Functional security auxiliary authentication method and system based on large language model

Abstract

The invention discloses a functional security auxiliary authentication method and system based on a large language model. Belongs to the technical field of function security authentication. The method comprises the steps that an input document is obtained, the input document comprises a demand document and a design document corresponding to a to-be-authenticated system, the input document comprises key information manually marked in advance, and the key information at least comprises security demand chapters and interface definitions; the input document is analyzed based on a pre-trained large language model and a pre-constructed knowledge base, an analysis result is obtained, and the analysis result represents whether the to-be-authenticated system can correctly process the risk of function failure and fault or not; triggering a manual auditing interface to enable an authenticator to correct the risk level and the failure mode description; the big language model optimizes the analysis result according to manual feedback. According to the invention, whether the system can correctly process the risk of potential function failure and fault can be automatically judged, manual authentication is assisted, and the authentication efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of functional safety certification technology, specifically relating to a functional safety auxiliary certification method and system based on a large language model. Background Technology

[0002] In critical fields such as industrial automation, aerospace, automotive electronics, and intelligent driving, functional safety is a core technology that ensures that systems or equipment can respond correctly in a predictable manner when failures occur. International standard IEC 61508, as well as ISO 26262 for intelligent driving and IEC 61511 for process industries, have standardized functional safety certification, requiring systems to have the ability to handle errors, hardware failures, and operational stress.

[0003] In the early stages, functional safety certification involved small-scale systems with a limited number of faults, and manual operation could meet the certification requirements. However, with the rapid iteration of cutting-edge technologies such as autonomous driving, system complexity has increased exponentially, and the traditional manual certification model has gradually revealed its drawbacks. On the one hand, manual verification is extremely inefficient. Taking Failure Mode and Effects Analysis (FMEA) of ISO 26262 as an example, manually building a traceability matrix containing thousands of requirements is time-consuming and inefficient. On the other hand, certification work requires a high level of interdisciplinary knowledge, while domain experts are scarce. According to relevant reports, some companies rely heavily on third parties for functional safety certification and lack independent certification capabilities.

[0004] While large language models have achieved remarkable results in natural language processing and knowledge reasoning, their application in functional safety certification is extremely limited. Current technology has not yet explored the use of large language models to analyze functional safety data, and there is a lack of systems supporting automated certification, making it difficult to meet the industry's urgent need for efficient and intelligent certification tools. Therefore, developing functional safety-assisted certification methods and systems based on large language models has become a key direction for breaking through the bottlenecks of traditional certification and improving certification efficiency and accuracy. Summary of the Invention

[0005] One objective of this invention is to provide a functional safety-assisted authentication method and system based on a large language model, which can solve the aforementioned technical problems in the prior art.

[0006] According to a first aspect of the present invention, a functional safety-assisted authentication method based on a large language model is provided, comprising:

[0007] Obtain input documents, which include the requirements document and design document corresponding to the system to be certified. The input documents contain key information that has been manually annotated in advance, and the key information includes at least the security requirements section and interface definition.

[0008] The input document is analyzed based on a pre-trained large language model and a pre-built knowledge base to obtain the analysis results. The analysis results indicate whether the system to be certified can correctly handle the risk of functional failure and malfunction.

[0009] The manual review interface is triggered, allowing certification personnel to revise the risk level and failure mode description.

[0010] The large language model is optimized based on human feedback and analysis results.

[0011] Optionally, the knowledge base construction process includes:

[0012] Extract the normative clauses from the industry standards corresponding to the system to be certified;

[0013] Import publicly available industry-specific failure cases corresponding to the aforementioned standard clauses for the system to be certified;

[0014] Integrate the manufacturer-defined design standards of the systems to be certified.

[0015] Optionally, the analysis of the input document based on the pre-trained large language model and the pre-built knowledge base yields analysis results, including:

[0016] After the large language model identifies multiple requirements in the requirements document, it triggers a knowledge base retrieval based on the multiple requirements to query the specification clauses in the knowledge base corresponding to each requirement.

