the required parameters. A series of parameter verifications are performed before testing to ensure that all setpoints are accurately achieved. Based on the test objectives and product specifications, test scripts are pre-written or configured. These scripts define the test steps, actions to be performed, and expected results. The scripts include commands and instructions from automated testing software, which can simulate user operations and monitor the system. Once the test environment is ready, the test script is launched. The script will execute a series of operations in a predetermined order, such as opening the application, inputting data, and simulating user interaction. The script can be executed locally or on a remote server to simulate different network conditions and usage scenarios. During test execution, system logs and performance counters are monitored to track the script's execution status in real time. This includes tracking the script's progress, detecting errors and abnormal behavior, and using the built-in functions of automated testing tools to capture screenshots, log output, and network requests during execution for troubleshooting and performance analysis. The execution performance of the test script is evaluated based on predefined performance metrics, including response time, throughput, and resource utilization. The execution results and performance metrics of the test script are recorded, including success, failure, error messages, and performance data.

[0093] S23, Using rough sets Q The learning algorithm adjusts the workload and energy consumption of the test equipment, optimizing the utilization rate of the test equipment and the power supply.

[0094] Preferably, rough sets are used. Q Learning algorithms adjust the workload and energy consumption of test equipment, optimizing equipment utilization and power supply, including the following steps:

[0095] S231. Determine the initial state and action set of the test equipment, and initialize... Q The table and parameters define the state space and action set, and set the initial parameters. Q Values, learning rate, and discount factor.

[0096] S232. Monitor and record the current workload and energy consumption status of the test equipment, and collect the data as status input.

[0097] S233, According to the current Q The function selects an action from the action set, executes the selected action, and observes the reward and the next state.

[0098] S234, According to rough set Q Learning update rules adjustment Q value.

[0099] Preferably, based on rough sets Q Learning update rules adjustment Q The expression for the value is:

[0100] ;

[0101] In the formula, Indicates that it is in the execution state. e Take action below d Expected returns;

[0102] Indicates the learning rate;

[0103] r Indicates the execution of an action e In state d The reward received directly;

[0104] Indicates the discount factor;

[0105] n Indicates the number of steps;

[0106] Indicates the next execution state e All actions in ' d 'of Q Maximum value;

[0107] v ′ represents the average inclusion value calculated based on rough set theory;

[0108] e Indicates the execution of an action;

[0109] e ′ indicates the next execution state;

[0110] d Indicates an action;

[0111] d ′ indicates all actions.

[0112] It needs to be explained that determining the initial state, such as whether the test equipment is idle, under medium or high load, and defining the action set, including increasing or decreasing power output, starting or stopping additional modules, etc., specifically includes: creating... Q A table is used to store the expected reward for each state and action pair. QThe table is a two-dimensional array, where rows represent possible states and columns represent possible actions; initialization. Q All values ​​in the table are zero or initial values ​​set based on prior knowledge; a learning rate (e.g., 0.1) is set to adjust... Q The rate at which values ​​are updated; setting a discount factor (e.g., 0.9) to calculate the present value of future rewards.

[0113] Use sensors and monitoring tools to monitor the workload and energy consumption of the test equipment in real time; the collected data may include indicators such as CPU utilization, memory usage, energy consumption, and temperature. Record the collected data as... Q For learning status input, data recording should ensure the accuracy of timestamps to facilitate analysis of status changes over time and to utilize the current state. Q The table determines the action to be performed in a specific state; typically, the selection... Q The action with the greatest value, but a trade-off between exploration and exploitation is needed. Sometimes random actions are chosen to explore unknown strategies. Executing selected actions, such as adjusting the power settings of test equipment, observing changes in system workload and energy consumption after the action is executed, and the rewards obtained.

[0114] Rough set theory can be used to handle uncertainty and imprecision, improving performance. Q Learning efficiency; based on rough sets Q The learning update rule uses the observed reward and the next state to adjust the Q value of the current state and action.

[0115] S235, according to the adjusted Q Adjust the workload and energy settings of the test equipment, monitor the performance and energy consumption of the test equipment after adjustment, and evaluate the optimization effect.

