Finished vehicle testing method and device based on LLM (Logistics Language Model), medium and equipment
Through the automated testing method based on the LLM large language model, structured test cases are generated and automated tests are performed, the problem of manual testing dependence in the existing technology is solved, the testing efficiency and accuracy are improved, the functional points of the entire vehicle are covered, and the functional points are reduced. Cost and labor consumption.
Patent Information
- Application Number
- CN202410983060.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
AI Technical Summary
In the prior art, vehicle testing and vehicle quality review rely on manual testing, resulting in high upstream and downstream communication costs, difficult project progress, difficult to control review costs, and unavailable functional quality, and lack of automation to improve testing efficiency and accuracy.
Using the vehicle testing method based on the LLM large language model, the product function requirements document is semantic analysis, structured test cases are generated, and the signal simulation tool is used to perform automated tests to evaluate whether the test results are of quality. Review criteria.
Implement automated vehicle testing, greatly improving the efficiency of test case generation and testing process efficiency, shortening the test cycle, improving the accuracy and resource utilization rate of the test, covering the functional points of the entire vehicle, reducing manual testing time and labor costs .
Smart Images

Figure CN118939541A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a vehicle testing method and device, medium and equipment based on an LLM large language model. Background Art
[0002] With the continuous development of automobile electronic intelligence, the number of body components has gradually increased, the vehicle functions have been continuously improved, and the electrical architecture evolution of the vehicle system has undergone earth-shaking changes, and CC domain control; Autosar and other system architecture models have emerged, but the corresponding vehicle quality review means are relatively simple, and only integration testing and vehicle audit review are performed at key project nodes such as OTS, PP, and SOP nodes. As the project progresses, the tighter the quality review time, the more quality problems are discovered later, the larger the scope of impact, the more serious the losses caused, the less log data support, the shorter the analysis and repair allowable time, and the higher the risk to the project. At present, vehicle testing and vehicle quality review are almost entirely dependent on manual testing, and due to the single description of vehicle function standards and product requirement documents, the test reviewer may have a biased understanding of the document, and the boundary check of the vehicle function may be omitted. In order to ensure the quality of the vehicle function, multiple rounds of review need to be repeated, resulting in a large amount of development and testing manpower and time consumption. Summary of the invention
[0003] Based on this, it is necessary to propose a vehicle testing method and device, medium and equipment based on the LLM large language model to address the above technical issues, and an automated vehicle testing method based on large model-generated test cases to replace the traditional real vehicle reviewer integration test. This can fundamentally solve the problems of huge upstream and downstream communication costs, unpredictable project progress, uncontrollable review costs, and unguaranteed functional quality during vehicle development.
[0004] In a first aspect, the present application provides a vehicle testing method based on an LLM large language model, the method comprising:
[0005] When receiving product function requirements from the user, extract the product requirement document corpus contained in the product function requirements as input data for the pre-established LLM large language model; perform semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases;
[0006] Convert the structured test case into a test script, and perform automated testing in the test case execution scenario simulated by the signal simulation tool;
[0007] Evaluate whether the test results in different execution scenarios meet the quality review standards.
[0008] Optionally, the pre-establishment of the LLM large language model includes:
[0009] Perform corpus structured processing on historical vehicle model data, generate vehicle knowledge graph, and build product demand corpus;
[0010] Obtain the test cases written by reading the product requirement document recorded in the review case table in the historical vehicle model data; map the test case narrative text to the product requirement corpus, and embed a word segmenter in the product requirement corpus, and generate the product requirement document and the test case into a predefined causal relationship vector table format through the word segmenter;
[0011] Inputting the causal relationship vector table of the product requirement document and the test case into the vehicle knowledge graph to form a reasoning context of the test case;
[0012] The causal relationship vector corresponding to the reasoning context of the test case is modeled to obtain the LLM large language model.
[0013] Optionally, the historical vehicle model data includes: vehicle controller VDC functional standards, product requirement documents, review case tables and review quality report data; wherein, the review case tables are used to record test cases written by reading product requirement documents.
[0014] Optionally, the corpus structured processing of the historical vehicle model data to generate a vehicle knowledge graph includes:
[0015] The product requirement documents in the base historical vehicle model data are decomposed into functional long texts, combined with the vehicle controller VDC functional standards to generate a vehicle knowledge graph.
