Test case generation apparatus, method, and computing device

Abstract

The application relates to the technical field of testing, and discloses a test case generation device, a method and a computing device. The device comprises a test module and a test defect mining module, and the test module and the test defect mining module are in communication connection. The test module is used for generating a basic test case of a target test object on a vehicle, executing the basic test case, and sending a first instruction to the test defect mining module. The test defect mining module is used for collecting a time sequence signal stream of a target signal in a basic test case execution process in response to the first instruction, generating a natural language description text corresponding to the time sequence signal stream, determining a test defect of the basic test case based on the natural language description text, and feeding back the test defect to the test module. The test module is further used for generating an enhanced test case based on the test defect, and executing the enhanced test case to compensate for the test defect. The technical scheme provided in the application embodiment improves the accuracy of the test defect and the coverage capability of the test case for complex vehicle working conditions.

Description

Technical Field

[0001] This invention relates to the field of testing technology, and more specifically to a test case generation device, method, and computing equipment. Background Technology

[0002] A test case is a set of input conditions, execution conditions, expected results, and execution steps designed for a specific objective. In the field of vehicle chassis testing, test cases include the initial state of the vehicle or controller, control commands or external stimuli issued to the controller, and the physical response or signal state generated under the action of control commands or external stimuli.

[0003] As the functions of vehicle chassis domain control systems become increasingly complex, the number of test cases is growing exponentially. Traditional methods of manually writing test cases are no longer sufficient to meet the requirements of development efficiency and coverage. Therefore, inputting requirement documents into large language models to generate test cases has become an important means of improving the efficiency of test case generation.

[0004] However, current test cases are mainly generated by static text knowledge. Large language models are unable to understand the real physical state of the vehicle chassis domain control system execution process based on massive multi-dimensional signals, and cannot meet the requirements of complex vehicle systems for test coverage and reliability. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this application is to provide a test case generation device, method and computing device, which improves the accuracy of test defects and the ability of test cases to cover complex vehicle operating conditions.

[0006] In a first aspect, embodiments of this application provide a test case generation device, the device comprising: a test module and a test defect mining module, the test module and the test defect mining module being communicatively connected.

[0007] The aforementioned testing module is used to generate basic test cases for the target test objects on the vehicle, execute the basic test cases, and send the first instruction to the test defect discovery module.

[0008] The aforementioned test defect mining module is used to collect the timing signal stream of the target signal during the execution of the basic test cases in response to the first instruction, generate the natural language description text corresponding to the timing signal stream, determine the test defects of the basic test cases based on the natural language description text, and feed the test defects back to the test module.

[0009] The aforementioned testing module is also used to generate enhanced test cases based on test defects, and execute the enhanced test cases to compensate for the test defects.

[0010] The target signal is the signal in the vehicle's multidimensional signal that is related to the performance of the target test object.

[0011] The technical solution provided in this application converts the multi-dimensional time-series signal stream generated during vehicle execution into natural language description text through a test defect mining module. This enables the large language model to directly understand the vehicle's actual test data based on the natural language description text and identify test defects based on the vehicle's actual test data. Consequently, the test module generates enhanced test cases based on the test defects and executes the enhanced test cases to compensate for the test defects. This method has strong self-optimization capabilities and improves the accuracy of test defects and the coverage of test cases for complex vehicle operating conditions.

[0012] One possible implementation involves vehicle multidimensional signals including: vehicle dynamics signals, vehicle bus signals, and control signals from the vehicle controller. Test defects include at least one of the following: first-type test function points that violate physical constraints in the domain knowledge base of the target test object; second-type test function points with potential risks; and test coverage blind spots in the target test signal combination. The target test signal combination is a combination of test signals whose test coverage is lower than a first preset test coverage threshold. In the above implementation, by identifying and compensating for test defects such as violations of physical constraints, potential risks, and test coverage blind spots, invalid test cases that violate physical laws are effectively avoided, potential risks are reduced, and test case coverage is improved.

[0013] One possible implementation of the aforementioned test defect mining module includes: a data acquisition unit, a natural language conversion unit, and a test defect mining unit. The natural language conversion unit is communicatively connected to both the data acquisition unit and the test defect mining unit. The data acquisition unit acquires a timing signal stream in response to a first instruction and sends the timing signal stream to the natural language conversion unit. The natural language conversion unit determines the associated test signals of the test function points based on the test function points of the basic test cases, determines the indicator characteristics of the statistical indicators of the associated test signals, determines the natural language description text based on the indicator characteristics and a domain knowledge base, and sends the natural language description text to the test defect mining unit. The test defect mining unit determines the test defects of the basic test cases based on the natural language description text and feeds the test defects back to the test module. This implementation, based on indicator characteristics and a domain knowledge base, converts high-frequency abstract physical signals into natural language description text that can be understood by the model, achieving automatic supplementation of test cases, breaking down the barrier between the physical execution layer and the semantic generation layer, and further improving the accuracy and coverage of test defects determined based on the natural language description text.

[0014] One possible implementation involves a basic test case comprising at least one test function point, and associated test signals comprising at least one test signal. A target test signal among the at least one test signal is associated with a target test function point among the at least one test function point. Specifically, the aforementioned test defect mining unit is used to: determine whether the target test function point satisfies a first condition based on natural language description text; and, if the target test function point satisfies the first condition, determine that the target test function point is a first type of test function point. The first condition includes: the physical relationship between multiple associated test signals of the target test function point does not conform to the defined physical relationship of multiple associated test signals of the target test function point in the physical constraints; and the signal value of the target test signal exceeds the signal limit value of the target test signal in the physical constraints. This implementation further reduces invalid test cases by identifying test function points that violate physical constraints or have abnormal signal values ​​through physical constraints and signal limitations.

