Automobile safety risk analysis method and electronic equipment

Through automated automobile safety risk analysis methods, using product design requirements description documents and target safety risk data, inputting safety risk analysis models is solved, and the problems of low efficiency, high cost and safety protection lag caused by manual analysis in the existing technology are achieved. Efficient, comprehensive and timely security risk treatment.

CN119089450APending Publication Date: 2024-12-06Z-ONE TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411192535.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The safety risk analysis methods of existing automotive business systems rely on manual analysis, resulting in low efficiency, high cost and lagging safety protection in the R&D and design process, and the risk problems cannot be handled in a timely manner.

Method used

By determining the product design requirements description documents and target safety risk data of the automotive business system, input the automobile safety risk analysis model, automatically analyze safety risks, provide safety risk analysis results and response measures, and realize the pre-design and protection of safety risks.

Benefits of technology

It improves the efficiency and comprehensiveness of safety risk analysis, reduces costs, promptly handles safety risks, avoids safety protection lag and secondary development and testing costs in the later stage of R&D, and improves the safety design capabilities of automotive business systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119089450A_ABST
    Figure CN119089450A_ABST
Patent Text Reader

Abstract

The invention discloses an automobile safety risk analysis method and electronic equipment. The method comprises the following steps: determining a product design demand description file corresponding to an automobile service system; according to the product design demand description file, determining target security risk data corresponding to the product design demand description file from security risk data included in a security risk database; and inputting the product design demand description file and the target safety risk data into the automobile safety risk analysis model to obtain an automobile safety risk analysis result. Thus, in the early stage of research and development of the automobile service system, the automobile safety risk analysis result can be obtained based on the product design demand description file, manual automobile safety risk analysis is not needed, and the automobile safety risk analysis efficiency is higher and more comprehensive. Moreover, the automobile safety risk analysis is carried out at the early stage of research and development of the automobile service system, so that the existing risk problem can be processed in time, the design efficiency of the automobile service system is higher, the cost is lower, and the safety protection is more timely.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of automobile safety risk analysis, and in particular to an automobile safety risk analysis method and electronic equipment. Background Art

[0002] The "new four trends" of automobiles, represented by electrification, intelligence, networking, and sharing, are accelerating the reshaping of the automotive industry chain and driving the automotive industry towards digital transformation. The concept of software-defined cars has become a new engine for traditional automakers to break out of the impasse. As a result, current cars contain more and more functions, and more and more parties are involved in data interaction in automotive business systems, and the scenarios are becoming more and more complex, which exposes cars to various security risks. In particular, if there are security risks in the data of automotive business systems, it will cause the leakage of car owner information and even have a great impact on the safety of car driving.

[0003] The existing automotive security analysis methods for automotive business systems are mainly carried out by automotive business system security engineers in the late stage of automotive business system development, based on security expertise and traditional technical means such as rule matching and static analysis, to scan and analyze security vulnerabilities in the automotive business system, first identify the security risk points of the automotive business system, and then reinforce and transform the security defects of the automotive business system based on the security risk points to eliminate the risks and threats of security attacks. However, taking the car-cloud business scenario as an example, the automotive business system faces many security risk points, and the means of attack are constantly changing. Relying on security engineers to identify security risk points will require a high level of professionalism from security engineers, and the cycle of manual security risk point analysis and design will also be very long, resulting in low efficiency and high cost in the automotive business system development and design links. In addition, after the automotive security risk analysis is carried out in the late stage of automotive business system development, the security reinforcement of the automotive business system will bring the cost of secondary development and testing, and there is a lag in security protection, and there are other problems and challenges such as the risks of the automotive business system have not been completely eliminated or eliminated in a timely manner. Summary of the invention

[0004] The present application provides an automobile safety risk analysis method and electronic equipment, which are used to solve the problem in the prior art that automobile business system safety engineers manually perform safety risk point analysis in the late stage of research and development, and the safety analysis cycle is long, resulting in low efficiency and high cost in the research and development and design links of the automobile business system, and lagging safety protection.

