A performance test system for EMMC particle controller

Abstract

The present invention discloses a performance testing system for an EMMC particle controller, which relates to the technical field of embedded system performance testing and prediction. The system comprises a preprocessing module for collecting and preprocessing operating data and user historical operation record data of the EMMC particle controller; a feature extraction module for extracting features from the operating data and user historical operation record data to obtain operating feature vectors and user feature vectors; a performance prediction module for constructing an EMMC performance prediction model to obtain EMMC failure modes and performance prediction values; a user analysis module for identifying user behavior patterns based on user feature vectors; and an operation scenario module for identifying key operation scenarios based on user behavior patterns and failure patterns. The present invention constructs an EMMC performance prediction model to implement fault prediction based on multi-dimensional features, particularly for nonlinear behavior prediction under high load or abnormal operating conditions, thereby reducing false alarm rates and missed alarm rates.

Description

Technical Field

[0001] The present invention relates to the technical field of embedded system performance testing and prediction, in particular to a performance testing system for an EMMC particle controller. Background Art

[0002] EMMC (Embedded Multi Media Card), also known as an embedded multimedia controller, is a packaged storage solution that integrates a memory chip, controller, and interface. As a highly efficient storage solution, embedded multimedia controllers are widely used in smartphones, tablets, and other mobile devices. With the increasing popularity and functional complexity of mobile devices, the performance requirements for EMMC controllers are also increasing.

[0003] While existing EMMC performance testing has made some progress, it still has shortcomings. First, there is a lack of in-depth analysis of user behavior patterns. Traditional testing methods are mostly based on predefined operating scenarios and fail to fully consider users' individual needs and behavioral habits. This leads to significant discrepancies between test results and actual user experience, failing to fully reflect the performance of EMMC in real-world usage environments. Furthermore, existing models have limited fault prediction capabilities. In most cases, fault prediction models rely solely on static operational characteristics for modeling, ignoring the changing trends of operational characteristics over time and the complex relationships between them. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a performance testing system for an EMMC particle controller to solve the problem of lack of in-depth analysis of user behavior patterns.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] The present invention provides a performance testing system for an EMMC particle controller, which comprises:

[0008] Collect the operating data of the EMMC particle controller and the user's historical operation record data, and perform pre-processing;

[0009] Extract features from the operation data and user historical operation record data to obtain operation feature vectors and user feature vectors;

[0010] Construct an EMMC performance prediction model to obtain the failure mode and performance prediction value of EMMC;

[0011] Identify user behavior patterns based on user feature vectors;

[0012] Identify critical operation scenarios based on user behavior patterns and failure modes;

[0013] Develop personalized testing plans based on key operating scenarios;

[0014] Execute personalized test plans and monitor the execution process;

[0015] Generate a test report based on the execution results.

[0016] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, wherein: the operating data includes temperature data, voltage data and read / write speed data;

[0017] The user's historical operation record data includes operation frequency, operation type, operation time and operation result.

[0018] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, the preprocessing specifically includes the following steps:

[0019] All collected data are processed to remove noise points, fill missing values ​​and normalize data.

[0020] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, wherein: feature extraction is performed on the operation data and the user historical operation record data to obtain the operation feature vector and the user feature vector, specifically including the following steps:

[0021] Use fast Fourier transform to extract high-dimensional features of running data and form running feature vectors;

[0022] The number of operations using a sliding window method to capture the operation frequency;

[0023] Count the operation ratios of different operation types;

[0024] Calculate the time interval between each operation;

[0025] Count the ratio of successful and failed operations;

[0026] The number of operations, operation ratio, operation time interval and the ratio of successful and failed operations are combined to form a user feature vector.