[0015] One possible implementation is that the aforementioned test defect mining unit is specifically used to: determine whether the natural language description text indicates the occurrence of the target scene; and, if the natural language description text indicates the occurrence of the target scene, determine the test function point corresponding to the identified dangerous event as a second type of test function point. Here, the target scene is a scene that meets the dangerous triggering conditions associated with the identified dangerous event. This implementation, by recognizing potential defects or abnormally high-risk scenes already exhibited in the data through natural language description text, can effectively capture implicit relationships and semantic ambiguities between data in conjunction with the context, making it less likely to miss test function points corresponding to identified dangerous events, thus improving the accuracy of the second type of test function points.

[0016] One possible implementation is that the aforementioned test defect mining unit is specifically used to: determine the range of tested signal values ​​for any combination of test signals based on natural language description text; determine the test coverage of any combination of test signals based on the range of tested signal values ​​and the full range of signal values ​​for any combination of test signals; and determine whether any combination of test signals is a target combination of test signals based on the test coverage of any combination of test signals. Wherein, any combination of test signals consists of at least one test signal from the associated test signals. This implementation determines test coverage based on natural language description text, automatically identifies parameter blind spots, facilitates the generation of supplementary test cases for parameter blind spots, and enables iterative evolution of test cases.

[0017] One possible implementation involves the aforementioned testing module, comprising a test case generation unit and a test execution unit, with the test case generation unit communicatively connected to the test execution unit. The test case generation unit acquires the basic test documents and the corresponding professional documents for the target test object. Based on the basic test documents, professional documents, and a domain knowledge base, it generates basic test cases that do not violate physical constraints. It then sends these basic test cases to the test execution unit for execution. The basic test documents include test requirements documents, test case design specifications, and historical test case examples. The professional documents include the technical documents for the target test object, the technical documents for its components, and the dynamic model of the dynamic system in which the target test object resides. This implementation, by generating and executing basic test cases that do not violate physical constraints based on the basic test documents, professional documents, and domain knowledge base, reduces invalid test cases and improves testing efficiency.

[0018] One possible implementation involves the test case generation unit specifically used to: extract test items covered by the test requirements document; input the first text vector corresponding to the test item and the second text vector corresponding to the professional document into the retrieval model to obtain text vector segments in the professional document that are related to the test item and conform to physical constraints. Based on the text vector segments, the third text vector corresponding to the test base document, and the prompt word template text vector, basic test cases are generated. Here, the test item indicates the test function point and the corresponding execution preconditions. The retrieval model is fine-tuned using a domain knowledge base based on a large language model. This implementation generates basic test cases by retrieving text vector segments that conform to physical constraints through the retrieval model. The physical constraints correspond to the test items, resulting in a higher degree of matching between the generated basic test cases and the test items.

[0019] In one possible implementation, the test case generation unit is further configured to: generate enhanced test cases based on target information, and send the enhanced test cases to the test execution unit for execution. The test execution unit is further configured to: after executing the enhanced test cases, send a first instruction to the test defect mining module. The test defect mining module is further configured to, in response to the first instruction, collect the timing signal stream of the target signal during the execution of the enhanced test cases, and determine the test defects of the enhanced test cases based on the timing signal stream of the target signal during the execution of the enhanced test cases. If the test defect of the enhanced test cases does not meet the stop feedback condition, the test defect of the enhanced test cases is fed back to the test case generation unit. The target information includes both the test defects and the test defects of the enhanced test cases. This implementation generates enhanced test cases for test defects, and then re-determines and compensates for the test defects of the enhanced test cases, achieving automatic evolution of test requirements and automatic improvement of test case quality and test coverage.

[0020] One possible implementation includes stopping the feedback condition as follows: there is no target test signal combination, and no new first-type test function points and second-type test function points are added in consecutive rounds of testing of the predicted number.

[0021] Secondly, this application provides a test case generation method, which is applied to the test defect mining module in the test case generation device of any of the above implementations. The method includes: In response to the first instruction sent by the testing module, the system acquires the time-series signal stream of the target signal during the execution of the basic test cases, generates natural language description text corresponding to the time-series signal stream, identifies test defects in the basic test cases based on the natural language description text, and feeds these defects back to the testing module. The testing module then generates enhanced test cases based on these defects and executes the enhanced test cases to compensate for the defects. The target signal is a signal related to the performance of the target test object from the vehicle's multi-dimensional signals.

[0022] One possible implementation involves vehicle multidimensional signals including: vehicle dynamics signals, vehicle bus signals, and control signals from the vehicle controller. Test defects include at least one of the following: a first type of test function point that violates the physical constraints of the domain knowledge base of the target test object; a second type of test function point with potential risks; and a test coverage blind spot in the target test signal combination. The target test signal combination is a test signal combination whose test coverage is lower than a first preset test coverage threshold.

[0023] One possible implementation is that the natural language description text corresponding to the aforementioned generated time-series signal stream can be specifically implemented as follows: based on the test function points of the basic test cases, determine the associated test signals of the test function points, determine the indicator characteristics of the statistical indicators of the associated test signals, and determine the natural language description text based on the indicator characteristics and the domain knowledge base.

[0024] One possible implementation involves a basic test case comprising at least one test function point, and associated test signals comprising at least one test signal. A target test signal among the at least one test signal is associated with a target test function point among the at least one test function point. Specifically, the above-mentioned determination of test defects in basic test cases based on natural language description text can be implemented as follows: Based on the natural language description text, determine whether the target test function point satisfies a first condition. If the target test function point satisfies the first condition, determine that the target test function point is a first type of test function point. The first condition includes: the physical relationship between multiple associated test signals of the target test function point does not conform to the defined physical relationship of multiple associated test signals of the target test function point in the physical constraints; and the signal value of the target test signal exceeds the signal limit of the target test signal in the physical constraints.

[0025] One possible implementation of the above-mentioned test defect determination based on natural language description text for basic test cases can be specifically implemented as follows: determine whether the natural language description text indicates that the target scenario has occurred; if the natural language description text indicates that the target scenario has occurred, determine the test function point corresponding to the identified dangerous event as a second type of test function point. Here, the target scenario is a scenario that meets the dangerous triggering conditions associated with the identified dangerous event.