[0016] Optionally, the step of generating the product requirement document and the test case into a predefined format of the causal relationship vector table by a word segmenter includes: obtaining actions and conditions included in the test case by identifying the test case;
[0017] Mapping the actions and conditions included in the test case to obtain a mapping relationship between the test case actions and conditions, and adding script information to the mapping relationship between the test case actions and conditions to obtain a mapping relationship carrying the script information;
[0018] A causal relationship vector table is generated according to the mapping relationship between the test case actions and conditions, and the mapping relationship carrying the script information.
[0019] Optionally, performing semantic analysis on the product requirement document corpus by using the LLM large language model to obtain structured test cases includes:
[0020] The input product requirement document corpus is converted into vector data through deep learning and input into the Tranformer encoder to learn the correlation between the product requirement document and the test case to obtain the product requirement representation features; all product requirement representation features of the product requirement document corpus are merged, and the merged product requirement representation features are defined as structured test cases.
[0021] Optionally, the learning of the correlation between the product requirement document and the test case includes: setting a causal relationship vector for the product requirement document and the reasoning context of each test case through a Tranformer encoder to represent the correlation between the product requirement and the test case;
[0022] All product requirement representation features related to their respective causal relationship vectors are aggregated in turn according to the number of nodes in the reasoning context of the test case to obtain the causal relationship of the neighboring node product requirements; by combining the causal relationship of product requirements, the product requirement representation features corresponding to the test case are obtained.
[0023] In a second aspect, the present application provides a vehicle testing device based on the LLM large language model, comprising:
[0024] The test case generation module is used to extract the product requirement document corpus contained in the product requirement document as input data of the pre-established LLM large language model when receiving the product function requirement from the user; perform semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases;
[0025] An automated testing module, used to convert the structured test cases into test scripts, and perform automated testing in the test case execution scenarios simulated by the signal simulation tool;
[0026] The evaluation module is used to evaluate whether the test results in different execution scenarios meet the quality review standards.
[0027] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any one of the methods described in the first aspect.
[0028] In a fourth aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods in the first aspect when executing the computer program.
[0029] Compared with the closest prior art, the present invention has the following beneficial effects:
[0030] The present invention proposes a whole vehicle testing method and device, medium and equipment based on the LLM large language model, which reuses historical test cases and review data of other models, builds a model based on deep learning, improves the model's understanding of quality review standards, generates rich test cases in structured and complex scenarios, and completely covers all vehicle functional points.
[0031] The present invention proposes a whole vehicle testing method and device, medium and equipment based on the LLM large language model, which are mainly divided into two parts: test case generation and automated testing. In the test case generation stage, through the LLM large language model, rich test cases in structured and complex scenarios can be automatically generated in a short time, completely covering the functional points of the whole vehicle, greatly improving the efficiency of test managers in writing test cases. In the automated testing stage, the test scripts are directly translated through the test cases to execute automated testing, which saves manual testing time, improves the accuracy of complicated test steps, and greatly improves the efficiency of the test process.
[0032] The automated testing solution provided by the present invention uses programming human-computer interaction external devices and signal simulation tools to simulate the execution scenarios of multiple test cases, replacing or assisting the reviewer in simulating the cockpit interaction test scenarios of complex and cumbersome processes such as user operation of the vehicle computer and driving the vehicle. Through the automated testing process provided by this application, the test cycle can be shortened, the test efficiency can be greatly improved, and the regression test time can be shortened. The resource utilization rate of the test vehicle and equipment is high, making the cumbersome path test in the complex scene of the whole vehicle more accurate and efficient. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.
[0034] Figure 1 The flowchart of the vehicle testing method based on the LLM large language model in one embodiment;
[0035] Figure 2 Generate an architecture diagram for a test case in an embodiment;
[0036] Figure 3 A flowchart of automated vehicle quality testing in one embodiment;
[0037] Figure 4 It is a schematic diagram of the structure of a whole vehicle testing device based on the LLM large language model in one embodiment;
[0038] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0039] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0040] The embodiment of the present application provides a vehicle testing method and device based on the LLM large language model, a medium and equipment, which are mainly used in vehicle equipment. The method can be applied to a terminal, a server, or a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be, but is not limited to, various computers, laptops, smart phones, tablet computers, etc.