[0116] Preferably, according to the adjusted Q Adjusting the workload and power settings of the test equipment, monitoring the performance and energy consumption of the test equipment after adjustment, and evaluating the optimization effect include the following steps:

[0117] S2351, from Q The table identifies the optimal action for each state, and the operating parameters of the test equipment are adjusted accordingly to match the recommended optimal action;

[0118] S2352. Implement parameter adjustments for the test equipment and simultaneously initiate real-time monitoring of the workload, energy consumption, and performance indicators of the adjusted test equipment.

[0119] S2353. Collect the adjusted performance data, analyze the performance data, and determine the actual impact of the adjustment on the performance of the test equipment.

[0120] S2354. Compare performance data before and after the adjustment to evaluate improvements in energy efficiency and workload management.

[0121] Preferably, collecting and analyzing the adjusted performance data to determine the actual impact of the adjustment on the performance of the test equipment includes the following steps:

[0122] S23531. Determine key performance indicators and set the frequency and duration of data collection;

[0123] S23532. Real-time collection of performance data during equipment operation;

[0124] S23533. Analyze performance data using time series analysis and configure data charts;

[0125] S23534. Based on the analysis results of performance data, evaluate the effectiveness of each adjustment measure, and compile the analysis results and data charts into a detailed report.

[0126] S236. Repeat steps S232-S235 until the predetermined optimization goal is achieved.

[0127] It needs to be explained that, from Q The table identifies the optimal action for each state, which can be found by searching in each state row. Q The action with the highest value is the optimal action, which reflects the most effective equipment operation strategy identified based on historical data and the learning process. Based on the identified optimal action, the operating parameters of the test equipment are adjusted, such as adjusting energy input, modifying operating modes, or adjusting operating speed. These adjustments aim to optimize equipment performance and energy efficiency while minimizing power consumption and increasing output. Parameter adjustments are implemented in the equipment control system to ensure all changes are safe and correctly executed. This requires manual adjustments by technicians or remote updates of equipment configuration via automated systems. Real-time monitoring of the adjusted equipment is initiated, including workload, energy consumption, and performance indicators. Monitoring software and hardware are used to track and record equipment status in real time. The equipment's workload can be measured by analyzing the frequency and complexity of its processing tasks. Energy consumption can be monitored using power meters and energy consumption monitors. Performance indicators may include processing speed, error rate, response time, etc., which are typically obtained by the equipment's built-in measurement tools.

[0128] When identifying key performance indicators (KPIs), such as energy efficiency, throughput, and error rate, accurate assessment of equipment performance is crucial. Setting the frequency and duration of data collection ensures data sufficiency and real-time performance. During equipment operation, performance data is collected in real-time through the equipment's built-in data log function or an external data acquisition system, ensuring continuity and consistency. Utilizing time-series analysis to analyze the collected performance data helps identify performance trends and patterns over time. Time-series analysis can reveal performance fluctuations and potential cyclical changes. The analysis results should be formatted into easily understandable charts, such as line graphs, bar charts, or pie charts, so that even non-technical personnel can quickly grasp the results. Based on the performance data analysis, the actual effect of each adjustment measure is evaluated, and performance data before and after the adjustment is compared to quantify the degree of improvement. The analysis results and data charts are compiled into a detailed report, which should include the adjustment measures, performance changes, evaluation results, and recommendations.

[0129] S24. Conduct preliminary analysis and verification of the test results, collect feedback data, and ensure that each test achieves the preset goals.

[0130] S3. Based on the automatic execution of working status and performance tests, according to the characteristics of the modular test environment, different environmental conditions are applied to each test module at the same time, and multiple tests are executed.

[0131] Preferably, based on automatically executed working status and performance tests, and according to the characteristics of the modular test environment, simultaneously applying varying environmental conditions to each test module and performing multiple tests includes the following steps:

[0132] S31. Based on the actual usage scenarios of electronic products, determine the simulated environmental conditions and set performance evaluation standards;

[0133] S32. Adjust the environmental conditions of the test room according to the requirements of each test module;

[0134] S33. Start the test program for the electronic product in the modular test environment chamber and record the running data of all modules;

[0135] S34. During the test, change the environmental conditions according to the predetermined plan, and monitor the impact of the environmental changes on the performance of each module; at the same time, conduct multiple performance tests to comprehensively evaluate the product performance.

[0136] S35. Collect performance data of all test modules and evaluate the performance of each module under various environmental conditions through statistical and comparative analysis.