[0026] One possible implementation method for confirming a target test signal combination includes: determining the range of tested signal values ​​for any test signal combination based on natural language description text; determining the test coverage of any test signal combination based on the range of tested signal values ​​and the full range of signal values ​​for any test signal combination; and determining whether any test signal combination is a target test signal combination based on the test coverage of any test signal combination. Wherein, any test signal combination consists of at least one test signal from the associated test signals.

[0027] The technical effects of any implementation method in the second aspect can be found in the technical effects of any implementation method in the first aspect mentioned above, and will not be repeated here.

[0028] Thirdly, this application provides a test case generation method. The method is applied to the test module of the test case generation device in any of the above implementations. The method includes: generating basic test cases for the target test object on the vehicle; executing the basic test cases; and sending a first instruction to the test defect discovery module. Enhanced test cases are generated based on test defects fed back by the test defect discovery module, and the enhanced test cases are executed to compensate for the test defects.

[0029] In some embodiments, the above-mentioned generation of basic test cases for the target test object on the vehicle can be specifically implemented as follows: obtaining the basic test document and the professional document corresponding to the target test object, and generating basic test cases that do not violate physical constraints based on the basic test document, the professional document, and the domain knowledge base. The basic test document includes a test requirement document, a test case design specification document, and historical test case examples; the professional document includes the technical document of the target test object, the technical document of the components in the target test object, and the dynamic model of the dynamic system in which the target test object is located.

[0030] One possible implementation, based on the test foundation document, professional documents, and domain knowledge base, to generate basic test cases that do not violate physical constraints, can be specifically implemented as follows: Extract test items covered by the test requirements document; input the first text vector corresponding to the test item and the second text vector corresponding to the professional document into the retrieval model to obtain text vector segments in the professional document that are related to the test item and conform to physical constraints. Based on the text vector segments, the third text vector corresponding to the test foundation document, and the prompt word template text vector, basic test cases are generated. Here, the test item indicates the test function point and the corresponding execution preconditions, and the retrieval model is obtained by fine-tuning the large language model using the domain knowledge base.

[0031] One possible implementation, where enhanced test cases are generated based on test defects reported by the test defect mining module, and these enhanced test cases are executed to compensate for the test defects, can be specifically implemented as follows: Enhanced test cases are generated based on target information, and then executed. After executing the enhanced test cases, a first instruction is sent to the test defect mining module, causing the module to determine the test defects of the enhanced test cases in response to the first instruction, and to report the test defects of the enhanced test cases if the test defects do not meet the stop feedback conditions. The target information includes both the test defects and the test defects of the enhanced test cases.

[0032] One possible implementation includes stopping the feedback condition as follows: there is no target test signal combination, and no new first-type test function points and second-type test function points are added in consecutive rounds of testing of the predicted number.

[0033] The technical effects of any implementation method in the third aspect can be found in the technical effects of any implementation method in the first aspect mentioned above, and will not be repeated here.

[0034] Fourthly, embodiments of this application provide a computing device, which includes a processor and a memory. The memory stores at least one computer program, which is loaded and executed by the processor to implement the test case generation method of any implementation of the second or third aspect described above.

[0035] Fifthly, this application provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the test case generation method of any implementation of the second or third aspect described above.

[0036] Sixthly, this application provides a computer program product, which includes a computer program or instructions, and when the computer program or instructions are executed by a processor, implements a test case generation method that implements any of the second or third aspects described above.

[0037] The solutions provided in aspects four through six above can implement the test case generation method of any implementation method of aspect two or three above, and their specific implementations will not be described in detail here. The technical effects corresponding to any implementation method of the solutions provided in aspects four through six above can be referred to the technical effects corresponding to any implementation method of aspect one above, and will not be described in detail here.

[0038] It should be noted that any of the possible implementations of any of the above aspects can be combined, provided that the solutions do not contradict each other. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application will be described below.

[0040] Figure 1 This is a schematic diagram of the structure of a test case generation device disclosed in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a test module disclosed in an embodiment of this application; Figure 3 This is a flowchart illustrating a method for generating basic test cases disclosed in an embodiment of this application; Figure 4 This is a schematic diagram of a data acquisition process disclosed in an embodiment of this application; Figure 5This is a flowchart illustrating an enhanced test case generation method disclosed in an embodiment of this application; Figure 6 This is a flowchart illustrating a closed-loop iteration method for coverage disclosed in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a test defect mining module disclosed in an embodiment of this application; Figure 8 This is a flowchart illustrating a test case generation method disclosed in an embodiment of this application; Figure 9 This is a flowchart illustrating another test case generation method disclosed in an embodiment of this application; Figure 10 This is a schematic diagram of the structure of a computing device disclosed in an embodiment of this application.

[0041] Explanation of reference numerals in the attached figures: 100. Test module; 10. Test Execution Unit; 11. Test Case Generation Unit; 200. Test the defect detection module; 20. Data Acquisition Unit; 21. Natural Language Conversion Unit; 22. Test Defect Discovery Unit. Detailed Implementation

[0042] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0043] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0044] In the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.

[0045] To facilitate understanding, the terms used in the embodiments of this application will be explained first.

[0046] Test cases are sets of input conditions, execution conditions, expected results, and execution steps designed for a specific objective. Specifically, for vehicle testing scenarios, test cases are sets of specific operational steps, input data, and expected results designed to verify whether a vehicle or onboard system meets design requirements, safety standards, and user needs. For example, vehicle test cases include test items (such as braking from a stationary vehicle under adaptive cruise control), preconditions (such as vehicle battery status, gear status, weather conditions, and road conditions), operational steps (such as activating adaptive cruise control (ACC) and setting the following distance to level 2), expected results (such as smoothly stopping at a distance of 2.5 meters from the vehicle in front with a deceleration ≤ 3 m / s²), and judgment criteria (such as braking distance and braking response time).