[0041] In one embodiment, Figure 1 As shown, a whole vehicle testing method based on a large model test case according to the following specific implementation scheme is provided, and the method comprises the following features:
[0042] S101, when receiving the product function requirement of the user, extract the product requirement document corpus contained in the product function requirement as the input data of the pre-established LLM large language model; perform semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases;
[0043] S102 converts the structured test case into a test script, and performs automated testing in the test case execution scenario simulated by the signal simulation tool;
[0044] S103 evaluates whether the test results in different execution scenarios meet the quality review standards.
[0045] In step S101 of the above embodiment, the generation of test cases is centered on the LLM large language model. The existing product requirement documents and the test cases written by reviewers after reading these documents are used as the basis for model construction. The product requirement corpus is converted into vector data through deep learning, and the action and condition mapping relationship of the test case is decomposed to construct an LLM large language model that generates structured test cases with full coverage of multiple scenario boundaries based on product requirement textual reasoning.
[0046] In the above embodiment, the pre-establishment of the LLM large language model in step S101 includes:
[0047] Perform corpus structured processing on historical vehicle model data, generate vehicle knowledge graph, and build product demand corpus;
[0048] Obtain the test cases written by reading the product requirement document recorded in the review case table in the historical vehicle model data; map the test case narrative text to the product requirement corpus, and embed a word segmenter in the product requirement corpus, and generate the product requirement document and the test case into a predefined causal relationship vector table format through the word segmenter;
[0049] Inputting the causal relationship vector table of the product requirement document and the test case into the vehicle knowledge graph to form a reasoning context of the test case;
[0050] The causal relationship vector corresponding to the reasoning context of the test case is modeled to obtain the LLM large language model.
[0051] In the above embodiment, the format of the predefined causal relationship vector table refers to a format included in the general LLM large language model.
[0052] The historical vehicle model data includes: vehicle controller VDC function standards, product requirement documents, review case tables and review quality report data; wherein, the review case tables are used to record test cases written by reading product requirement documents.
[0053] In the above embodiment, the historical vehicle model data is subjected to corpus structured processing to generate a vehicle knowledge graph, including:
[0054] The product requirement documents in the base historical vehicle model data are decomposed into functional long texts, combined with the vehicle controller VDC functional standards to generate a vehicle knowledge graph.
[0055] In the above embodiment, the step of generating the product requirement document and the test case into a predefined causal relationship vector table format by a word segmenter includes: identifying the test case to obtain the actions and conditions included in the test case;
[0056] Mapping the actions and conditions included in the test case to obtain a mapping relationship between the test case actions and conditions, and adding script information to the mapping relationship between the test case actions and conditions to obtain a mapping relationship carrying the script information;
[0057] A causal relationship vector table is generated according to the mapping relationship between the test case actions and conditions, and the mapping relationship carrying the script information.
[0058] In the above embodiment, the semantic analysis of the product requirement document corpus is performed by the LLM large language model to obtain structured test cases including:
[0059] The input product requirement document corpus is converted into vector data through deep learning and input into the Tranformer encoder to learn the correlation between the product requirement document and the test case to obtain the product requirement representation features; all product requirement representation features of the product requirement document corpus are merged, and the merged product requirement representation features are defined as structured test cases.
[0060] In the above embodiment, the learning of the correlation between the product requirement document and the test case includes: setting a causal relationship vector for the product requirement document and the reasoning context of each test case through a Tranformer encoder to represent the correlation between the product requirement and the test case;
[0061] All product requirement representation features related to their respective causal relationship vectors are aggregated in turn according to the number of nodes in the reasoning context of the test case to obtain the causal relationship of the neighboring node product requirements; by combining the causal relationship of product requirements, the product requirement representation features corresponding to the test case are obtained.