[0137] It needs to be explained that, based on the actual usage scenario of the electronic product, such as outdoor, industrial, or home environments, corresponding environmental conditions are determined for simulation. These include factors such as temperature, humidity, vibration, dust, and light, and the range of environmental conditions to be simulated is determined, for example, temperature from -20°C to 50°C, and humidity from 10% to 95%. Performance evaluation standards are set, including the electronic product's response time, battery life, and mechanical strength, ensuring that these standards can comprehensively evaluate the product's performance under different environmental conditions. The environmental conditions in the modular testing environment chamber are adjusted according to the needs of each test module, such as adjusting temperature and humidity controllers and setting the vibration level of the vibration table. An automated control system is used to precisely adjust and maintain the required environmental conditions to maintain test consistency and repeatability. In the modular testing environment chamber, the tests for the electronic product are initiated. The testing procedure includes software applications and hardware operations, ensuring that all testing equipment and software are correctly configured to execute the predetermined testing tasks. It records the operational data of all modules during the testing process, including but not limited to performance indicators and real-time changes in environmental conditions. Data acquisition systems, such as sensors and data loggers, are used to ensure that all critical information is captured and recorded. Environmental conditions are gradually or suddenly changed during the testing process according to a predetermined plan, such as temperature jumps or gradual increases in humidity. Multiple performance tests are conducted simultaneously, such as stress tests and durability tests, to comprehensively evaluate product performance. The impact of environmental changes on the performance of each test module is monitored, and an advanced monitoring system is used to track and analyze data in real time. Performance data of all test modules under different environmental conditions is collected, and statistical methods and comparative analysis are used to evaluate the performance of each module under different environmental conditions.

[0138] The system comprehensively evaluates the performance of electronic products in a simulated environment and guides further improvements and optimizations through precise data analysis. This not only ensures that products maintain performance under extreme and changing environmental conditions but also helps R&D teams understand potential weaknesses and areas for improvement.

[0139] S4. Collect key data generated during multiple tests and analyze the key data using distributed reinforcement learning algorithms to identify design defects in electronic products and formulate corresponding improvement measures. Specifically, this includes: collecting test data on performance, durability, and other key indicators in multiple tests of electronic products; configuring Spark cluster resources according to the scale and complexity of the collected test data; using the ReliefF algorithm to evaluate the importance of the collected test data and creating multiple targeted feature subspaces based on the evaluation results; training multiple classification and regression decision trees in multiple feature subspaces using bootstrap sampling methods; integrating all classification and regression decision trees to obtain the final random forest model; using the final random forest model to analyze the test data, predict potential design defects, identify design defects related to defects, and optimize for the identified design defects.

[0140] It is important to explain that the test plan includes the types of tests (e.g., performance tests, durability tests), test methods, number of tests, and frequency. This ensures the test plan comprehensively covers the product's key performance indicators, such as speed, accuracy, and stability. Each test is executed according to the plan, including both manual and automated tests. During testing, various testing tools and instruments are used to collect data, such as performance analysis tools and data loggers. Test data on performance, durability, and other key indicators is collected in real-time to ensure accuracy and completeness for subsequent analysis and evaluation. The scale and complexity of the collected test data are assessed, including the amount, diversity, and structure of the data. Understanding the data characteristics is crucial for configuring appropriate Spark cluster resources. Based on the data's scale and complexity, appropriate Spark cluster resources are configured, including selecting suitable hardware resources (e.g., CPU, memory, storage) and configuring the appropriate software environment (e.g., Spark version, cluster architecture). Cluster resources are optimized and adjusted based on the cluster's operating status to improve performance and efficiency.

[0141] Preferably, the ReliefF algorithm is used to evaluate the importance of the collected test data, and multiple targeted feature subspaces are created based on the evaluation results. Creating multiple feature subspaces includes the following steps:

[0142] Extract feature data from the collected test data;

[0143] A preset number of instances are randomly selected from the test data, and for each selected instance, similar nearest neighbors and dissimilar nearest neighbors are found;

[0144] Update the weights of each feature data based on the feature differences between the target instance and its similar and dissimilar neighbors;

[0145] By repeatedly executing steps S433-S434, multiple instances are evaluated, and the average weight of each feature throughout the process is calculated as the importance score.

[0146] Based on importance scores, all features are sorted, and multiple feature subspaces are created by selecting the features with the highest importance.