[0047] Test defects: These are instances discovered during actual testing based on test cases where the system's actual performance deviates from the expected results in the test cases. For example: during ACC (Adaptive Cruise Control) following another vehicle, if the vehicle in front is stationary, the vehicle fails to trigger the Automatic Emergency Braking (AEB) system, resulting in a direct collision. Other examples include malfunctions in the vehicle's chassis system, such as abnormal steer-by-wire or steer-by-wire actuation.

[0048] Large language models (LLMs) are artificial intelligence neural network models trained on massive amounts of text data, possessing a huge number of parameters (typically reaching billions or even trillions). They are capable of reasoning and learning in conjunction with context, understanding and generating natural language descriptive text. Natural language descriptive text is written text using languages ​​used in everyday human communication (such as Chinese and English) to describe a thing, process, state, or concept. For example, "This car runs very fast" is natural language descriptive text, while the program code "speed>100" is non-natural language. Compared to non-natural language, natural language descriptive text is more expressive and can describe abstract concepts, emotions, and complex logic.

[0049] In related technologies, large language models are used in conjunction with static requirements documents, design specifications, and historical test case libraries to directly generate test cases and execution scripts based on text semantics. However, relying on static text knowledge to generate test cases makes it difficult for large language models to understand the actual physical state of the vehicle chassis domain control system's execution process based on massive amounts of multidimensional signals. Consequently, they cannot accurately identify test defects (such as substandard performance or potential anomalies) based on actual test data, and thus cannot meet the test coverage and reliability requirements of complex vehicle systems.

[0050] Therefore, embodiments of this application provide a test case generation device and method with strong self-optimization capabilities, which can improve the accuracy of test defects and the coverage of test cases for complex vehicle operating conditions. The technical solution provided in this application can be applied to testing fields such as bench testing, closed-site testing, and open road testing. The test objects include vehicle electronic-controlled air suspension (ECAS) and electronic-continuous damping control (ECDC) systems, etc., and this application does not limit the application to these categories.

[0051] The embodiments of this application are described below with reference to the accompanying drawings.

[0052] For example, please refer to Figure 1 , Figure 1 This is a schematic diagram of the structure of a test case generation device disclosed in an embodiment of this application. The test case generation device provided in this embodiment includes a test module 100 and a test defect discovery module 200, which are communicatively connected.

[0053] The test module 100 is used to generate basic test cases for the target test object on the vehicle, execute the basic test cases, and send a first instruction to the test defect mining module 200. It is also used to generate enhanced test cases based on the test defects fed back by the test defect mining module 200, and execute the enhanced test cases to make up for the test defects.

[0054] The target test signal combination is the test signal combination with a test coverage lower than the first preset test coverage threshold. Test coverage represents the proportion and depth of test requirements, functional points, code paths, input combinations, or system states covered by executed test cases. Specifically, the test signal combination with a test coverage lower than the first preset test coverage threshold is a set of test signals that are rarely triggered, never triggered, or triggered but not thoroughly verified in the existing test case library. These typically correspond to blind spots, corner cases, or potentially high-risk states in the vehicle system's operation.

[0055] For example, the test defect mining unit 22 determines the range of tested signal values ​​for any combination of test signals based on natural language description text. Based on the range of tested signal values ​​and the total range of signal values ​​for any combination of test signals, the test coverage of any combination of test signals is determined. Based on the test coverage of any combination of test signals, it is determined whether any combination of test signals is a target combination of test signals.

[0056] For example, the range of tested signal values ​​and their combinations with operating conditions are statistically analyzed and compared with the full parameter space (the full range of signal values). This generates a test coverage report, identifies parameter blind spots, and determines the target test signal combination. For instance, the range of tested signal values ​​(e.g., measured vehicle speed 40-100km / h, normal / low gear vehicle height) is compared with the full parameter space (e.g., vehicle speed 0-200km / h, normal / low / high / extremely high vehicle height). This generates a test coverage report, identifies the "high speed + extremely high vehicle height" range as a parameter blind spot, and determines the test signal combination corresponding to "high speed + extremely high vehicle height" as the target test signal combination.

[0057] Each test signal combination consists of at least one of the associated test signals, and the associated test signals are determined based on the test function points of the basic test cases.

[0058] For example, the basic test case includes at least one test function point, the associated test signal includes at least one test signal, and the target test signal in the at least one test signal is associated with the target test function point in the at least one test function point.

[0059] In some embodiments, such as Figure 2 As shown, the test module 100 includes a test case generation unit 11 and a test execution unit 10, and the test case generation unit 11 and the test execution unit 10 are communicatively connected.

[0060] The test case generation unit 11 is used to obtain the basic test document and the professional document corresponding to the target test object, and generate basic test cases that do not violate physical constraints based on the basic test document, the professional document, and the domain knowledge base. It then sends the basic test cases to the test execution unit 10 so that the test execution unit 10 can execute the basic test cases.

[0061] The testing foundation documents include test requirements documents, test case design specifications documents, and historical test case examples. The specialized documents include technical documents of the target test object, technical documents of the components within the target test object, and the dynamic model of the dynamic system in which the target test object resides. Examples include the vehicle chassis controlled object and the dynamic model of the vehicle, as well as technical documents of its components.

[0062] For example, for a vehicle's active suspension system, the basic testing documents include: ECAS / ECDC functional requirements specifications (such as "vehicle height adjustment function" and "automatic damping adjustment on continuous bumpy roads"), test case design specifications, and a historical reference test case library. The chassis-related professional documents include vehicle dynamics model descriptions (including the air spring stiffness-height nonlinear mapping formula and damping force-velocity characteristic curves), controlled object model documents (such as the maximum air compressor flow rate limit and solenoid valve response delay time), and component technical documents (such as the height sensor travel range of 0-200mm and the air tank safety pressure relief threshold of 1.2MPa). These documents contain the air pressure, fluid dynamics, and mechanical motion physical laws and constraints that the vehicle chassis must adhere to.