[0027] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, wherein: constructing an EMMC performance prediction model to obtain the performance prediction value of EMMC specifically includes the following steps:

[0028] Define failure modes based on operational feature vectors;

[0029] Perform nonlinear mapping on the operating characteristic vector to obtain the operating characteristic transformation vector;

[0030] The running feature vector, weight matrix and running feature transformation vector are linearly combined to obtain the calculation result of linear prediction, and the Sigmoid function is applied to normalize the calculation result of linear prediction;

[0031] Do the dot product calculation between the running feature vector and the nonlinear weight vector, and introduce nonlinear transformation to obtain the nonlinear adjustment term;

[0032] By combining the prediction results of the linear part and the nonlinear adjustment term, the failure mode score of EMMC is obtained, which is expressed as:

[0033] ;

[0034] in, Indicates the The score of each failure mode, represents the Sigmoid function, represents the running feature vector, represents the transpose of the running eigenvector, Indicates the The weight matrix of the failure modes, represents the running feature transformation vector, Indicates the nonlinear adjustment term for the The weight of the failure mode, represents the nonlinear weight vector;

[0035] Use the Softmax function to convert the failure mode score into failure probability;

[0036] Based on the failure probability, the performance prediction value of EMMC is obtained.

[0037] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, wherein: the performance prediction value of the EMMC is obtained, specifically comprising the following steps:

[0038] Calculate the weighted average of the running eigenvectors;

[0039] The optimization algorithm is combined with the weighted average of the operating characteristic vector to calculate the failure probability and obtain the performance prediction value of the EMMC, which is expressed as follows:

[0040] ;

[0041] in, Indicates that the EMMC particle controller is in time The performance prediction value of Indicates the number of features involved in the calculation, Indicates the The normalized running characteristics, represents the weighted average of the normalized running characteristics, represents the square of the standard deviation of the normalized running characteristic, represents the total number of failure modes, Indicates the The weight of each failure mode, represents the regularization parameter, Indicates the Failure mode at time probability of failure.

[0042] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, wherein: identifying user behavior patterns and preference analysis based on user feature vectors specifically includes the following steps:

[0043] Using the clustering cost function, the initial cluster center is selected from the user feature vector, and the Euclidean distance from each data to the cluster center is calculated;

[0044] The distance from each data point to the cluster center, assigning the data point to the nearest cluster;

[0045] After completing the cluster assignment, the center of each cluster is calculated as the mean of all data points belonging to the cluster until the clustering cost function reaches the minimum value, which is expressed as:

[0046] ;

[0047] in, Indicates the The clustering cost function at the iteration, represents the total number of clusters, Indicates the The first iteration A set of data points in a cluster, represents a data point, Indicates the The feature vector of the data point, Indicates the The first iteration The center of the cluster;

[0048] When the clustering cost function reaches the minimum value, the cluster center and cluster affiliation complete convergence;

[0049] The cluster center is a characteristic representation of the user behavior pattern, and the cluster affiliation is the behavior pattern to which the user behavior characteristic belongs.

[0050] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, the key operation scenarios are identified based on user behavior patterns and failure probabilities, specifically including the following steps:

[0051] Set minimum support and minimum confidence;

[0052] Convert each user's operation records into transaction format;

[0053] The transaction format includes all operations within a period of time;

[0054] For each user behavior pattern, FP-Growth algorithm is applied to mine frequent item sets and association rules;

[0055] Use statistical analysis to find the operating scenarios that affect the performance of the EMMC granular controller from the mined association rules;

[0056] Define a weight function based on behavior patterns and failure probabilities to calculate the priority of each operation scenario;

[0057] According to the calculated priorities, all key operation scenarios are sorted and H key operation scenarios are selected as high-priority key operation scenarios.

[0058] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, a personalized test plan is formulated based on key operation scenarios, specifically including the following steps:

[0059] Build a test framework and configure fault injection tools;

[0060] Design specific test cases based on high-priority critical operation scenarios and gradually expand to other test scenarios.

[0061] As a preferred solution of the performance test system of the EMMC particle controller of the present invention, wherein: generating a test report specifically includes the following steps:

[0062] Aggregate all test logs and performance monitoring data into a central database;

[0063] Use data analysis tools to analyze test data and generate charts and statistical reports.