[0147] It needs to be explained that all relevant feature data is extracted from the test dataset. The feature data includes numerical features, categorical features, and calculated derived features. The extracted feature data is preprocessed, including normalization and handling of missing values. A preset number of instances are randomly selected from the test dataset. For each selected instance, a distance metric is used to find nearest neighbors of the same class and nearest neighbors of different classes. Nearest neighbors of the same class are the nearest neighbor samples that belong to the same class as the target instance, and nearest neighbors of different classes are the nearest neighbor samples that belong to different classes as the target instance.

[0148] For each selected target instance, calculate its feature differences with its nearest neighbors of the same and different classes. Feature differences are obtained by calculating the differences in feature values ​​between instances, such as the difference in the value of a target instance and its nearest neighbors on a certain feature. Based on the calculated feature differences, update the weights of each feature. Typically, the weights of features that can effectively distinguish between similar and dissimilar samples will be increased. The weight update formula is based on the rules of the ReliefF algorithm, and multiple instances are evaluated. This repeated execution helps ensure the stability and reliability of the evaluation results. Calculate the average weight of each feature throughout the process. This can be achieved by averaging the feature weights obtained from evaluating all instances. Sort all features according to the calculated importance score. The higher the score, the more important the feature. The sorting process can be implemented using a standard sorting algorithm. Based on the sorting results, select the most important features to create multiple feature subspaces. Each feature subspace includes a set of important features for different analysis or model training tasks. The number of features selected can be determined according to specific needs, such as selecting the top 10 most important features.

[0149] Δ W ( f =diff( x , NM , f ) ( x , NH , f );

[0150] diff x , NM , f ) represents the target instance xIts nearest outlier (NM) in terms of features f The absolute value of the difference;

[0151] Δ W ( f ) represents features f The change in weight;

[0152] diff x , NH , f ) represents the target instance x Its nearest neighbor (NH, Nearest Hit) in terms of characteristics f The absolute value of the difference.

[0153] The ReliefF algorithm is a machine learning algorithm for feature selection, primarily used to identify the most discriminative features in classification tasks. It is an extension of the Relief algorithm, which can better handle noisy data, multi-class classification problems, and imbalanced datasets. The core idea of ​​the ReliefF algorithm is to determine the importance of each feature by evaluating its ability to distinguish neighboring samples.

[0154] Preferably, training multiple classification and regression decision trees using bootstrap sampling in multiple feature subspaces, and integrating all classification and regression decision trees to obtain the final random forest model includes the following steps:

[0155] Select an appropriate training dataset from the multiple feature subspaces created;

[0156] During the training of each decision tree, a bootstrap sampling method is used to randomly draw samples from the training dataset with replacement.

[0157] The dataset obtained through bootstrap sampling is used to independently train each classification and regression decision tree, allowing each tree to learn the different features of the feature data.

[0158] All trained classification and regression decision trees are merged into a random forest model;

[0159] Based on the training results and preliminary validation feedback, the parameters of the random forest model were adjusted.

[0160] It's important to explain that a suitable dataset for training classification and regression decision trees is selected from multiple feature subspaces. Each subspace contains different feature combinations, and for different prediction tasks or data features, the selected datasets are preprocessed, including handling missing values ​​and normalization, to optimize model training. During the training of each decision tree, a bootstrap sampling method is used, which involves randomly sampling with replacement from the entire training dataset to create multiple different training subsets. Each subset may contain duplicate samples, which helps improve the model's generalization ability. Using the datasets obtained through bootstrap sampling, each classification and regression decision tree is trained independently. Each tree only accesses the features in its sampled dataset, learning the relationship between these features and the prediction target. In relational, classification, and regression decision trees, the best splitting feature is selected at each node during training. This is achieved by calculating information gain or Gini impurity. All independently trained classification and regression decision trees are then ensembled into a single random forest model. Random forests improve the overall accuracy and stability of predictions by integrating the predictions of multiple trees. Ensemble methods typically involve simple voting or averaging predictions, with each tree's prediction considered an independent vote. Preliminary performance evaluations are conducted based on the model's performance on the training and validation sets, focusing on key metrics such as accuracy, recall, and F1 score. Based on the training results and preliminary validation feedback, the parameters of the random forest model are adjusted, including the number of trees, tree depth, and the minimum number of samples required for splitting.