[0017] Determine whether each requirement matches the corresponding specification clause, and obtain the analysis results for each requirement.

[0018] Optionally, the method further includes:

[0019] If it is determined that the requirements do not match the corresponding specifications, a risk analysis should be conducted based on the corresponding failure cases.

[0020] The verification process corresponding to the requirements in the custom design standard is invoked to generate a risk report.

[0021] Optionally, the large language model optimizes the analysis results based on human feedback, including:

[0022] New keywords are provided to the large language model based on human feedback.

[0023] The large language model re-analyzes the input document based on the new keywords to obtain optimized analysis results.

[0024] Optionally, after obtaining the analysis results, the method further includes:

[0025] Based on the model analysis results and review comments, generate a functional safety analysis report with confidence level annotations;

[0026] Submit the functional safety analysis report to professionals for revision and record the revision history;

[0027] The final revised functional safety analysis report will be submitted to professionals for review before being made available to users.

[0028] According to a second aspect of the present invention, a system for applying the functional safety-assisted authentication method based on a large language model as described in the first aspect of the present invention is provided, including an authentication assistance module;

[0029] The authentication assistance module is used to obtain input documents, which include the requirements document and design document corresponding to the system to be authenticated. The input documents contain key information that has been manually annotated in advance, and the key information includes at least the security requirements section and interface definition.

[0030] The authentication assistance module is also used to analyze the input document based on the pre-trained large language model and the pre-built knowledge base to obtain the analysis results. The analysis results indicate whether the system to be certified can correctly handle the risk of functional failure and malfunction. It triggers the manual review interface to allow the certifier to revise the risk level and failure mode description. The large language model optimizes the analysis results based on the manual feedback.

[0031] Optionally, it may also include a display module and a storage module;

[0032] The display module is used to output functional safety analysis reports and provide visual displays;

[0033] The storage module is used to store pre-trained model parameters, knowledge base data, and generated report data.

[0034] According to a third aspect of the present invention, an electronic device is provided, including a processor and a memory, wherein the memory stores a program or instructions executable by the processor, and the program or instructions, when executed by the processor, implement the steps of the functional safety-assisted authentication method based on a large language model as described in the first aspect of the present invention.

[0035] According to a fourth aspect of the present invention, a readable storage medium is provided, wherein a program or instructions are stored therein, which, when executed, implement the steps of the functional safety-assisted authentication method based on a large language model as described in the first aspect of the present invention.

[0036] The beneficial effects of this invention are as follows: This invention analyzes requirement documents and design documents using a large language model to automatically determine whether the system can correctly handle potential functional failures and malfunctions, assisting manual certification and improving certification efficiency. It utilizes external knowledge base technology to assist reasoning, improving the accuracy and reliability of functional safety analysis. The system supports method execution through the collaboration of multiple modules, ensuring the efficiency and stability of functional safety analysis and report generation. The method and system support multi-scenario industrial applications, reducing reliance on domain experts and empowering SMEs to achieve independent certification. Attached Figure Description

[0037] Figure 1 This is a flowchart of the functional safety-assisted authentication method based on a large language model in an embodiment of the present invention. Detailed Implementation

[0038] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0039] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0040] Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0041] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0042] In the specification of this invention, the terms "first" and "second" may explicitly or implicitly include one or more of the same feature. In the description of this invention, unless otherwise stated, "multiple" means two or more. Furthermore, in the specification, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0043] The purpose of this invention is to address the problems of high reliance on manual labor and low efficiency in existing functional safety certification processes by providing a large language model functional safety certification method and system. This method analyzes requirements documents and design documents using a large language model to determine the risk of potential functional failures and malfunctions and generate reports. At the same time, the system modules support the execution of the method, assisting manual certification and improving certification efficiency and accuracy.

[0044] like Figure 1 As shown in the figure, this embodiment introduces a functional safety-assisted authentication method based on a large language model, including steps 1100-1400.

