Performance test analysis positioning method and system based on artificial intelligence

Abstract

The invention relates to a performance test analysis positioning method based on artificial intelligence, and belongs to the technical field of software performance test problem positioning. The method comprises the steps that historical performance test multi-source data is acquired, a preset performance problem library is constructed through performance data matching, a performance problem analysis initial model is obtained through fine adjustment of a pre-trained Bert large model according to the preset performance problem library, pruning is conducted through an L1 norm pruning device, fine adjustment is conducted again, and a performance problem analysis model is obtained; obtaining performance test information and outputting a performance problem category through the performance problem analysis model; according to the performance test information and the performance problem category, performance problem positioning information is obtained through a pattern matching retrieval algorithm, the recognition precision and positioning efficiency of the performance test problem are remarkably improved, and the problem that performance problem analysis excessively depends on expert experience is effectively solved. And efficient and reliable performance test problem automatic analysis is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of software performance test problem positioning, and in particular relates to a performance test analysis and positioning method based on artificial intelligence. Background Art

[0002] Software performance testing is a key step in ensuring that the software system operates properly under predetermined performance standards. It involves evaluating the system's responsiveness, stability, resource utilization, and scalability under various workloads. When performance testing finds problems, in-depth analysis of these problems is essential to ensure that the software meets user needs and expectations. Traditional performance testing analysis techniques usually include baseline comparison, trend analysis, bottleneck analysis, log analysis, load analysis, response time analysis, and code review.

[0003] However, traditional performance testing analysis techniques often rely too much on the personal experience of performance testing experts, which may lead to inconsistency and lack of objectivity in the analysis results. In addition, although there are a variety of performance testing and analysis tools on the market, these tools may not comprehensively cover all types of performance issues or may provide inaccurate analysis results in specific situations. Summary of the invention

[0004] In order to solve the above-mentioned problems existing in the prior art, the present invention proposes a performance test analysis and positioning method based on artificial intelligence. The method implements an intelligent analysis algorithm based on a large model and a performance problem library, which can automatically analyze performance test problems, thereby replacing traditional manual analysis methods, promoting the automation of performance testing processes, and effectively solving the problem of excessive reliance on expert experience in performance problem analysis, significantly improving the recognition accuracy and positioning efficiency of performance test problems, and effectively improving the development efficiency and product quality of software products.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A performance test analysis and positioning method based on artificial intelligence, comprising:

[0007] Acquire historical performance test multi-source data, and build a preset performance problem library through performance data matching based on the historical performance test multi-source data, wherein the historical performance test multi-source data includes historical performance test information, historical performance analysis information, and historical performance problem solutions;

[0008] According to the preset performance problem library, a performance problem analysis model is obtained by secondary fine-tuning of the pre-trained Bert large model;

[0009] Acquire performance test information, the performance test information including performance test data, error log data, and configuration data, and output performance problem categories through the performance problem analysis model according to the performance test information, the performance problem categories including code problem categories, configuration problem categories, and system problem categories;

[0010] Performance problem location information is obtained according to the performance test information and the performance problem category through a pattern matching retrieval algorithm.

[0011] Preferably, the step of building a preset performance problem library by matching performance data based on the historical performance test multi-source data includes:

[0012] Extracting historical performance key indicator data through the LoadRunner tool based on historical performance test information, wherein the historical performance key indicator data includes historical performance test data, historical error log data, historical configuration data, and historical application code data;

[0013] According to the historical key performance indicator data, noise and abnormal values ​​are removed by a data cleaning tool to obtain standard data of the historical key performance indicators;

[0014] A preset performance problem library is obtained through annotation matching according to the historical performance key indicator standard data, the historical performance analysis information, and the historical performance problem solutions.

[0015] Preferably, obtaining a preset performance problem library through annotation matching based on the historical performance key indicator standard data, the historical performance analysis data, and the historical performance problem solutions includes:

[0016] According to the historical performance key indicator standard data, the historical performance analysis data, and the historical performance problem solution, performance matching information is obtained by annotating and matching with a data annotation tool;

[0017] The preset performance problem library is obtained by storing the performance matching information in a JSON file, and the annotation matching includes code problem annotation matching, configuration problem annotation matching, and system problem annotation matching.