[0161] It's important to clarify that the test data must have undergone appropriate preprocessing, such as handling missing values ​​and normalization, to ensure its format matches the training data. The same feature subspace used when training the random forest model should be employed to ensure the test data contains all relevant features. The trained random forest model should be loaded, and the data prepared for analysis. The random forest model should be used to predict performance issues or design flaws. Predictions should be based on patterns learned by the model, identifying features similar to known flaws. The model's predictions should be analyzed to identify potential design flaws. This involves a detailed review of the model output and the use of decision tree path analysis to understand which features most significantly influence flaw predictions, identifying features associated with identified flaws. Features provide insights into which aspects of the design might lead to performance degradation. Identified flaws should be fed back to the design team for product design optimization. This includes adjusting design parameters that affect performance and reliability, and creating detailed analysis reports, including model-predicted flaws, key performance-affecting factors, and recommended design adjustments.

[0162] The distributed reinforcement learning algorithm is one of the core supporting algorithms of this invention, and its operating environment needs to be compatible with the modular testing system and Spark cluster resource configuration of this invention. To enable those skilled in the art to clearly understand and implement this algorithm, the environment on which it is based is described in detail below:

[0163] 1. Hardware operating environment: This distributed reinforcement learning algorithm is based on the Spark cluster hardware architecture configured in this invention. The specific environment parameters are as follows:

[0164] The cluster nodes must contain at least one master node and three or more slave nodes; the master node hardware must have ≥8 CPU cores (≥2.8GHz), ≥32GB of memory, ≥1TB of storage (SSD), and ≥10Gbps of network bandwidth; each slave node hardware must have ≥6 CPU cores (≥2.4GHz), ≥16GB of memory, ≥500GB of storage, and ≥1Gbps of network bandwidth; each node of the auxiliary hardware must be configured with an environmental sensor data interface (supporting RS485 / Ethernet protocols) to acquire environmental parameters (temperature, humidity, etc.) of the modular test environment chamber in real time, as part of the algorithm input data.

[0165] 2. Software support environment:

[0166] The operating system's master and slave nodes uniformly use Linux CentOS 7.9 or Ubuntu 20.04 LTS; the distributed computing framework is Spark 3.2.0 or later, configured with YARN as the resource scheduler to ensure resource isolation and efficient allocation between algorithm and test data processing tasks; the algorithm's dependency library is Python 3.8+, equipped with TensorFlow 2.8.0 (distributed version) or PyTorch 1.12.0 (supporting distributed training), used to implement parameter updates and iterations of the reinforcement learning model; the data storage environment uses HDFS (Hadoop Distributed File System) 3.3.0 or later, used to store massive amounts of critical data such as performance and durability collected during testing, supporting the algorithm's rapid reading and analysis of historical data; the communication protocol uses RPC (Remote Procedure Call) to synchronize algorithm parameters between cluster nodes, and uses REST API interfaces to implement data interaction between the algorithm and other modules of the testing system (such as the dynamic environment testing module and the anomaly detection module).

[0167] 3. The interaction environment between the algorithm and the system:

[0168] Data input environment: The algorithm reads the test data (feature data preprocessed by the ReliefF algorithm) stored in HDFS through the DataFrame interface of the Spark cluster, and at the same time receives the dynamic environmental parameters (such as temperature change rate, humidity fluctuation value, etc.) of the modular test environment in real time through the sensor interface. The two types of data together constitute the input dataset of the algorithm.

[0169] Data output environment: The analysis results of the algorithm (including the results of potential design defect identification and feature importance weight update data) are pushed to the random forest model of the data analysis module through a message queue (Kafka 2.8.0) for model parameter optimization. At the same time, it is synchronized to the anomaly detection and response module to provide a basis for decision-making for test parameter adjustment.

[0170] Running a collaborative environment: The training and inference process of the algorithm is synchronized with the dynamic testing process of the testing system. Through the dynamic resource adjustment function of the Spark cluster, computing resources are allocated in real time according to the generation speed of test data to ensure that the algorithm running does not affect the continuity of the testing process (such as avoiding delays in environmental parameter adjustment due to excessive CPU resources occupied by the algorithm).

[0171] S5. Implement a real-time anomaly detection mechanism to continuously monitor abnormal situations during the testing process and adjust test parameters immediately when abnormal situations occur.