[0045] Step 1100: Obtain the input document, which includes the requirements document and design document corresponding to the system to be certified. The input document contains key information that has been manually annotated in advance, and the key information includes at least the security requirements section and interface definition.

[0046] Step 1200: Analyze the input document based on the pre-trained large language model and the pre-built knowledge base to obtain the analysis results. The analysis results indicate whether the system to be certified can correctly handle the risk of functional failure and malfunction.

[0047] Step 1300: Trigger the manual review interface to allow certification personnel to revise the risk level and failure mode description.

[0048] Step 1400: The large language model is optimized based on human feedback analysis results.

[0049] Manually annotated key information is used to inform the large language model which content requires special attention. The input document contains multiple chapters, some of which require special attention; these chapters are manually annotated.

[0050] For example, in automotive electronics, the input documentation includes the ISO 26262 safety requirements specification, braking system control code, and CAN bus interface definition. In industrial automation, the input documentation includes PLC programming manuals and IEC 61508 SIL level requirements.

[0051] The system will automatically identify whether the document structure conforms to relevant standards. If it does not, it will prompt the user to manually supplement the missing content.

[0052] This invention is applicable to functional safety certification across various industries. The certification standards differ for different application scenarios, resulting in different knowledge bases. For example, in automotive electronics, the system utilizes automotive-specific knowledge bases, such as an ABS system failure case library, with a large language model focusing on analyzing the matching of braking failure modes (such as pedal signal loss) with ASIL levels. In industrial automation scenarios, it utilizes an industrial knowledge base containing PID controller failure modes to automate the assessment of safety integrity levels.

[0053] The analysis results provided by the large language model need further human verification. If there are any discrepancies between the results and the human verification, the large language model needs to re-analyze the input document to optimize the results. Alternatively, the inconsistent parts can be analyzed manually, and the inconsistent content can be optimized and updated to obtain the final result.

[0054] In this embodiment, the knowledge base construction process includes: extracting normative clauses from the industry standards corresponding to the system to be certified; importing publicly available failure cases corresponding to the normative clauses in the industry corresponding to the system to be certified; and integrating the design standards customized by the manufacturer of the system to be certified.

[0055] Industry standards typically regulate related products from multiple perspectives, resulting in various specifications and clauses. Different specifications and clauses may target different components of the product, such as mechanical structural parts, electrical components, or even control software.

[0056] Product failures can be caused by multiple factors, such as hardware or software malfunctions. The imported failure cases in the knowledge base will also include various examples, each corresponding to different failure scenarios.

[0057] Functional safety certification is specific to a particular manufacturer's actual product. Even for the same type of product, different manufacturers may have different standards. Therefore, to improve the applicability of this invention, custom design standards from various manufacturers have been added to the knowledge base.

[0058] Specifically, taking automotive braking system certification as an example, the knowledge base construction process is as follows: extracting standard clauses such as "the braking system must meet the ASIL-D level safety requirements" from the ISO 26262 standard; importing publicly available industry failure cases of "brake booster pump failure leading to abnormal pedal force"; and integrating the automaker's customized enterprise standard of "dual brake sensor redundancy design".

[0059] Meanwhile, this invention also supports manually adding enterprise privatization standards or customized cases, which complement the system's knowledge base.

[0060] In this embodiment, step 1200 includes: after the large language model identifies multiple requirements in the requirements document, it triggers a knowledge base retrieval based on the multiple requirements to query the normative clauses in the knowledge base corresponding to each requirement; it determines whether each requirement matches the corresponding normative clauses and obtains the analysis results corresponding to each requirement.

[0061] After the user inputs a document, the large language model parses the document, extracting relevant requirements. It then retrieves corresponding specification clauses from the knowledge base to determine if a match is found.

[0062] For example, when analyzing a braking system requirements document, the large language model recognizes the description "single power supply for brake pedal travel sensor," automatically triggering a knowledge base search to match the clause in ISO 26262 that "safety-critical sensors require dual power supply redundancy." Since ISO 26262 specifies the need for dual power supply redundancy, while the requirements document only specifies a single power supply for the brake pedal travel sensor, the two are incompatible.