[0064] The beneficial effects of this invention include: using a clustering cost function to identify user behavior patterns, accurately capturing users' individual needs and behavioral habits, providing an important basis for developing personalized test plans, and improving the consistency between test results and actual user experience. The personalized test plans developed in this way ensure flexibility and adaptability, and can dynamically adjust test strategies based on different user needs and operating scenarios, improving overall performance and user experience. In addition, an EMMC performance prediction model is constructed, enabling fault prediction based on multi-dimensional features, significantly improving the accuracy of fault prediction, especially for nonlinear behavior prediction under high load or abnormal operating conditions, reducing false alarm and missed alarm rates. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0066] Figure 1 Schematic diagram of the performance test system of the EMMC particle controller in Example 1.

[0067] Figure 2 A schematic diagram of generating a personalized solution in Example 1. DETAILED DESCRIPTION

[0068] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0069] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0070] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0071] Example 1, with reference to Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a performance testing system for an EMMC particle controller, comprising the following steps:

[0072] The preprocessing module collects the operating data of the EMMC particle controller and the user's historical operation record data and performs preprocessing;

[0073] Collect temperature data of the EMMC particle controller to monitor whether the EMMC is overheating during operation and ensure that it operates within a safe temperature range.

[0074] Collect voltage data from the EMMC particle controller to evaluate the stability of the power supply and prevent performance degradation or hardware damage caused by voltage fluctuations.

[0075] Collect read and write speed data of the EMMC granular controller to measure the response time and throughput of the EMMC under different operating conditions and evaluate its performance.

[0076] Further explanation provides accurate data support to help analyze the performance of EMMC under different load conditions and provide a basis for optimized design.

[0077] Count the frequency of users performing specific operations and understand their common operating habits.

[0078] Record the types of operations performed by users, such as file transfers and application launches, to identify high-frequency key operations.

[0079] Record the timestamp of each operation and analyze user behavior patterns and operation time periods.

[0080] Record the success or failure of each operation and evaluate the reliability and success rate of the operation.

[0081] It is further explained that through detailed user operation records, the usage habits and preferences of different users can be identified, providing data support for the formulation of personalized testing plans.

[0082] All data are stored in a central database to facilitate subsequent data processing and analysis.

[0083] The Z-score method is used to identify and eliminate abnormal data points to ensure the authenticity and reliability of the data.

[0084] Interpolation method is used to fill in the data with missing values ​​to ensure the integrity of the data.

[0085] Min-Max normalization is used to convert data of different scales to the same scale, eliminating dimensional differences and facilitating subsequent feature extraction and model construction.

[0086] It is further explained that through detailed data collection and preprocessing steps, not only the comprehensive monitoring and capture of the EMMC particle controller's operating status and user behavior patterns are achieved, but also the consistency, integrity and reliability of the data are ensured through centralized management and preprocessing.

[0087] The feature extraction module extracts features from the operation data and user historical operation record data to obtain the operation feature vector and user feature vector;

[0088] Use Fast Fourier Transformation to transform time series operational data (such as temperature, voltage, read / write speed, etc.) into the frequency domain, capturing both periodic and non-periodic components. This allows identification of high- and low-frequency features hidden in the time series data, which correspond to different types of failure modes or performance bottlenecks.

[0089] It is further shown that frequency domain features can capture complex patterns in time series data that are difficult to find intuitively, thereby improving the model's ability to identify different types of faults.

[0090] The sliding window method can be used to dynamically monitor the user's operation frequency in different time periods and capture short-term and long-term change trends.

[0091] Categorize historical user operations into different types (such as file transfers, application launches, read and write operations), and calculate the percentage of each type. Based on the percentage of operations, assess which operations have the greatest impact on the performance of the EMMC granular controller and determine the focus of testing.

[0092] By calculating the time interval between each operation, we can analyze the time distribution of user operations and identify user behavior patterns.

[0093] Count the success rate of each operation, understand which operations are most likely to fail, and identify potential problem points.

[0094] The features of multiple dimensions, such as the number of operations, operation ratio, time interval and success rate, are combined to form a comprehensive user feature representation.

[0095] It is further explained that through the fusion of multi-dimensional features, a detailed user portrait is formed, which provides a scientific basis for personalized testing.