[0172] It should be explained that a real-time anomaly detection system needs to be designed during continuous monitoring testing. This system should be able to receive and analyze the data generated during the testing process. The anomaly detection system should include a data receiving module, a data processing module, an anomaly detection algorithm, and a response mechanism. Appropriate anomaly detection algorithms should be selected based on the characteristics of the test data, such as statistical methods, machine learning models, or rule-based methods. The algorithm should be able to quickly identify data points that deviate from the normal range, ensuring that all test equipment can send data to the anomaly detection system in real time. Necessary preprocessing should be performed on the collected data, such as scaling, noise reduction, and missing value imputation, to improve data quality and ensure the effectiveness of the anomaly detection algorithm. The anomaly detection system continuously monitors the incoming data stream, applying the anomaly detection algorithm in real time to identify potential anomalies. Once an anomaly is detected, it will determine whether to trigger a response based on preset rules or thresholds. After confirming the anomaly, it will automatically adjust test parameters or send an alarm to the operator. Adjustments may include modifying environmental conditions, changing the test frequency, or pausing the test for further analysis. All detected anomaly events and their contextual information should be recorded for subsequent analysis to identify the cause of the anomaly. The performance of the anomaly detection system should be evaluated regularly, including detection accuracy, response time, and stability.

[0173] According to another embodiment of the invention, such as Figure 2As shown, a test control system for electronic products is also provided. The electronic product test control system includes: a test environment construction module 1, an automated test execution module 2, a dynamic environment test module 3, a data analysis module 4, and an anomaly detection and response module 5.

[0174] The test environment construction module 1, automated test execution module 2, dynamic environment test module 3, data analysis module 4, and anomaly detection and response module 5 are connected in sequence.

[0175] Test environment construction module 1 is used to set test conditions according to the expected use environment of electronic products, build a modular test environment chamber, and set and adjust specific environmental parameters for each test module in the test environment chamber;

[0176] The automated test execution module 2 is used to automatically execute working status and performance tests in a modular test environment using pre-configured test scripts, and optimize test sequences and resource configurations.

[0177] Dynamic environment testing module 3 is used for automatically executed working status and performance testing. Based on the characteristics of the modular test environment chamber, it applies changing environmental conditions to each test module and performs multiple tests.

[0178] Data Analysis Module 4 is used to collect test data on performance, durability, and other key indicators in multiple tests of electronic products; configure Spark cluster resources according to the scale and complexity of the collected test data; use the ReliefF algorithm to evaluate the importance of the collected test data and create multiple targeted feature subspaces based on the evaluation results; train multiple classification and regression decision trees in multiple feature subspaces using bootstrap sampling methods; integrate all classification and regression decision trees to obtain the final random forest model; use the final random forest model to analyze the test data, predict potential design defects, identify design defects related to defects, and optimize for the identified design defects;

[0179] The anomaly detection and response module 5 is used to implement a real-time anomaly detection mechanism, continuously monitor abnormal conditions during the testing process, and immediately adjust the test parameters when an abnormal condition occurs.

[0180] In summary, by utilizing the above-described technical solution of this invention, the present invention ensures the repeatability and consistency of testing by precisely setting and adjusting environmental parameters through an automated control system, which helps to accurately evaluate the performance and stability of products under different environments; and by utilizing rough sets... QLearning algorithms optimize the workload and energy consumption of testing equipment, effectively improving energy efficiency and equipment utilization while reducing energy waste. Real-time monitoring and data analysis enable timely identification and response to abnormal states and performance degradation during testing, accelerating fault diagnosis and repair. Detailed performance data and adjustment effect evaluation reports provide feedback for product design and testing processes, supporting continuous improvement and iterative development. Rapid adjustment of automated test scripts and environmental parameters enhances the automation and flexibility of the testing process, adapting to rapidly changing testing needs and complex testing scenarios. This invention can accurately simulate varying environmental conditions, such as temperature, based on the actual usage scenarios of electronic products. Humidity, vibration, and other factors enhance the practicality and reliability of test results. Simultaneously, applying varying environmental conditions to each test module and executing multiple tests allows for a comprehensive evaluation of the product's performance under different environmental conditions, enhancing its adaptability and durability. By collecting key data and utilizing distributed reinforcement learning algorithms, not only can design defects be identified, but these defects can also be targeted for optimization, promoting continuous improvement in product design. Implementing a real-time anomaly detection mechanism enables timely identification and adjustment of abnormal situations during testing, ensuring efficient test operation and accurate test data. Optimizing test sequences and resource allocation reduces resource waste and improves the cost-effectiveness of the testing process.