[0062] In one embodiment, Figure 2 As shown, for the generation of test cases, the following embodiments are proposed:
[0063] 1.1 Build a corpus of product requirement documents, decompose the complete functional description document of the module into long text of unit functions, convert the technical terms of the whole vehicle controller assembly into tokens, and build a knowledge graph of the whole vehicle through the encoder-only models encoder;
[0064] 1.2 Test case document construction vector table, mapping the test case narrative text to the product document corpus, and embedding the product requirement-test case statement into a causal relationship vector table through the word segmenter;
[0065] 1.3 The "Requirement-Use Case" decoder generates an LLM model from 1.1 Product-Technical Terms Encoder and 1.2 Vector Table, realizing the function of inputting product requirement document corpus and outputting structured test cases;
[0066] 1.4 A test script generator may be used to translate the structured test cases in 1.3 into test scripts. The test scripts are predefined text formats that support automated testing by automated testing systems.
[0067] 1.5 Human-computer interaction external equipment and signal simulation system: parse the 1.4 test script, collect preset interactive test meta-instructions, and generate a series of vehicle operation instructions that can be executed in complex scenarios;
[0068] 1.6 Review and arbitration system: After executing the test instructions, the sensor equipment detects the results, organizes the test case pass rate, and outputs the functional item quality review report.
[0069] In the above embodiment, step S102 can convert the use case into a test script and input it into the automated test system. The automated test system simulates the test scenario through human-computer interaction external devices and signal simulation tools, collects test results through sensor devices, and calculates whether the test case passes the one-stop quality review process.
[0070] Specifically, the automated testing system in the embodiment of the present invention is composed of a test script interpreter and executor, a programmable human-computer interaction external device, and a signal simulation tool. The test script controls the external tool to generate a virtual user to operate the vehicle computer / drive the vehicle, and at the same time simulates the vehicle control signal to create a vehicle digital twin test system, and realistically restore the test case execution scenario; then, the test results are input into the arbitration system through the acoustic, optical, and electromagnetic sensor equipment and the image acquisition equipment to determine whether the current test case has passed. Specifically, during the test process, the application of sensors such as image acquisition and processing, potential detection, and acoustic, optical, and electromagnetic induction can replace manual detection to verify the test case operation results, and capture data in real time to accurately locate the problem points.
[0071] Based on the above specific implementation scheme, combined with Figure 3 As shown, the present application provides the following embodiments for determining whether the current test case meets the standard:
[0072] The function development manager inputs the VDC function standards into the vehicle technology knowledge graph in advance to form a reasoning context;
[0073] The product requirement document PRD of each module is input into the LLM Tranformer encoder to construct the product requirement representation features.
[0074] The assessor uses the VDC functional standards and PRD to define the vehicle inspection standards for the arbitration system, such as the status of windows, doors, seat adjustment, air conditioning, audio, lights, etc.
[0075] Product managers and interaction designers define the vehicle inspection standards for the review and arbitration system, such as the vehicle page interaction process and vehicle operation performance indicators;
[0076] The quality reviewer or test manager presets the vehicle test scenario parameters such as smoke test, integration test, audit review standard test, etc. and inputs them into the system. After the test case generation system converts the VDC-PRD document into a structured test case table, the reviewer or test manager reviews the case table.
[0077] The reviewer inputs the reviewed use case table into the test script generator and translates it into a test script;
[0078] The human-computer interaction device and signal simulation system run the test script and collect data such as images, sounds, and component potential signals in the test vehicle through sensors;
[0079] The test results are input into the review and arbitration system to judge whether the current use case has passed, collect vehicle logs and vehicle status data, report to the workflow system, and organize the test case pass rates corresponding to the required functional items, and output the functional item quality review report.
[0080] The above-mentioned embodiment uses the review and arbitration system to collect vehicle data, use case evaluation, vehicle system monitoring, log capture and other one-stop review processes, which can greatly shorten the time spent on the testing phase in vehicle development.
[0081] The above-mentioned automated vehicle testing method can fundamentally solve the problems of huge upstream and downstream communication costs, unpredictable project progress, uncontrollable review costs, and unguaranteed functional quality during the vehicle development process.