[0018] Preferably, the code problem annotation matching includes memory leak problems, deadlock problems, circular reference problems, high algorithm complexity problems, and unreasonable data structure problems;

[0019] The configuration problem annotation matching includes improper system parameter settings, database connection pool configuration errors, unreasonable thread pool configuration problems, and middleware configuration errors;

[0020] The system problem annotation matching includes hardware failure problems, network delay problems, and system resource shortage problems.

[0021] Preferably, the performance problem analysis model is obtained by secondary fine-tuning of the pre-trained Bert large model according to the preset performance problem library, including:

[0022] According to the preset performance problem library, a preset performance problem data set is obtained by performing word segmentation and encoding through a Tokenization API, wherein the Tokenization API is a Tokenization API provided by the pre-trained Bert large model;

[0023] According to the preset performance problem data set, the Bert performance test large model is obtained by fine-tuning the training parameters of the pre-trained Bert large model, wherein the training parameters of the pre-trained Bert large model include a learning rate, a batch size, and a number of iterations;

[0024] According to the Bert performance test large model, a Bert performance test large model evaluation result is obtained by evaluating the Bert performance test large model through the Bert performance test large model evaluation index; according to the Bert performance test large model evaluation result, the model parameters are updated through the Adam optimizer to obtain the performance problem analysis initial model; the Bert performance test large model evaluation index includes a cross entropy loss function and a Bert performance test large model evaluation index; the Bert performance test large model evaluation index includes a Bert performance test large model accuracy, a Bert performance test large model recall rate, and a Bert performance test large model F1 value;

[0025] The performance problem analysis initial model is pruned by an L1 norm pruner and fine-tuned again to obtain a performance problem analysis model.

[0026] Preferably, obtaining the performance problem location information through a pattern matching retrieval algorithm according to the performance test information and the performance problem category includes:

[0027] Constructing a preset performance problem vector library through a vector model according to the preset performance problem library;

[0028] Performance problem location information is obtained through vector retrieval according to the performance test information and the performance problem category.

[0029] Preferably, constructing a preset performance problem vector library through a vector model according to the preset performance problem library includes:

[0030] Extracting performance problem document text information through the BeautifulSoup library according to the preset performance problem library;

[0031] Obtaining a performance problem text block by text segmentation according to the text information of the performance problem document;

[0032] The preset performance problem vector library is obtained by converting the performance problem text block into a vector form and creating a text block vector index.

[0033] An artificial intelligence-based performance test analysis and positioning system using the artificial intelligence-based performance test analysis and positioning method, comprising a performance test data collection module, a performance problem library construction module, a performance problem analysis module, and a performance problem positioning module;

[0034] The performance test data acquisition module is used to collect performance test information and perform data preprocessing;

[0035] The performance problem library building module is used to build a preset performance problem library based on historical performance test information, historical performance analysis information, and historical performance problem solutions;

[0036] The performance problem analysis module is used to receive performance test information and output performance problem categories through a performance problem analysis model;

[0037] The performance problem locating module is used to determine the performance problem locating information according to the performance test information and the performance problem category through a pattern matching retrieval algorithm.

[0038] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned artificial intelligence-based performance testing, analysis and positioning method when executing the program.

[0039] A storage medium containing computer executable instructions, characterized in that the computer executable instructions are used to execute the above-mentioned performance test analysis and positioning method based on artificial intelligence when executed by a computer processor.

[0040] The beneficial effects of the present invention are:

[0041] (1) The initial model for performance problem analysis is obtained by fine-tuning the pre-trained Bert model. The pre-trained Bert model already has good language comprehension capabilities. After fine-tuning the pre-trained Bert model, it can be more adapted to the specific language and context in the performance testing field, thereby improving the accuracy of the initial model for performance problem analysis.

[0042] (2) The initial performance problem analysis model is pruned and fine-tuned again using an L1 norm pruner to obtain a performance problem analysis model, thereby reducing model complexity, saving storage space and computing resources, reducing the risk of overfitting, and enhancing the generalization ability of the model.

[0043] (3) By building a preset performance problem library, the accumulation and reuse of historical performance test problems and historical performance problem solutions are promoted, thereby improving the efficiency of knowledge management.