[0096] The performance prediction module builds the EMMC performance prediction model to obtain the failure mode and performance prediction value of EMMC. It specifically includes the following steps:

[0097] The failure mode is defined according to the running feature vector, the possible failure types are classified, and the corresponding feature representation is defined for them, ensuring that all possible failure modes are taken into account, thereby improving the generalization ability of the model.

[0098] Through nonlinear mapping, the complex nonlinear relationship between operating features is captured and combined into an operating feature transformation vector.

[0099] It further shows that nonlinear mapping can better capture dynamic changes and complex behaviors.

[0100] A weight matrix is ​​set based on the operating feature vector and the operating feature transformation vector to measure the importance of different operating features to a specific failure mode.

[0101] According to the running feature vector, the running feature transformation vector and the weight matrix, the linear prediction calculation result is obtained through linear combination, and the Sigmoid function is used to normalize the linear prediction result to the range of [0,1] to facilitate subsequent probability conversion.

[0102] Establish the initial nonlinear weight vector and define the loss function;

[0103] The nonlinear weight vector is obtained by minimizing the loss function through the convergence of the optimization algorithm;

[0104] Through dot product calculation, the relationship between the running feature vector and the nonlinear weight vector is quantified to obtain the nonlinear adjustment term;

[0105] It is further shown that nonlinear adjustment terms help capture subtle changes and differences, improving the model's ability to understand complex scenarios.

[0106] By combining the prediction results of the linear part and the nonlinear adjustment term, the failure mode score of EMMC is obtained, which is expressed as:

[0107] ;

[0108] in, Indicates the The score of each failure mode, represents the Sigmoid function, represents the running feature vector, represents the transpose of the running eigenvector, Indicates the The weight matrix of the failure modes, represents the running feature transformation vector, Indicates the nonlinear adjustment term for the The weight of the failure mode, represents the nonlinear weight vector.

[0109] Failure mode score The value range depends on ,when greater than zero, The range of is [0, +∞]; when is equal to zero, The range of is [0,1]. less than zero, The value range of is [-∞,1], and it is adjusted according to the needs of specific business scenarios. size.

[0110] The Softmax function is used to convert the failure mode score into the failure probability, which is expressed as follows:

[0111] ;

[0112] in, Indicates the Failure modes in a given operating characteristic vector The probability under the condition of , N represents the total number of failure modes.

[0113] For the convenience of calculation and representation, Convert to .

[0114] Each running feature is normalized and the weighted average of the running features is calculated. The expression is:

[0115] ;

[0116] in, The weighted average of the running features after normalization, represents the number of features, Indicates the The weight of the normalized running features, Indicates the Normalized running characteristics.

[0117] The optimization algorithm is combined with the weighted average of the operating characteristic vector to calculate the failure probability and obtain the performance prediction value of the EMMC, which is expressed as follows:

[0118] ;

[0119] in, Indicates that the EMMC particle controller is in time The performance prediction value of Indicates the number of features involved in the calculation, Indicates the The normalized running characteristics, represents the weighted average of the normalized running characteristics, represents the square of the standard deviation of the normalized running characteristic, represents the total number of failure modes, Indicates the The weight of each failure mode, represents the regularization parameter, Indicates the Failure mode at time probability of failure.

[0120] This performance prediction value reflects the EMMC particle controller's overall health and potential risks.

[0121] A health threshold is set up, and the health status of the EMMC granular controller is obtained based on the performance prediction value within the range of the health threshold. The setting of the health threshold is customized according to the usage scenario and range.

[0122] The user analysis module identifies user behavior patterns based on user feature vectors.

[0123] Randomly select the initial cluster center from the user feature vector;

[0124] For each data point, calculate the Euclidean distance to all initial cluster centers and assign each data point to the cluster with the closest distance based on the Euclidean distance.

[0125] Further explanation: By minimizing the distance from each data point to the nearest cluster center, data points are ensured to be reasonably assigned to the most similar clusters, thereby improving the accuracy of clustering.

[0126] For each cluster, calculate its new center as the mean of all data points belonging to that cluster.

[0127] It is further explained that by continuously updating the cluster centers, the clustering results are gradually optimized so that the cluster centers more accurately reflect the characteristics of the data points in the cluster.