[0063] For example, the test case generation unit 11 extracts the test items covered by the test requirements document, and inputs the first text vector corresponding to the test item and the second text vector corresponding to the professional document into the retrieval model to obtain text vector segments in the professional document that are related to the test item and conform to physical constraints. Based on the text vector segments, the third text vector corresponding to the test base document, and the prompt word template text vector, basic test cases are generated.

[0064] Among them, the test items are used to indicate the test function points and the corresponding execution preconditions of the test function points, and the retrieval model is obtained by fine-tuning the large language model using a domain knowledge base.

[0065] For example, such as Figure 3 As shown, the method for generating basic test cases includes the following steps: Step S301: Extract the functional points and prerequisites from the active suspension requirements document.

[0066] For example: Extract the function point "Automatically lower the vehicle to a lower gear when driving at high speed (vehicle speed > 120km / h)".

[0067] Step S302: Vectorize the chassis professional documents and store them in the vector database.

[0068] For example, the retrieval model retrieves physical constraints such as "upper limit of compressor charging pressure build-up rate (e.g., 0.1 MPa / s)", "maximum height adjustment rate (e.g., 15 mm / s)" and "minimum safe air pressure of air spring" from the vector database, and obtains text vector segments in professional documents that are related to the test items and conform to the physical constraints.

[0069] Step S303: Combine the requirement text with the text vectors that meet the physical constraints retrieved from the vector database using the prompt template text vector to generate basic test cases. For example, the prompt may stipulate that: "When outputting the expected body height change curve, the descent rate shall not exceed 15mm / s, and the exhaust valve opening time must meet the flow formula constraint."

[0070] For example, the test execution unit 10 executes the generated basic test cases on a simulation platform or ring test bench (hardware), simultaneously collecting multi-dimensional dynamic data streams during execution. For instance, such as... Figure 4 As shown, during the test execution, vehicle dynamics signals (such as vehicle vertical acceleration, vehicle pitch rate, vehicle roll rate, suspension dynamic deflection, etc.), vehicle bus signals (such as vehicle speed, current vehicle height, driving mode (comfort / sport), and continuous damping control (CDC) damping request current value on the controller area network with flexible data-rate (CANFD) network), and controller internal signals (such as air compressor operating temperature, solenoid valve opening and closing duty cycle, and real-time air tank pressure obtained through the universal measurement and calibration protocol (XCP) debugging interface).

[0071] In some embodiments, the test case generation unit 11 further generates enhanced test cases based on the target information and sends the enhanced test cases to the test execution unit 10 so that the test execution unit 10 executes the enhanced test cases.

[0072] The target information includes test defects and test defects that enhance test cases.

[0073] Optionally, test defects include first-type test function points that violate the physical constraints in the domain knowledge base of the target test object, second-type test function points that have potential risks, and test coverage blind spots of the target test signal combination, etc.

[0074] For example, such as Figure 5As shown, when the test defect includes the first type of test function points, test cases that conform to the physical constraints in the domain knowledge base of the target test object are generated through large language model reasoning, resulting in enhanced test cases. For example, for the abnormal situation where the exhaust valve is fully open but the height does not decrease, a fault-tolerant test case of "simulating excessive back pressure due to exhaust valve pipeline blockage" is generated. When the test defect includes the second type of test function points, the potential risks corresponding to the second type of test function points are transformed into new test requirements, and test cases targeting the potential risks are generated based on the new test requirements, resulting in enhanced test cases. For example, for "compressor thermal management defects," a set of boundary test cases designed to verify "the height maintenance and degradation strategy after compressor shutdown under continuous turbulent conditions" is automatically generated. When the test defect includes test coverage blind spots of the target test signal combination, exploratory test cases are generated in blind spots with low test coverage (such as below the second preset test coverage threshold) based on reinforcement learning strategies, resulting in enhanced test cases. For example, for parameter blind spots with low test coverage, a reinforcement learning strategy (using coverage improvement as a reward function) is used to generate exploratory test cases in the insufficiently covered "high speed + extremely high body" range (testing the risk of deterioration in handling stability caused by excessively low air spring stiffness under this condition), and outputs enhanced test cases.

[0075] In some embodiments, after executing the enhanced test case, the test execution unit 10 sends a first instruction to the test defect discovery module 200. This causes the test defect discovery module 200 to respond to the first instruction by acquiring the timing signal stream of the target signal during the execution of the enhanced test case, and determining the test defect of the enhanced test case based on the timing signal stream of the target signal during the execution of the enhanced test case. If the test defect of the enhanced test case does not meet the stop feedback condition, the test defect of the enhanced test case is fed back to the test case generation unit.

[0076] The stopping feedback conditions include: there is no target test signal combination, and no new first-type test function points and second-type test function points are added in consecutive rounds of testing of the predicted number.

[0077] For example, test execution unit 10 executes enhanced test cases, re-acquiring, analyzing, and enhancing signals until the coverage report shows that the full parameter space coverage reaches the first preset test coverage threshold (e.g., 90%), and there are no new physical constraint violations or potential risk points in two consecutive iterations, achieving continuous closed-loop evolution of test case quality and coverage. For example: Figure 6 As shown, the closed-loop iterative method for coverage includes: Step S601: Generate enhanced test cases.

[0078] For example, new enhanced test cases are generated based on test defects in the base test cases or the previous enhanced test cases.

[0079] Step S602: Execute the enhanced test cases.

[0080] Step S603: Generate a coverage report.

[0081] Step S604: Is the stop feedback condition met? If yes, proceed to step S605; otherwise, proceed to step S601.

[0082] Step S605: End.