[0181] Although the present invention has been disclosed above with reference to preferred embodiments, the embodiments are merely examples for illustrative purposes and are not intended to limit the present invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention should be determined by the claims.

Claims

1. A test control method for electronic products, characterized in that, The electronic product testing and control method includes the following steps: S1. Set test conditions according to the expected use environment of electronic products, construct a modular test environment chamber, and set and adjust specific environmental parameters for each test module in the test environment chamber; S2. In the modular test environment, pre-configured test scripts are used to automatically execute working status and performance tests, and optimize test sequences and resource configurations. S3. Based on the automatic execution of working status and performance tests, according to the characteristics of the modular test environment, different environmental conditions are applied to each test module at the same time, and multiple tests are performed. S4. In multiple tests of electronic products, collect test data on performance, durability, and other key indicators; configure Spark cluster resources according to the scale and complexity of the collected test data; use the ReliefF algorithm to evaluate the importance of the collected test data, and create multiple targeted feature subspaces based on the evaluation results; train multiple classification and regression decision trees in multiple feature subspaces using bootstrap sampling methods; integrate all classification and regression decision trees to obtain the final random forest model; use the final random forest model to analyze the test data, predict potential design defects, identify design defects related to defects, and optimize for the identified design defects. S5. Implement a real-time anomaly detection mechanism to continuously monitor abnormal situations during the testing process and adjust test parameters immediately when abnormal situations occur. The process of automatically executing operational status and performance tests using pre-configured test scripts in a modular testing environment, and optimizing test sequences and resource configurations, includes the following steps: S21. In the modular test environment chamber, set the required environmental parameters to predefined standard values ​​according to the test requirements; S22. Run the pre-configured test script and monitor the script's execution status and performance metrics; S23, Using rough sets Q The learning algorithm adjusts the workload and energy consumption of the test equipment, optimizing the utilization of the test equipment and the power supply. S24. Conduct preliminary analysis and verification of the test results, collect feedback data, and ensure that each test achieves the preset objectives; The use of rough sets Q Learning algorithms adjust the workload and energy consumption of test equipment, optimizing equipment utilization and power supply, including the following steps: S231. Determine the initial state and action set of the test equipment, and initialize... Q The table and parameters define the state space and action set, and set the initial parameters. Q Values, learning rate, and discount factor; S232. Monitor and record the current workload and energy consumption status of the test equipment, and collect data as status input; S233, According to the current Q A function that selects an action from the action set, executes the selected action, observes the reward obtained, and the next state; S234, According to rough set Q Learning update rules adjustment Q value; S235, according to the adjusted Q Adjust the workload and power settings of the test equipment, monitor the performance and energy consumption of the test equipment after adjustment, and evaluate the optimization effect; S236. Repeat steps S232-S235 until the predetermined optimization goal is achieved; According to rough set Q Learning update rules adjustment Q The expression for the value is: ； In the formula, Indicates that it is in the execution state. e Take action below d Expected returns; Indicates the learning rate; r Indicates the execution of an action e In state d The reward received directly; Indicates the discount factor; n Indicates the number of steps; Indicates the next execution state e All actions in ' of Q Maximum value; v ′ represents the average inclusion value calculated based on rough set theory; e Indicates the execution of an action; e ′ indicates the next execution state; d Indicates an action; It represents all actions.

2. The test control method for electronic products according to claim 1, characterized in that, The adjusted Q Adjusting the workload and power settings of the test equipment, monitoring the performance and energy consumption of the test equipment after adjustment, and evaluating the optimization effect include the following steps: S2351, from Q The table identifies the optimal action for each state, and the operating parameters of the test equipment are adjusted accordingly to match the recommended optimal action; S2352. Implement parameter adjustments for the test equipment and simultaneously initiate real-time monitoring of the workload, energy consumption, and performance indicators of the adjusted test equipment. S2353. Collect the adjusted performance data, analyze the performance data, and determine the actual impact of the adjustment on the performance of the test equipment. S2354. Compare performance data before and after the adjustment to evaluate improvements in energy efficiency and workload management.