[0128] Define the clustering cost function to measure the quality of the current clustering result, and its expression is:

[0129] ;

[0130] in, Indicates the The clustering cost function at the iteration, represents the total number of clusters, Indicates the The first iteration A set of data points in a cluster, represents a data point, Indicates the The feature vector of the data point, Indicates the The first iteration The center of a cluster.

[0131] When the clustering cost function reaches the minimum value, it is considered that the cluster center and cluster affiliation have completed convergence.

[0132] The cluster center represents the characteristic representation of the user behavior pattern, while the cluster affiliation reflects the behavior pattern to which the user behavior feature belongs.

[0133] It is further explained that by minimizing the clustering cost function, the final clustering result is ensured to reach the optimal state, thereby improving stability and reliability.

[0134] The operation scenario module identifies key operation scenarios based on user behavior patterns and failure modes.

[0135] Set the minimum support and support threshold for mining frequent item sets and association rules to ensure that the mined rules are statistically significant.

[0136] Specifically, by setting the minimum support, we can exclude those item sets or rules that appear very rarely in the data set, ensuring that the mined patterns have statistical significance and practical application value.

[0137] The purpose of setting the minimum support threshold is to filter out unimportant or uncommon patterns while controlling computational complexity. A lower support threshold usually mines more frequent itemsets, but they may contain many meaningless patterns and generate more rules with lower quality.

[0138] Therefore, by comparing the effects of thresholds, we can find a balance that ensures that sufficient and meaningful patterns are mined while controlling computational complexity. Furthermore, an appropriate threshold can significantly reduce the number of candidate item sets, improve algorithm execution efficiency, and reduce computational costs.

[0139] Convert each user's operation records into a transaction format, where each transaction contains all operations within a period of time. Transaction formatting helps better organize and process operation data, facilitating subsequent association rule mining.

[0140] Based on each user's behavior pattern, the FP-Growth algorithm is applied to mine frequent itemsets and association rules. FP-Growth is an efficient frequent pattern mining algorithm that can quickly find frequent itemsets in large-scale datasets.

[0141] Using statistical analysis methods, we can identify operational scenarios that affect the performance of the EMMC granular controller from the mined association rules. For example, certain operational combinations may lead to a decrease in EMMC performance or an increase in the frequency of failures.

[0142] Calculate the frequency weight based on the number of data points in the user behavior pattern, input the feature representation of each user behavior pattern into the performance prediction model, and calculate the corresponding failure probability;

[0143] Multiply the occurrence frequency of each behavior pattern and the corresponding failure probability to obtain the impact weight of all behavior patterns;

[0144] The influence weights of the obtained behavior patterns are sorted, and the characteristic representation of the behavior patterns is output as the key operation scenarios.

[0145] For example, high-frequency write operations and long-term high-load random reads.

[0146] Select H key operation scenarios from the key operation scenarios as high-priority key operation scenarios. The number of high-priority key operation scenarios is customized according to actual conditions.

[0147] The test plan module develops personalized test plans based on key operation scenarios.

[0148] Establish a test environment, including hardware platform, software environment, monitoring tools, and logging;

[0149] Establish a test framework that supports batch execution of test cases to improve test efficiency. Ensure that the test framework is highly configurable, allowing users to adjust test parameters (such as load intensity, operation frequency, etc.) as needed.

[0150] Choose appropriate fault injection tools, such as ChaosMonkey and Gremlin, which provide a rich set of fault types (such as network latency, disk IO errors, memory leaks, etc.).

[0151] Configure the fault injection strategy, including fault type, fault frequency, fault scope, and recovery mechanism.

[0152] Design specific test cases based on high-priority critical operational scenarios, such as:

[0153] High-frequency write operations: simulate frequent write operations to evaluate their impact on EMMC performance.

[0154] Long-term high-load random read: Simulates long-term high-load random read operations to evaluate their impact on EMMC stability and response time.

[0155] Combined operations: Simulates a combination of multiple operations (such as read-after-write, mixed read-write, etc.) to evaluate their impact on EMMC resource allocation and performance.