[0083] The target signal is the signal related to the performance of the target test object from the vehicle's multi-dimensional signals. Optionally, the vehicle multi-dimensional signals include: vehicle dynamics signals, vehicle bus signals, and control signals from the vehicle controller.

[0084] The test defect mining module 200 is used to collect the timing signal stream of the target signal during the execution of the basic test case in response to the first instruction, generate the natural language description text corresponding to the timing signal stream, determine the test defects of the basic test case based on the natural language description text, and feed the test defects back to the test module 100.

[0085] In some embodiments, such as Figure 7 As shown, the test defect mining module 200 includes a data acquisition unit 20, a natural language conversion unit 21, and a test defect mining unit 22. The natural language conversion unit 21 is communicatively connected to the data acquisition unit 20 and the test defect mining unit 22, respectively.

[0086] The acquisition unit 20 is used to acquire the timing signal stream in response to the first instruction and send the timing signal stream to the natural language conversion unit.

[0087] Natural Language Conversion Unit 21 is used to determine the associated test signals of the test function points based on the basic test cases, determine the indicator characteristics of the statistical indicators of the associated test signals, determine the natural language description text based on the indicator characteristics and the domain knowledge base, and send the natural language description text to Test Defect Mining Unit 22. For example, feature extraction and slicing of high-frequency time series signals, and mapping the indicator characteristics to natural language description text based on the knowledge base of the vehicle chassis domain.

[0088] Specifically, the natural language conversion unit 21 preprocesses the acquired real-time data stream (e.g., denoising and timestamp alignment), then performs condition slicing and feature extraction (e.g., extracting steady-state mean, overshoot, and oscillation decay rate) to obtain indicator features. Semantic mapping is then performed on these indicator features based on a chassis domain knowledge base. For example, if the extracted signal feature is "After the height control command is issued, the vehicle height remains unchanged and the air tank pressure continues to drop," it is mapped to the natural language description: "An abnormal operating condition where the air supply system has a serious leak, preventing the vehicle from being raised."

[0089] The test defect mining unit 22 is used to determine the test defects of the basic test cases based on natural language description text, and to feed the test defects back to the test module 100.

[0090] As an example, the test defect mining unit 22 determines whether the target test function point meets the first condition based on the natural language description text. If the target test function point meets the first condition, it determines that the target test function point is a first type of test function point.

[0091] The first condition includes: the physical relationship between multiple associated test signals of the target test function point does not conform to the physical constraints that define the physical relationship between multiple associated test signals of the target test function point, and the signal value of the target test signal exceeds the signal limit of the target test signal in the physical constraints.

[0092] For example, based on a professional knowledge base in the field of vehicle chassis, it is used to verify whether the relationship between multiple target test function points (measured signals) conforms to the basic physical equations, or to verify whether the control commands exceed the actuator limits. If the relationship between multiple target test function points does not conform to the basic physical equations, or the control commands exceed the actuator limits, the target test function point is determined to be a first-type test function point. For example, verifying whether the continuous operating time of the compressor exceeds the limit specified by the overheat protection model, or verifying whether the CDC shock absorber requests saturated current on a continuously undulating road surface, but the actual damping force does not change accordingly according to the valve flow equation.

[0093] As an example, the test defect mining unit 22 determines whether the natural language description text indicates that the target scenario has occurred. If the natural language description text indicates that the target scenario has occurred, the test function point corresponding to the identified hazardous event is identified as a second-type test function point. For example, by analyzing the natural language description text through a large language model, the test function point corresponding to the identified hazardous event that has shown potential defects or anomalies is identified as a second-type test. Specifically, in the "continuous pothole road medium speed driving" test, it was found that the ECAS system frequently triggered inflation and deflation to maintain vehicle height, causing the compressor to overheat and shut down, which was identified as a potential risk point of "compressor thermal management logic defect under extreme continuous excitation".

[0094] The target scenario is a scenario that meets the danger triggering conditions associated with the calibrated dangerous event.

[0095] For example, please refer to Figure 8 , Figure 8 This is a flowchart illustrating a test case generation method disclosed in an embodiment of this application. The method is applied to the test defect mining module in the aforementioned test case generation device, or to other devices capable of performing test defect mining.

[0096] Step S801: In response to the first instruction sent by the test module, collect the timing signal stream of the target signal during the execution of the basic test cases.

[0097] The target signal is the signal in the vehicle's multidimensional signal that is related to the performance of the target test object.

[0098] Optionally, the vehicle multidimensional signals include: vehicle dynamics signals, vehicle bus signals, and control signals from the vehicle controller.

[0099] The target test signal combination is a test signal combination whose test coverage is lower than the first preset test coverage threshold.

[0100] For example, the method for confirming a target test signal combination includes: determining the range of tested signal values ​​for any test signal combination based on natural language description text; determining the test coverage of any test signal combination based on the range of tested signal values ​​and the full range of signal values ​​for any test signal combination; and determining whether any test signal combination is a target test signal combination based on the test coverage of any test signal combination.

[0101] Each test signal combination consists of at least one test signal from the associated test signals.

[0102] Step S802: Generate natural language description text corresponding to the time-series signal stream.

[0103] For example, based on the test function points of the basic test cases, the associated test signals of the test function points are determined, the indicator characteristics of the statistical indicators of the associated test signals are determined, and the natural language description text is determined based on the indicator characteristics and the domain knowledge base.

[0104] Step S803: Determine the test defects of the basic test cases based on the natural language description text, and feed the test defects back to the test module so that the test module can generate enhanced test cases based on the test defects and execute the enhanced test cases to make up for the test defects.

[0105] Optionally, test defects include: first-type test function points that violate the physical constraints in the domain knowledge base of the target test object, second-type test function points that have potential risks, and test coverage blind spots of the target test signal combination, etc.

[0106] In some embodiments, the basic test case includes at least one test function point, the associated test signal includes at least one test signal, and the target test signal in the at least one test signal is associated with the target test function point in the at least one test function point.