[0063] If the large language model recognizes the description "two redundant power supplies for brake pedal travel sensor" and it conforms to the provisions of ISO26262 standard, then the matching result is a match.

[0064] If it is determined that the requirements do not match the corresponding specification clauses, a risk analysis is performed based on the corresponding failure cases; the verification process corresponding to the requirements in the custom design standard is invoked to generate a risk report.

[0065] For example, when the power supply method of the brake pedal travel sensor does not match the requirements in the ISO 26262 standard, the impact analysis of "single power supply failure leading to sensor signal interruption" in historical cases is referenced, the redundancy design verification process in the enterprise standard is invoked, a risk report of "power supply architecture does not meet ASIL-D requirements" is generated, and a dual power supply retrofit solution is recommended.

[0066] In this embodiment, step 1400 includes: providing new keywords to the large language model based on human feedback; the large language model re-analyzing the input document based on the new keywords to obtain optimized analysis results.

[0067] When the analysis results output by the large language model contain inconsistencies with the results of human review, new keywords are provided to the large language model. These new keywords indicate the inconsistencies, allowing the large language model to re-analyze these inconsistencies and obtain optimized analysis results.

[0068] In this embodiment, after obtaining the analysis results, the method further includes: generating a functional safety analysis report with confidence level annotation based on the model analysis results and review comments; submitting the functional safety analysis report to professionals for revision and recording the revision history; and submitting the final revised functional safety analysis report to professionals for confirmation before allowing users to view it.

[0069] This invention fully leverages the advantages of large language models and knowledge base technologies to automate the functional safety certification process, significantly improving certification efficiency and accuracy. The system supports the execution of the method through knowledge base modules and certification assistance modules, ensuring the stability and reliability of functional safety analysis and report generation. The method and system support multi-scenario industrial applications, reducing labor costs and reliance on domain experts, filling a gap in the application of artificial intelligence in the field of functional safety certification.

[0070] This embodiment describes a system for applying the functional safety-assisted authentication method based on a large language model according to any embodiment of the present invention, including an authentication assistance module;

[0071] The authentication assistance module is used to obtain input documents, which include the requirements document and design document corresponding to the system to be authenticated. The input documents contain key information that has been manually annotated in advance, and the key information includes at least the security requirements section and interface definition.

[0072] The authentication assistance module is also used to analyze the input document based on the pre-trained large language model and the pre-built knowledge base to obtain the analysis results. The analysis results indicate whether the system to be certified can correctly handle the risk of functional failure and malfunction. It triggers the manual review interface to allow the certifier to revise the risk level and failure mode description. The large language model optimizes the analysis results based on the manual feedback.

[0073] The system also includes a display module and a storage module;

[0074] The display module is used to output functional safety analysis reports and provide visual displays;

[0075] The storage module is used to store pre-trained model parameters, knowledge base data, and generated report data.

[0076] This embodiment introduces an electronic device, including a processor and a memory. The memory stores programs or instructions that can be executed by the processor. When the program or instructions are executed by the processor, they implement the steps of the functional safety-assisted authentication method based on a large language model as described in any embodiment of the present invention.

[0077] This embodiment introduces a readable storage medium that stores a program or instructions, which, when executed, implement the steps of the functional safety-assisted authentication method based on a large language model as described in any embodiment of the present invention.

[0078] While specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should understand that modifications can be made to the above embodiments without departing from the scope and spirit of the invention.

[0079] Those skilled in the art will recognize that the modules and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software 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 implementations should not be considered beyond the scope of this invention.

[0080] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described apparatus and equipment can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0081] In the 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 is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0082] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.

[0083] In addition, the functional modules in the embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0084] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. 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.