[0044] (4) The performance problem location information is matched and retrieved through the pattern matching retrieval algorithm, and the most matching problem text block is quickly locked in the preset performance problem vector library, ensuring the accuracy and efficiency of the performance problem location information, reducing manual intervention, and shortening the performance problem location time. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0046] Figure 1 The present invention is a flowchart of a performance testing, analysis and positioning method based on artificial intelligence. DETAILED DESCRIPTION

[0047] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0048] See also Figure 1 , a performance test analysis and positioning method based on artificial intelligence, comprising:

[0049] S1: Acquire multi-source data of historical performance tests, and build a preset performance problem library through performance data matching according to the multi-source data of historical performance tests, wherein the multi-source data of historical performance tests includes historical performance test information, historical performance analysis information, and solutions to historical performance problems;

[0050] S2: According to the preset performance problem library, a performance problem analysis model is obtained by secondary fine-tuning of the pre-trained Bert large model;

[0051] S3: Acquire performance test information, the performance test information includes performance test data, error log data, and configuration data, and output performance problem categories through the performance problem analysis model according to the performance test information, the performance problem categories including code problem categories, configuration problem categories, and system problem categories;

[0052] S4: Obtaining performance problem location information according to the performance test information and the performance problem category through a pattern matching retrieval algorithm.

[0053] In step S1, building a preset performance problem library through performance data matching based on the historical performance test multi-source data specifically includes the following steps:

[0054] S101: extracting historical performance key indicator data through the LoadRunner tool according to historical performance test information, wherein the historical performance key indicator data includes historical performance test data, historical error log data, historical configuration data, and historical application code data; the historical performance test data includes historical throughput data, historical performance inflection point data, historical response time data, and historical system resource bottleneck data; the historical system resource bottleneck data includes historical CPU sampling data, historical heap memory model sampling data, and historical thread data; the historical configuration data includes historical system configuration data and historical middleware configuration data;

[0055] S102: Using a data cleaning tool to remove noise and abnormal values ​​from the historical performance key indicator data to obtain historical performance key indicator standard data;

[0056] S103: Obtain a preset performance problem library through annotation matching according to the historical performance key indicator standard data, the historical performance analysis information, and the historical performance problem solutions.

[0057] The step S103 comprises the following steps:

[0058] S103-1: Annotate and match the historical performance key indicator standard data, the historical performance analysis data, and the historical performance problem solution using a data annotation tool to obtain performance matching information;

[0059] S103-2: obtaining the preset performance problem library through JSON file storage according to the performance matching information, wherein the annotation matching includes code problem annotation matching, configuration problem annotation matching, and system problem annotation matching;

[0060] The code problem annotation matching includes memory leak problems, deadlock problems, circular reference problems, high algorithm complexity problems, and unreasonable data structure problems;

[0061] The configuration problem annotation matching includes improper system parameter settings, database connection pool configuration errors, unreasonable thread pool configuration problems, and middleware configuration errors;

[0062] The system problem annotation matching includes hardware failure problems, network delay problems, and system resource shortage problems.

[0063] It should be noted that the historical error log data needs to extract error information from application logs and system logs and classify and summarize them; the historical system configuration data specifically involves database connection pool configuration, thread pool configuration, and system parameter settings; the historical middleware configuration data specifically involves Web container configuration, message queue configuration, and cache configuration.

[0064] In this embodiment, the static code analysis results of the historical application code data are also stored in the preset performance problem library.

[0065] In step S2, the performance problem analysis model is obtained by fine-tuning the pre-trained Bert large model according to the preset performance problem library, which specifically includes the following steps:

[0066] S201: performing word segmentation and encoding according to the preset performance problem library through a Tokenization API to obtain a preset performance problem data set, wherein the Tokenization API is a Tokenization API provided by a pre-trained big Bert model;

[0067] S202: obtaining a Bert performance test large model by fine-tuning the training parameters of a pre-trained Bert large model according to the preset performance problem data set, wherein the training parameters of the pre-trained Bert large model include a learning rate, a batch size, and a number of iterations;