[0107] As an example, the system determines whether the target test function point meets the first condition based on the natural language description text. If the target test function point meets the first condition, it is determined that the target test function point is a first type of test function point.

[0108] The first condition includes: the physical relationship between multiple associated test signals of the target test function point does not conform to the physical constraints that define the physical relationship between multiple associated test signals of the target test function point, and the signal value of the target test signal exceeds the signal limit of the target test signal in the physical constraints.

[0109] As another example, determine whether the natural language description text indicates that the target scenario has occurred. If the natural language description text indicates that the target scenario has occurred, determine the test function point corresponding to the hazard event as the second type of test function point.

[0110] The target scenario is a scenario that meets the danger triggering conditions associated with the calibrated dangerous event.

[0111] For example, please refer to the figure. Figure 9 This is a flowchart illustrating another test case generation method disclosed in an embodiment of this application. The method is applied to the test module in the test case generation device described above, or to other devices capable of generating test cases.

[0112] Step S901: Generate basic test cases for the target test object on the vehicle, execute the basic test cases, and send the first instruction to the test defect discovery module.

[0113] For example, the test base document and the professional document corresponding to the target test object are obtained, and basic test cases that do not violate physical constraints are generated based on the test base document, professional document and domain knowledge base.

[0114] The basic testing documents include test requirements documents, test case design specifications documents, and historical test case examples. The professional documents include technical documents of the target test object, technical documents of the components in the target test object, and the dynamic model of the dynamic system in which the target test object is located.

[0115] Specifically, test items covered by the test requirements document are extracted. The first text vector corresponding to the test item and the second text vector corresponding to the professional document are input into the retrieval model to obtain text vector segments in the professional document that are related to the test item and conform to physical constraints. Based on the text vector segments, the third text vector corresponding to the test base document, and the prompt word template text vector, basic test cases are generated.

[0116] Among them, the test items are used to indicate the test function points and the corresponding execution preconditions of the test function points, and the retrieval model is obtained by fine-tuning the large language model using a domain knowledge base.

[0117] Step S902: Generate enhanced test cases based on the test defects fed back by the test defect mining module, and execute the enhanced test cases to make up for the test defects.

[0118] For example, enhanced test cases are generated based on target information, and then executed. After executing the enhanced test cases, a first instruction is sent to the test defect discovery module, so that the test defect discovery module determines the test defects of the enhanced test cases in response to the first instruction, and provides feedback on the test defects of the enhanced test cases if the test defects of the enhanced test cases do not meet the stop feedback conditions.

[0119] The target information includes test defects and test defects that enhance test cases.

[0120] The conditions for stopping feedback include: there is no target test signal combination, and no new first-type test function points and second-type test function points are added in consecutive rounds of testing of the predicted number.

[0121] The foregoing mainly describes the solutions provided by the embodiments of this application from the perspective of apparatus and methods. To achieve the above functions, test case generation includes the corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0122] This application embodiment can exemplarily divide the test case generation module according to the above-described vehicle test case generation method. For example, the test case generation device may include functional modules corresponding to each functional division, or two or more functions may be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, there may be other division methods.

[0123] This application also provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the test case generation method provided in the above-described method embodiments.

[0124] Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as a read-only memory (ROM), random access memory (RAM), magnetic tape, floppy disk, and optical data storage device.

[0125] This application also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, they implement the test case generation method provided in the above-described method embodiments.

[0126] It should be noted that when one or more instructions in the computer-readable storage medium or computer program product are executed by the processor of a computing device, they implement the various processes of the above-described method embodiments and achieve the same technical effects as the above-described methods. To avoid repetition, they will not be described again here.

[0127] This application also provides a computing device. Based on the aforementioned test case generation method, this application exemplarily divides the computing device into functional modules. For example, the computing device may include functional modules corresponding to each functional division, or two or more functions may be integrated into a single processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application is illustrative and represents only one logical functional division; in actual implementation, other division methods may be used.

[0128] like Figure 10As shown in the illustration, a computing device disclosed in this application includes a processor 1001, a bus 1002, a communication interface 1003, and a memory 1004. The processor 1001, the memory 1004, and the communication interface 1003 communicate with each other via the bus 1002. It should be understood that this application does not limit the number of processors and memories in a network device.

[0129] Bus 1002 can be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or a Universal Serial Bus (USB), etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 10 The bus 1002 may be represented by a single line, but this does not mean that there is only one bus or one type of bus. The bus 1002 may include a path for transmitting information between various components of the network device (e.g., memory 1004, processor 1001, communication interface 1003).

[0130] The processor 1001 may include any one or more processors such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0131] The memory 1004 may include volatile memory, such as random access memory (RAM). The processor 1001 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0132] The communication interface 1003 uses transceiver modules, such as, but not limited to, network interface cards and transceivers, to enable communication between the computing device and other devices or communication networks.

[0133] The memory 1004 stores executable program code, and the processor 1001 executes the executable program code to implement the functions of the aforementioned method embodiments. That is, the memory 1004 stores instructions for executing the above-described test case generation method.

[0134] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0136] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0137] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0138] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Those skilled in the art can understand that implementing all or part of the processes of the above embodiments and making equivalent changes according to the claims of this application still fall within the scope of this application.

Claims

1. A test case generation device, characterized in that, The test case generation device includes: a test module and a test defect discovery module; the test module and the test defect discovery module are communicatively connected. The testing module is used to generate basic test cases for the target test object on the vehicle, execute the basic test cases, and send a first instruction to the test defect discovery module. The test defect mining module is used to respond to the first instruction to collect the time-series signal stream of the target signal during the execution of the basic test case; wherein, the target signal is a signal related to the performance of the target test object in the vehicle multi-dimensional signal; generate natural language description text corresponding to the time-series signal stream; determine the test defect of the basic test case based on the natural language description text, and feed the test defect back to the test module; The testing module is further configured to generate enhanced test cases based on the test defects; and execute the enhanced test cases to compensate for the test defects.