[0156] Extended test scenarios, such as boundary conditions, multi-user concurrency, and long runs.

[0157] It further explains that resources can be concentrated on the most critical scenarios to improve testing and maintenance efficiency and reduce unnecessary costs.

[0158] Monitoring module, executes personalized test plans, and monitors the execution process;

[0159] Execute test cases designed based on high-priority key operation scenarios and test cases with extended test scenarios, and monitor performance indicators in real time during execution.

[0160] Based on the test results, identify the performance bottlenecks of EMMC, determine which operation scenarios have the greatest impact on performance, summarize common failure modes and their triggering conditions, and provide improvement suggestions.

[0161] Based on the test results, the fault injection tool is updated to make it closer to real application scenarios and improve the accuracy and effectiveness of the test.

[0162] For example, new fault types are added and fault injection strategies are optimized.

[0163] The test report module generates a test report based on the execution results.

[0164] Collect detailed log files from each test node, including operation logs and error logs;

[0165] Collect performance indicator data from monitoring tools, such as response time, throughput, CPU usage, memory usage, etc.

[0166] Use secure transmission protocols to send test logs and performance monitoring data to a central database.

[0167] Use visualization tools to draw trend charts of performance indicators, analyze performance changes under different operating scenarios, identify performance bottlenecks, and determine which operating scenarios have the greatest impact on performance.

[0168] Generate comprehensive test reports including failure frequency, impact scope, and recovery time.

[0169] Further explanation: The results of this performance test can be intuitively understood, and the test experience can be summarized and the next test can be conducted based on the test results.

[0170] In summary, the present invention uses a clustering cost function to identify user behavior patterns, accurately capturing users' individual needs and behavioral habits. This provides an important basis for developing personalized test plans, improving the consistency between test results and actual user experience. The resulting personalized test plan ensures flexibility and adaptability, dynamically adjusting test strategies based on the needs of different users and operating scenarios, and improving overall performance and user experience. Furthermore, an EMMC performance prediction model is constructed, enabling fault prediction based on multidimensional features. This significantly improves the accuracy of fault prediction, particularly for nonlinear behavior prediction under high load or abnormal operating conditions, reducing false positive and false negative rates.

[0171] Example 2, referring to Table 1, is the second embodiment of the present invention. In order to further verify the technical solution of the present invention, experimental simulation data of the performance test system of the EMMC particle controller is provided.

[0172] To verify the effectiveness of the proposed EMMC granular controller performance testing system, a series of experiments were designed and conducted. The experimental subject was a certain model of EMMC granular controller (designated Model A), which is widely used for data storage in mobile devices. The experimental environment was set up under standard laboratory conditions to minimize the impact of external factors such as temperature and humidity on the experimental results.

[0173] Data collection and preprocessing, operational data: Built-in sensors continuously record the Model A's temperature, voltage, and read / write speed data under different workloads. This data is recorded every 1 second and transmitted to a central database.

[0174] User historical operation record data: Data on user operation frequency, type, time, and results are collected from multiple usage scenarios, covering both normal and abnormal operations.

[0175] Preprocessing: All raw data are processed by removing noise points, filling missing values ​​and normalization to ensure the consistency and accuracy of the data, providing a reliable basis for subsequent analysis.

[0176] Feature extraction, running feature vector: The fast Fourier transform (FFT) method is used to extract high-dimensional features from the running data, forming a running feature vector containing frequency domain information.

[0177] User feature vector: The sliding window method is used to capture the number of operations of the operation frequency, and the proportion, time interval and success and failure ratio of different types of operations are counted to construct the user feature vector, which reflects the user's operation habits and preferences.

[0178] Performance prediction model construction and failure mode definition: Based on the operating feature vector, multiple possible failure modes are defined, such as overheating protection activation and data loss caused by voltage fluctuations.

[0179] Nonlinear mapping and combination: The operating feature transformation vector is obtained through nonlinear mapping, and it is linearly combined with the original operating feature vector. Finally, the Sigmoid function is applied to normalize the result to generate the EMMC performance prediction value.