[0068] S203: According to the Bert performance test large model, a Bert performance test large model evaluation result is obtained by evaluating the Bert performance test large model through the Bert performance test large model evaluation index; according to the Bert performance test large model evaluation result, the model parameters are updated through the Adam optimizer to obtain the initial model for performance problem analysis; the Bert performance test large model evaluation index includes a cross entropy loss function and a Bert performance test large model evaluation index; the Bert performance test large model evaluation index includes a Bert performance test large model accuracy, a Bert performance test large model recall rate, and a Bert performance test large model F1 value;

[0069] S204: Pruning the initial performance problem analysis model through an L1 norm pruner and fine-tuning again to obtain a performance problem analysis model, the mathematical expression of which is:

[0070]

[0071] Among them, a is the training input of the initial model for performance problem analysis, b is the output target of the initial model for performance problem analysis, w is the trainable weight of the initial model for performance problem analysis, α is the balance factor of the initial model for performance problem analysis, and g(β) is the sparse scale factor penalty of the initial model for performance problem analysis.

[0072] In step S4, obtaining performance problem location information according to the performance test information and the performance problem category through a pattern matching retrieval algorithm specifically includes the following steps:

[0073] S401: constructing a preset performance problem vector library through a vector model according to the preset performance problem library;

[0074] Extracting performance problem document text information through the BeautifulSoup library according to the preset performance problem library, obtaining performance problem text blocks through text segmentation according to the performance problem document text information, and obtaining the preset performance problem vector library through vector conversion and creation of text block vector index according to the performance problem text blocks;

[0075] S402: Obtaining performance problem location information through vector retrieval according to the performance test information and the performance problem category;

[0076] According to the text block vector index, the closest text block vector is searched in the preset performance problem vector library through approximate nearest neighbor retrieval, which is the performance problem location information.

[0077] It should be noted that the approximate nearest neighbor search uses cosine similarity to measure the distance between text block vectors, and the mathematical expression is:

[0078]

[0079] Among them, pq is the dot product of text block vector p and text block vector q, ||p|| is the Euclidean norm of text block vector p, and ||q|| is the Euclidean norm of text block vector q.

[0080] In this embodiment, the code problem can be located to a specific function through a pattern matching retrieval algorithm.

[0081] A performance test analysis and positioning system based on artificial intelligence includes a performance test data acquisition module, a performance problem library construction module, a performance problem analysis module, and a performance problem positioning module;

[0082] The performance test data acquisition module is used to collect performance test information and perform data preprocessing;

[0083] The performance problem library building module is used to build a preset performance problem library based on historical performance test information, historical performance analysis information, and historical performance problem solutions;

[0084] The performance problem analysis module is used to receive performance test information and output performance problem categories through a performance problem analysis model;

[0085] The performance problem locating module is used to determine the performance problem locating information according to the performance test information and the performance problem category through a pattern matching retrieval algorithm.

[0086] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device.

[0087] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0088] The program code included on the computer readable medium can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or their combinations, and the programming language includes object-oriented programming languages-such as Java, Smalltalk, C++, and also includes conventional procedural programming languages-such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0089] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A performance test analysis and positioning method based on artificial intelligence, characterized in that: include: Acquire historical performance test multi-source data, and build a preset performance problem library through performance data matching based on the historical performance test multi-source data, wherein the historical performance test multi-source data includes historical performance test information, historical performance analysis information, and historical performance problem solutions; According to the preset performance problem library, a performance problem analysis model is obtained by secondary fine-tuning of the pre-trained Bert large model; Acquire performance test information, the performance test information including performance test data, error log data, and configuration data, and output performance problem categories through the performance problem analysis model according to the performance test information, the performance problem categories including code problem categories, configuration problem categories, and system problem categories; Performance problem location information is obtained according to the performance test information and the performance problem category through a pattern matching retrieval algorithm.

2. The performance test analysis and positioning method based on artificial intelligence according to claim 1 is characterized in that: The step of building a preset performance problem library by matching performance data based on the historical performance test multi-source data includes: Extracting historical performance key indicator data through the LoadRunner tool based on historical performance test information, wherein the historical performance key indicator data includes historical performance test data, historical error log data, historical configuration data, and historical application code data; According to the historical key performance indicator data, noise and abnormal values ​​are removed by a data cleaning tool to obtain standard data of the historical key performance indicators; A preset performance problem library is obtained through annotation matching according to the historical performance key indicator standard data, the historical performance analysis information, and the historical performance problem solutions.