[0085] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

[0086] It should be understood that the sequence numbers of the steps in the invention's content and embodiments do not absolutely imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention. The foregoing description of embodiments of this disclosure has been provided for illustrative and descriptive purposes. The foregoing description is not exhaustive and is not intended to limit this disclosure to the exact form disclosed. Various modifications and variations may exist based on the foregoing teachings, or various modifications and variations may be derived from the practice of this disclosure. These embodiments were chosen and described to illustrate the principles of this disclosure and its practical application, so that those skilled in the art can utilize this disclosure in various implementations and modifications suitable for the specific purpose of the concept.

Claims

1. A functional safety assisted authentication method based on a large language model, characterized in that, The method comprises the following steps: obtaining an input document, wherein the input document comprises a requirement document and a design document corresponding to a system to be certified, and the input document contains pre-annotated key information, wherein the key information at least comprises a security requirement chapter and an interface definition; analyzing the input document based on a pre-trained large language model and a pre-constructed knowledge base to obtain an analysis result, wherein the analysis result indicates whether the system to be certified can correctly handle the risk of functional failure and fault; triggering an artificial review interface to enable a certification personnel to correct a risk level and a failure mode description; optimizing the analysis result by the large language model according to artificial feedback.

2. The method of claim 1, wherein, The knowledge base construction process comprises the following steps: extracting specification clauses from industry standards corresponding to the system to be certified; importing fault cases corresponding to the specification clauses from industry public sources corresponding to the system to be certified; integrating design standards defined by a manufacturer of the system to be certified.

3. The method of claim 2, wherein, The analysis of the input document based on the pre-trained large language model and the pre-constructed knowledge base to obtain the analysis result comprises the following steps: after the large language model identifies multiple requirement contents in the requirement document, triggering knowledge base retrieval according to the multiple requirement contents to query specification clauses corresponding to each requirement content in the knowledge base; judging whether each requirement content matches the corresponding specification clause to obtain an analysis result corresponding to each requirement content.

4. The method of claim 3, wherein, The method further comprises the following steps: if it is judged that the requirement content does not match the corresponding specification clause, performing risk analysis according to the corresponding fault case; calling a verification process corresponding to the requirement content in the self-defined design standard to generate a risk report.

5. The method of claim 1, wherein, The optimization of the analysis result by the large language model according to artificial feedback comprises the following steps: providing new keywords to the large language model according to the artificial feedback content; reanalyzing the input document by the large language model according to the new keywords to obtain an optimized analysis result.

6. The method of claim 1, wherein, After obtaining the analysis result, the method further comprises the following steps: generating a functional safety analysis report with a confidence level according to the model analysis result and the review opinion; submitting the functional safety analysis report to a professional for revision and recording the revision history; submitting the functional safety analysis report after final revision to a professional for confirmation and then providing the functional safety analysis report to a user for viewing.

7. A system applying the functional safety assisted authentication method based on a large language model according to any one of claims 1-6, characterized in that, The method comprises the following steps: obtaining an input document, wherein the input document comprises a requirement document and a design document corresponding to a system to be certified, and the input document contains pre-annotated key information, wherein the key information at least comprises a security requirement chapter and an interface definition; analyzing the input document based on a pre-trained large language model and a pre-constructed knowledge base to obtain an analysis result, wherein the analysis result indicates whether the system to be certified can correctly handle the risk of functional failure and fault; 8. The system of claim 7, wherein, triggering an artificial review interface to enable a certification personnel to correct a risk level and a failure mode description; optimizing the analysis result by the large language model according to artificial feedback. The method further comprises the following steps: displaying a functional safety analysis report and a visual display by a display module; and storing the functional safety analysis report by a storage module. The storage module is used for storing pre-training model parameters, knowledge base data and generated report data.

9. An electronic device, comprising: The processor and the memory are included, and the memory stores programs or instructions which can be executed by the processor, and the programs or instructions are executed by the processor to realize the steps of the function safety auxiliary authentication method based on the large language model in any one of claims 1-6.

10. A readable storage medium, characterized by, The readable storage medium stores programs or instructions, and the programs or instructions are executed to realize the steps of the function safety auxiliary authentication method based on the large language model in any one of claims 1-6.

Images