[0180] Fault probability calculation: The Softmax function is used to convert each fault mode score into a fault probability, and a reasonable fault threshold is set to determine whether a specific fault occurs.

[0181] User behavior pattern recognition and cluster analysis: The K-means algorithm is used to cluster user feature vectors, and several user behavior pattern clusters are obtained. Each cluster represents a typical operating style or preference.

[0182] Identification of key operation scenarios and association rule mining: Based on user behavior patterns, the FP-Growth algorithm is applied to mine frequent item sets and association rules, from which key operation scenarios that affect EMMC performance are screened.

[0183] Prioritization: Define a weight function, calculate the priority of each operation scenario, and select the top H as high-priority key operation scenarios.

[0184] Test plan development and execution, personalized test case design: Detailed test cases are developed for high-priority critical operation scenarios, and gradually expanded to other scenarios to ensure comprehensive coverage.

[0185] Monitoring and logging: During the execution of personalized test plans, various performance indicators are monitored in real time, and all test logs and performance monitoring data are aggregated into a central database.

[0186] Test report generation, data analysis and visualization: Data analysis tools were used to conduct in-depth analysis of the collected data, generating charts and statistical reports that visually demonstrated the performance of the EMMC granular controller in different operating scenarios. The details are shown in Table 1 below:

[0187] Table 1 EMMC operation comparison table

[0188]

[0189] By comparing and analyzing the contents of the above table, it can be clearly seen that the EMMC particle controller performance test proposed in the present invention has significant advantages over the comparative technology:

[0190] Temperature control is optimized. Under the same workload, the average temperature of the present invention is reduced by 5.7°C (from 45.2°C to 39.5°C). This shows that the present invention can manage heat more effectively, reduce the risk of overheating, and extend the life of the equipment.

[0191] The voltage stability is improved, and the average voltage increases slightly from 3.25V to 3.30V, which shows that the present invention can slightly improve the power supply efficiency and reduce the possibility of voltage fluctuation while ensuring stable power supply.

[0192] The read and write speeds were significantly improved, with the average read and write speed increased by 6.5MB / s (from 28.7MB / s to 35.2MB / s), demonstrating the improvement of the present invention in optimizing the read and write paths and greatly improving data processing efficiency.

[0193] The failure rate was significantly reduced, from 12.3% to 5.4%, nearly halving the number of failures. This not only demonstrates the effectiveness of this invention in fault prediction and prevention, but also directly improves user experience and reliability.

[0194] The operation success rate was significantly improved from 87.7% to 94.6%, indicating that the present invention can better adapt to different user behavior patterns and provide more stable service quality.

[0195] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A performance test system for an EMMC particle controller, characterized in that: include: The preprocessing module collects the operating data of the EMMC particle controller and the user's historical operation record data and performs preprocessing; The feature extraction module extracts features from the operation data and user historical operation record data to obtain the operation feature vector and user feature vector; Performance prediction module, builds EMMC performance prediction model, obtains EMMC failure mode and performance prediction value; User analysis module, which identifies user behavior patterns based on user feature vectors; Operation scenario module, which identifies key operation scenarios based on user behavior patterns and failure modes; Test plan module, which develops personalized test plans based on key operation scenarios; Monitoring module, executes personalized test plans, and monitors the execution process; Test report module, generates test reports based on execution results; Constructing an EMMC performance prediction model to obtain the performance prediction value of EMMC includes the following steps: Define failure modes based on operational feature vectors; Perform nonlinear mapping on the operating characteristic vector to obtain the operating characteristic transformation vector; The running feature vector, weight matrix and running feature transformation vector are linearly combined to obtain the calculation result of linear prediction, and the Sigmoid function is applied to normalize the calculation result of linear prediction; Do the dot product calculation between the running feature vector and the nonlinear weight vector, and introduce nonlinear transformation to obtain the nonlinear adjustment term; By combining the prediction results of the linear part and the nonlinear adjustment term, the failure mode score of EMMC is obtained, which is expressed as: ； in, Indicates the The score of each failure mode, represents the Sigmoid function, represents the running feature vector, represents the transpose of the running eigenvector, Indicates the The weight matrix of the failure modes, represents the running feature transformation vector, Indicates the nonlinear adjustment term for the The weight of the failure mode, represents the nonlinear weight vector; Use the Softmax function to convert the failure mode score into failure probability; Based on the failure probability, the performance prediction value of EMMC is obtained; Identify key operational scenarios based on user behavior patterns and failure patterns, including the following steps: Set the minimum support and support threshold for mining frequent itemsets and association rules; Convert each user's operation records into transaction format; The transaction format includes all operations within a period of time; For each user behavior pattern, FP-Growth algorithm is applied to mine frequent item sets and association rules; Use statistical analysis to find the operating scenarios that affect the performance of the EMMC granular controller from the mined association rules; Define a weight function based on behavior patterns and failure probabilities to calculate the priority of each operation scenario; According to the calculated priorities, all key operation scenarios are sorted and H key operation scenarios are selected as high-priority key operation scenarios.