[0082] Based on the above specific implementation scheme, this application provides an embodiment of an automated vehicle quality review method for the following test cases, which includes:
[0083] 1. Device support
[0084] The implementation of the automated review process requires a "human-computer interaction simulation device", which simulates the driver / user operating the central control screen and performs interactive operations such as single-finger / multi-finger clicking, double-clicking, sliding, dragging, voice input, and physical operations of the central control hard buttons and square control hard buttons. This device has the test script execution function; at the same time, an "environmental signal acquisition device" is required to collect the vehicle status and signals after the "human-computer interaction simulation device" operates the vehicle computer to verify whether the operation is successful; the "environmental signal acquisition device" consists of an image acquisition sensor, a sound acquisition sensor, a potential signal sensor and an integrated verification system, among which the integrated verification system has the test script verification function.
[0085] 2. Verification Standards
[0086] Taking the cockpit function review of a certain car model as an example, the reviewer inputs the test script translated from the test case generated by the large model into the "human-computer interaction simulation device", and inputs the verification script corresponding to the test script into the environmental signal acquisition device. Taking the functional test script of the air conditioning application as an example:
[0087] The unified review function items include central control screen interaction: turn on / off the air conditioner, adjust the air conditioner temperature and air volume, adjust the air outlet angle, adjust the seat ventilation / heating / massage, and adjust the gear position; voice interaction: turn on / off the air conditioner, adjust the temperature and air volume, physical button interaction: adjust the seat back / cushion / lumbar support angle.
[0088] The collective review function items include central control screen / voice interaction: turn on / off nap mode, that is, turn on the air conditioner, automatically adjust the air conditioner temperature to 25 degrees and the air volume to 3 levels; unfold the seat back / lumbar support to the maximum angle, adjust the seat massage to the maximum level, close the windows, and play white noise on the car audio.
[0089] Correspondingly, the review and verification script of the air-conditioning application verifies the air-conditioning on / off, temperature / air volume detection, seat function adjustment and angle adjustment, window opening / closing, audio playback detection, etc. after the test script of the air-conditioning application is run; in the collective functional review, the operation of the air-conditioning, seats, windows, audio and other equipment is tested at the same time to see if they comply with the preset functional items, to verify whether this function item has passed.
[0090] 3. Implementation process
[0091] Take the air conditioning application collection review function item of a certain car model: nap mode as an example: first install the human-computer interaction simulation device on the central control screen, and install the hard button simulation device on the steering wheel; and place the four image collectors of the environmental signal acquisition device in the position facing the car window, install temperature and wind speed sensors in the cabin, and uniformly install the air conditioning compressor, seat motor airbag, window rocker motor. Potential signal sensors of electrical equipment such as audio amplifiers, and transmit the collected sensor signals back to the integrated verification system. Verification function item: Nap mode is turned on. After the human-computer interaction device turns on the car machine, simulate the user's operation on the central control screen to enter the air conditioning application-advanced function menu, click the nap mode button, or wake up "turn on nap mode" through the language input device. At this time, the environmental signal acquisition device collects the window status image movement status. Real-time temperature / air outlet wind speed data and potential signals of air conditioners / seats / windows / audio equipment are transmitted back to the integrated verification system according to the system time. Through image recognition, it is used to determine whether the window is closed, whether the temperature inside the car is 25 degrees, whether the wind speed at the air outlet is level 3, and whether the potential signal meets the potential change curve diagram for normal function execution.
[0092] The above process is repeated 2-10 times according to the review criteria, and reversely verified in the fault state of the electrical equipment. Combined with the previous test results, it is determined whether the air conditioning application-nap mode function item has passed the review.
[0093] Based on the same inventive concept, the embodiment of the present application also provides a whole vehicle testing device based on the LLM large language model and a whole vehicle testing method based on the LLM large language model. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of one or more whole vehicle testing devices based on the LLM large language model provided below can be referred to the limitations of the whole vehicle testing method based on the LLM large language model above, and will not be repeated here.
[0094] In one embodiment, the present invention also provides a vehicle testing device based on the LLM large language model, such as Figure 4 As shown, it includes: a test case generation module 110, an automated test module 120, and an evaluation module 130; wherein:
[0095] The test case generation module 110 is used to extract the product requirement document corpus contained in the product requirement document as input data of the pre-established LLM large language model when receiving the product function requirement from the user; perform semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases;
[0096] An automated testing module 120, configured to convert the structured test cases into test scripts, and to perform automated testing in the test case execution scenarios simulated by the signal simulation tool;
[0097] The evaluation module 130 is used to evaluate whether the test results in different execution scenarios meet the quality review standards.