2. The test case generation device according to claim 1, characterized in that, The vehicle multidimensional signals include: vehicle dynamics signals, vehicle bus signals, and control signals of the vehicle controller; The test defect includes at least one of the following: The first type of test function point that violates the physical constraints in the domain knowledge base of the target test object; The second type of test function point has potential risks; The test coverage blind zone of the target test signal combination; wherein, the target test signal combination is a test signal combination with a test coverage lower than a first preset test coverage threshold.

3. The test case generation device according to claim 2, characterized in that, The test defect mining module includes: a data acquisition unit, a natural language conversion unit, and a test defect mining unit; the natural language conversion unit is communicatively connected to both the data acquisition unit and the test defect mining unit. The acquisition unit is configured to, in response to the first instruction, acquire the time-series signal stream and send the time-series signal stream to the natural language conversion unit; The natural language conversion unit is used to determine the associated test signals of the test function points based on the test function points of the basic test cases; determine the indicator characteristics of the statistical indicators of the associated test signals; determine the natural language description text based on the indicator characteristics and the domain knowledge base; and send the natural language description text to the test defect mining unit. The test defect mining unit is used to determine the test defects of the basic test cases based on the natural language description text, and to feed back the test defects to the test module.

4. The test case generation device according to claim 3, characterized in that, The basic test cases include at least one test function point; the associated test signals include at least one test signal; the target test signal in the at least one test signal is associated with the target test function point in the at least one test function point; the test defect mining unit is specifically used for: Based on the natural language description text, determine whether the target test function point meets the first condition; If the target test function point satisfies the first condition, the target test function point is determined to be the first type of test function point; The first condition includes: The physical relationship between the multiple associated test signals of the target test function point does not conform to the defined physical relationship of the multiple associated test signals of the target test function point in the physical constraints. The signal value of the target test signal exceeds the signal limit of the target test signal in the physical constraints.

5. The test case generation device according to claim 3, characterized in that, The test defect mining unit is specifically used for: Determine whether the natural language description text indicates that a target scenario has occurred; wherein, the target scenario is a scenario that meets the danger triggering conditions associated with the identified dangerous event; When the natural language description text indicates that the target scenario has occurred, the test function point corresponding to the calibrated dangerous event is determined to be the second type of test function point.

6. The test case generation device according to claim 3, characterized in that, The test defect mining unit is specifically used for: Based on the natural language description text, the range of tested signal values ​​for any combination of test signals is determined; wherein, any combination of test signals consists of at least one test signal from the associated test signals; Based on the range of values ​​of the tested signals and the full range of values ​​of any combination of test signals, determine the test coverage of any combination of test signals. Based on the test coverage of any test signal combination, determine whether any test signal combination is the target test signal combination.

7. The test case generation apparatus according to any one of claims 2-6, characterized in that, The testing module includes a test case generation unit and a test execution unit; the test case generation unit and the test execution unit are communicatively connected. The test case generation unit is used to obtain the basic test document and the professional document corresponding to the target test object; Based on the aforementioned basic test document, the aforementioned professional document, and the aforementioned domain knowledge base, generate the aforementioned basic test cases that do not violate the aforementioned physical constraints; The basic test cases are sent to the test execution unit so that the test execution unit executes the basic test cases. The basic testing documents include test requirements documents, test case design specifications documents, and historical test case examples; the professional documents include technical documents of the target test object, technical documents of the components in the target test object, and the dynamic model of the dynamic system in which the target test object is located.

8. The test case generation device according to claim 7, characterized in that, The use case generation unit is specifically used for: Extract the test items covered by the test requirements document; wherein, the test items are used to indicate the test function points and the corresponding execution preconditions; The first text vector corresponding to the test item and the second text vector corresponding to the professional document are input into the retrieval model to obtain the text vector segments in the professional document that are related to the test item and meet the physical constraints; wherein, the retrieval model is obtained by fine-tuning the domain knowledge base based on the large language model; Based on the text vector segmentation, the third text vector corresponding to the test base document, and the prompt word template text vector, the basic test cases are generated.

9. The test case generation device according to claim 7, characterized in that, The use case generation unit is further configured to: Generate enhanced test cases based on target information; and send the enhanced test cases to the test execution unit so that the test execution unit executes the enhanced test cases; The test execution unit is further configured to: send the first instruction to the test defect mining module after executing the enhanced test case; the test defect mining module is further configured to: in response to the first instruction, collect the timing signal stream of the target signal during the execution of the enhanced test case; determine the test defect of the enhanced test case based on the timing signal stream of the target signal during the execution of the enhanced test case; and, if the test defect of the enhanced test case does not meet the stop feedback condition, feed back the test defect of the enhanced test case to the test case generation unit; The target information includes: the test defect and the test defect of the enhanced test case.

10. The test case generation device according to claim 9, characterized in that, The stop feedback conditions include: The target test signal combination does not exist, and no new test function points of the first type and the second type are added in the consecutive rounds of testing of the predicted number.

11. A test case generation method, characterized in that, A test defect mining module is applied in a test case generation device, the test case generation device including a test module and a test defect mining module, the test module being used to generate and execute basic test cases for target test objects on vehicles; The testing module and the test defect mining module are communicatively connected; The test case generation method includes: In response to the first instruction sent by the test module, the timing signal stream of the target signal during the execution of the basic test case is collected; wherein, the target signal is a signal related to the performance of the target test object in the vehicle multidimensional signal; Generate natural language description text corresponding to the time-series signal stream; Based on the natural language description text, the test defects of the basic test cases are determined, and the test defects are fed back to the test module, so that the test module generates enhanced test cases based on the test defects and executes the enhanced test cases to make up for the test defects.

12. A computing device, characterized in that, It includes a processor and a memory, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the test case generation method as described in claim 11.

Images