2. The performance testing system of the EMMC particle controller according to claim 1, characterized in that: The operating data includes temperature data, voltage data and read / write speed data; The user's historical operation record data includes operation frequency, operation type, operation time and operation result.

3. The performance test system of the EMMC particle controller according to claim 2, characterized in that: The pre-processing specifically comprises the following steps: All collected data are processed to remove noise points, fill missing values ​​and normalize data.

4. The performance test system of the EMMC particle controller according to claim 3, characterized in that: Feature extraction is performed on the operation data and the user's historical operation record data to obtain the operation feature vector and the user feature vector. The specific steps include the following: Use fast Fourier transform to extract high-dimensional features of running data and form running feature vectors; The number of operations using a sliding window method to capture the operation frequency; Count the operation ratios of different operation types; Calculate the time interval between each operation; Count the ratio of successful and failed operations; The number of operations, operation ratio, operation time interval and the ratio of successful and failed operations are combined to form a user feature vector.

5. The performance test system of the EMMC particle controller according to claim 4, characterized in that: The step of obtaining the performance prediction value of the EMMC specifically comprises the following steps: Calculate the weighted average of the running eigenvectors; The optimization algorithm is combined with the weighted average of the operating characteristic vector to calculate the failure probability and obtain the performance prediction value of the EMMC, which is expressed as follows: ； in, Indicates that the EMMC particle controller is in time The performance prediction value of Indicates the number of running features involved in the calculation, Indicates the The normalized running characteristics, represents the weighted average of the normalized running characteristics, represents the square of the standard deviation of the normalized running characteristic, represents the total number of failure modes, Indicates the The weight of the failure mode, represents the regularization parameter, Indicates the Failure mode in time probability of failure.

6. The performance test system of the EMMC particle controller according to claim 5, characterized in that: Identifying user behavior patterns based on user feature vectors includes the following steps: Using the clustering cost function, the initial cluster center is selected from the user feature vector, and the Euclidean distance from each data to the cluster center is calculated; Assign data points to the nearest cluster based on their distance from the cluster center. After completing the cluster assignment, the center of each cluster is calculated as the mean of all data points belonging to the cluster until the clustering cost function reaches the minimum value, which is expressed as: ； in, Indicates the The clustering cost function at the iteration, represents the total number of clusters, Indicates the The first iteration A set of data points in a cluster, represents a data point, Indicates the The feature vector of the data point, Indicates the The first iteration The center of the cluster; When the clustering cost function reaches the minimum value, the cluster center and cluster affiliation complete convergence; The cluster center is a characteristic representation of the user behavior pattern, and the cluster affiliation is the behavior pattern to which the user behavior characteristic belongs.

7. The performance test system of the EMMC particle controller according to claim 6, characterized in that: Develop a personalized test plan based on key operating scenarios, including the following steps: Build a test framework and configure fault injection tools; Design specific test cases based on high-priority critical operation scenarios and gradually expand to other test scenarios.

8. The performance testing system of the EMMC particle controller according to claim 7, characterized in that: Generating a test report specifically includes the following steps: Aggregate all test logs and performance monitoring data into a central database; Use data analysis tools to analyze test data and generate charts and statistical reports.

Images