3. The performance test analysis and positioning method based on artificial intelligence according to claim 2 is characterized in that: The preset performance problem library is obtained by labeling and matching the historical performance key indicator standard data, the historical performance analysis data, and the historical performance problem solution, including: According to the historical performance key indicator standard data, the historical performance analysis data, and the historical performance problem solution, performance matching information is obtained by annotating and matching with a data annotation tool; The preset performance problem library is obtained by storing the performance matching information in a JSON file, and the annotation matching includes code problem annotation matching, configuration problem annotation matching, and system problem annotation matching.

4. The performance test analysis and positioning method based on artificial intelligence according to claim 3 is characterized in that: The code problem annotation matching includes memory leak problems, deadlock problems, circular reference problems, high algorithm complexity problems, and unreasonable data structure problems; The configuration problem annotation matching includes improper system parameter settings, database connection pool configuration errors, unreasonable thread pool configuration problems, and middleware configuration errors; The system problem annotation matching includes hardware failure problems, network delay problems, and system resource shortage problems.

5. The performance test analysis and positioning method based on artificial intelligence according to claim 1 is characterized in that: The performance problem analysis model obtained by secondary fine-tuning of the pre-trained Bert large model according to the preset performance problem library includes: According to the preset performance problem library, a preset performance problem data set is obtained by performing word segmentation and encoding through a Tokenization API, wherein the Tokenization API is a Tokenization API provided by the pre-trained Bert large model; According to the preset performance problem data set, the Bert performance test large model is obtained by fine-tuning the training parameters of the pre-trained Bert large model, wherein the training parameters of the pre-trained Bert large model include a learning rate, a batch size, and a number of iterations; According to the Bert performance test large model, a Bert performance test large model evaluation result is obtained by evaluating the Bert performance test large model through the Bert performance test large model evaluation index; according to the Bert performance test large model evaluation result, the model parameters are updated through the Adam optimizer to obtain the performance problem analysis initial model; the Bert performance test large model evaluation index includes a cross entropy loss function and a Bert performance test large model evaluation index; the Bert performance test large model evaluation index includes a Bert performance test large model accuracy, a Bert performance test large model recall rate, and a Bert performance test large model F1 value; The performance problem analysis initial model is pruned by an L1 norm pruner and fine-tuned again to obtain a performance problem analysis model.

6. The performance test analysis and positioning method based on artificial intelligence according to claim 1 is characterized in that: The obtaining of the performance problem location information by using a pattern matching retrieval algorithm according to the performance test information and the performance problem category includes: Constructing a preset performance problem vector library through a vector model according to the preset performance problem library; Performance problem location information is obtained through vector retrieval according to the performance test information and the performance problem category.

7. The performance test analysis and positioning method based on artificial intelligence according to claim 6 is characterized in that: Constructing a preset performance problem vector library through a vector model according to the preset performance problem library includes: Extracting performance problem document text information through the BeautifulSoup library according to the preset performance problem library; Obtaining a performance problem text block by text segmentation according to the text information of the performance problem document; The preset performance problem vector library is obtained by converting the performance problem text block into a vector form and creating a text block vector index.

8. An artificial intelligence-based performance test analysis and positioning system using the artificial intelligence-based performance test analysis and positioning method according to claim 1, comprising a performance test data acquisition module, a performance problem library construction module, a performance problem analysis module, and a performance problem positioning module, characterized in that: The performance test data acquisition module is used to collect performance test information and perform data preprocessing; The performance problem library building module is used to build a preset performance problem library based on historical performance test information, historical performance analysis information, and historical performance problem solutions; The performance problem analysis module is used to receive performance test information and output performance problem categories through a performance problem analysis model; The performance problem locating module is used to determine the performance problem locating information according to the performance test information and the performance problem category through a pattern matching retrieval algorithm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the performance testing, analysis and positioning method based on artificial intelligence as described in any one of claims 1-7 is implemented.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions, when executed by a computer processor, are used to execute the artificial intelligence-based performance testing analysis and positioning method as described in any one of claims 1-7.

Images