[0098] At the same time, in one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a whole vehicle testing method based on the LLM large language model is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse.
[0099] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0100] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.
[0101] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.
[0102] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.
[0103] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0104] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0105] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A vehicle testing method based on LLM large language model, It is characterized in that The method comprises: When receiving product function requirements from the user, extract the product requirement document corpus contained in the product function requirements as input data for the pre-established LLM large language model; perform semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases; Convert the structured test case into a test script, and perform automated testing in the test case execution scenario simulated by the signal simulation tool; Evaluate whether the test results in different execution scenarios meet the quality review standards.
2. The method according to claim 1, It is characterized in that The pre-establishment of the LLM large language model includes: Perform corpus structured processing on historical vehicle model data, generate vehicle knowledge graph, and build product demand corpus; Obtain the test cases written by reading the product requirement document recorded in the review case table in the historical vehicle model data; Mapping the test case narrative text to a product requirement corpus, and embedding a tokenizer in the product requirement corpus, and generating the product requirement document and the test case into a format of a predefined causal relationship vector table through the tokenizer; Inputting the causal relationship vector table of the product requirement document and the test case into the vehicle knowledge graph to form a reasoning context of the test case; The causal relationship vector corresponding to the reasoning context of the test case is modeled to obtain the LLM large language model.
3. The method according to claim 2, It is characterized in that The historical vehicle model data includes: vehicle controller VDC function standards, product requirement documents, review case tables and review quality report data; wherein the review case tables are used to record test cases written by reading product requirement documents.
4. The method according to claim 3, It is characterized in that The corpus structured processing of the historical vehicle model data to generate the vehicle knowledge graph includes: Decompose the product requirement documents in the historical vehicle model data into functional long texts, and combine them with the vehicle controller VDC functional standards to generate a vehicle knowledge graph.
5. The method according to claim 2, It is characterized in that The method of generating the format of the predefined causal relationship vector table from the product requirement document and the test case by the word segmenter includes: obtaining the actions and conditions included in the test case by identifying the test case; Mapping the actions and conditions included in the test case to obtain a mapping relationship between the test case actions and conditions, and adding script information to the mapping relationship between the test case actions and conditions to obtain a mapping relationship carrying the script information; A causal relationship vector table is generated according to the mapping relationship between the test case actions and conditions, and the mapping relationship carrying the script information.
6. The method according to claim 1, It is characterized in that The method of performing semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases includes: The input product requirement document corpus is converted into vector data through deep learning and input into the Tranformer encoder to learn the correlation between the product requirement document and the test case to obtain the product requirement representation features; all product requirement representation features of the product requirement document corpus are merged, and the merged product requirement representation features are defined as structured test cases.
7. The method according to claim 6, It is characterized in that The learning of the correlation between the product requirement document and the test case includes: setting a causal relationship vector for the product requirement document and the reasoning context of each test case through a Tranformer encoder to represent the correlation between the product requirement and the test case; All product requirement representation features related to their respective causal relationship vectors are aggregated in turn according to the number of nodes in the reasoning context of the test case to obtain the causal relationship of the neighboring node product requirements; by combining the causal relationship of product requirements, the product requirement representation features corresponding to the test case are obtained.
8. A vehicle testing device based on LLM large language model, It is characterized in that include: The test case generation module is used to extract the product requirement document corpus contained in the product requirement document as input data of the pre-established LLM large language model when receiving the product function requirement from the user; perform semantic analysis on the product requirement document corpus through the LLM large language model to obtain structured test cases; An automated testing module, used to convert the structured test cases into test scripts, and perform automated testing in the test case execution scenarios simulated by the signal simulation tool; The evaluation module is used to evaluate whether the test results in different execution scenarios meet the quality review standards.
9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A computer device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Integrated test method for simulating semiconductor equipment and EAP system and related equipment
CN119377129A
Automatic test method and device for vehicle body controller
CN119987331A
Intelligent automobile key scene optimization generation method based on large model knowledge guidance
CN120144481A
Intelligent cockpit voice automatic recognition method and device
CN120236581A
Method and device for remotely testing vehicle machine in different places and medium
CN120849298A