3. The test control method for electronic products according to claim 2, characterized in that, The process of collecting and analyzing the adjusted performance data to determine the actual impact of the adjustments on the performance of the test equipment includes the following steps: S23531. Determine key performance indicators and set the frequency and duration of data collection; S23532. Real-time collection of performance data during equipment operation; S23533. Analyze performance data using time series analysis and configure data charts; S23534. Based on the analysis results of performance data, evaluate the effectiveness of each adjustment measure, and compile the analysis results and data charts into a detailed report.

4. The test control method for electronic products according to claim 1, characterized in that, The automated execution-based working state and performance testing, based on the characteristics of the modular test environment, simultaneously applies varying environmental conditions to each test module and performs multiple tests, including the following steps: S31. Based on the actual usage scenarios of electronic products, determine the simulated environmental conditions and set performance evaluation standards; S32. Adjust the environmental conditions of the test room according to the requirements of each test module; S33. Start the test program for the electronic product in the modular test environment chamber and record the running data of all modules; S34. During the test, change the environmental conditions according to the predetermined plan, and monitor the impact of the environmental changes on the performance of each module; at the same time, conduct multiple performance tests to comprehensively evaluate the product performance. S35. Collect performance data of all test modules and evaluate the performance of each module under various environmental conditions through statistical and comparative analysis.

5. The test control method for electronic products according to claim 1, characterized in that, The process of using the ReliefF algorithm to assess the importance of collected test data and creating multiple targeted feature subspaces based on the assessment results includes the following steps: Extract feature data from the collected test data; A preset number of instances are randomly selected from the test data, and for each selected instance, similar nearest neighbors and dissimilar nearest neighbors are found; Update the weights of each feature data based on the feature differences between the target instance and its similar and dissimilar neighbors; By repeatedly executing steps S433-S434, multiple instances are evaluated, and the average weight of each feature throughout the process is calculated as the importance score. Based on importance scores, all features are sorted, and multiple feature subspaces are created by selecting the features with the highest importance.

6. The test control method for electronic products according to claim 5, characterized in that, The process of training multiple classification and regression decision trees using bootstrap sampling in multiple feature subspaces and integrating all the classification and regression decision trees to obtain the final random forest model includes the following steps: Select an appropriate training dataset from the multiple feature subspaces created; During the training of each decision tree, a bootstrap sampling method is used to randomly draw samples from the training dataset with replacement. The dataset obtained through bootstrap sampling is used to independently train each classification and regression decision tree, allowing each tree to learn the different features of the feature data. All trained classification and regression decision trees are merged into a random forest model; Based on the training results and preliminary validation feedback, the parameters of the random forest model were adjusted.

7. A test control system for electronic products, used to implement the test control method for electronic products according to any one of claims 1-6, characterized in that, The electronic product test and control system includes: a test environment construction module, an automated test execution module, a dynamic environment test module, a data analysis module, and an anomaly detection and response module; The test environment construction module, automated test execution module, dynamic environment test module, data analysis module, and anomaly detection and response module are sequentially connected. The test environment construction module is used to set test conditions according to the expected use environment of electronic products, build a modular test environment chamber, and set and adjust specific environmental parameters for each test module in the test environment chamber; The automated test execution module is used to automatically execute working status and performance tests in a modular test environment using pre-configured test scripts, and optimize test sequences and resource configurations. The dynamic environment testing module is used for automatically executed working state and performance testing. Based on the characteristics of the modular test environment, it applies changing environmental conditions to each test module and executes multiple tests. The data analysis module is used to collect test data on performance, durability, and other key indicators in multiple tests of electronic products; configure Spark cluster resources according to the scale and complexity of the collected test data; use the ReliefF algorithm to evaluate the importance of the collected test data and create multiple targeted feature subspaces based on the evaluation results; train multiple classification and regression decision trees in multiple feature subspaces using bootstrap sampling methods; integrate all classification and regression decision trees to obtain the final random forest model; use the final random forest model to analyze the test data, predict potential design defects, identify design defects related to defects, and optimize for the identified design defects; The anomaly detection and response module is used to implement a real-time anomaly detection mechanism, continuously monitor abnormal conditions during the testing process, and immediately adjust test parameters when an anomaly occurs.

Images