[0005] Among them, by inputting the product design requirement description document corresponding to the automobile business system and the corresponding target safety risk data into the automobile safety risk analysis model, the safety risk analysis results can be obtained conveniently and accurately, without manual safety risk analysis, making the safety risk analysis more efficient and comprehensive. In addition, since this solution is based on the product design requirement description document corresponding to the automobile business system for automobile safety analysis, the above-mentioned automobile safety risk analysis can be carried out in the early stage of automobile business system development, that is, in the early stage of automobile business system development, the automobile safety risk analysis results can be obtained based on the product design requirement description document to adjust the automobile business system design, etc., and the existing risk problems can be dealt with in time, and then the automobile business system design can be carried out, so that the automobile business system design is efficient, low-cost and timely in safety protection, avoiding the existing technology of conducting automobile safety risk analysis in the late stage of automobile business system development. If there is a safety risk problem, the existing business system design needs to be tested after the corresponding treatment, and the problem of low efficiency, high cost and delayed safety protection.

[0006] To solve the above technical problems, in the first aspect, the implementation method of the present application discloses a method for automobile safety risk analysis, which method includes: determining a product design requirement description document corresponding to an automobile business system; according to the product design requirement description document, determining target safety risk data corresponding to the product design requirement description document from the security risk data included in a security risk database; inputting the product design requirement description document and the target safety risk data into an automobile safety risk analysis model to obtain an automobile safety risk analysis result.

[0007] The automobile safety risk analysis method provided by the implementation of the present application determines the product design requirement description document corresponding to the automobile business system, determines the corresponding target safety risk data according to the product design requirement description document, inputs the product design requirement description document and the target safety risk data into the automobile safety risk analysis model, and obtains the safety risk analysis result. In this way, in the early stage of the development of the automobile business system, the automobile safety risk analysis model can obtain the safety risk analysis result based on the product requirement description document. On the one hand, there is no need to manually perform safety risk analysis, making the safety risk analysis more efficient and more comprehensive. On the other hand, the automobile safety risk analysis result can be obtained based on the product design requirement description document in the early stage of the development of the automobile business system, so as to adjust the design of the automobile business system, deal with the existing risk problems in a timely manner, and then design the business system, so that the business system design is efficient, low-cost and timely in safety protection, avoiding the existing technology of conducting automobile safety risk analysis in the later stage of the development of the automobile business system. If there is a safety risk problem, the existing business system design needs to be tested after the corresponding processing. The problem of low efficiency, high cost and delayed safety protection.

[0008] According to another specific implementation of the present application, the implementation of the present application discloses an automobile safety risk analysis method, which also includes obtaining security risk data included in a security risk database in the following manner: determining target information related to automobile safety risks, the target information including automobile industry safety specification information, automobile safety field knowledge, automobile safety policies, and historical specification information for automobile business system design; preprocessing the target information to obtain multiple security risk data blocks; performing data segmentation processing on the security risk data blocks to obtain multiple security risk data slices; inputting the security risk data slices into a vectorized processing model for vectorized processing to obtain security risk data.

[0009] By adopting the above technical scheme, the target information related to automobile safety risks is processed to obtain vectorized safety risk data that can be input into the automobile safety risk analysis model, and the safety risk data is stored in the safety risk database. During the actual automobile safety risk analysis, the safety risk-related data for the product design requirement description document can be obtained by obtaining the corresponding target safety risk data, which enables the large model to perform safety risk analysis based on the product design requirement description document and obtain automobile safety risk results more comprehensively and quickly.

[0010] According to another specific implementation of the present application, the implementation of the present application discloses a method for analyzing automobile safety risks, in which preprocessing includes at least one of the following processes: data entity parsing processing; document division processing; data enhancement processing; special data processing; and metadata addition processing.

[0011] By adopting the above technical solution, the target information is preprocessed using a variety of preprocessing methods, so that the obtained risk data block is more comprehensive and more in line with the automobile safety risk requirements.

[0012] According to another specific implementation of the present application, the implementation of the present application discloses a method for analyzing automobile safety risks, in which data segmentation processing includes at least one of the following processing: sentence segmentation processing based on sentence segmentation; recursive segmentation processing based on block rules; semantic segmentation processing based on text similarity; special structure segmentation processing based on specific structured content.

[0013] By adopting the above technical solution, a variety of data segmentation methods are used to segment the safety risk data blocks, making the risk data pieces more comprehensive and more in line with automobile safety risk requirements.

[0014] According to another specific implementation of the present application, an implementation of the present application discloses an automobile safety risk analysis method, which inputs a safety risk data piece into a vectorized processing model for vectorized processing to obtain safety risk data, including obtaining the safety risk data in the following manner: inputting a safety risk data piece into at least one vectorized processing model, understanding the meaning of the safety risk data piece through the vectorized processing model, and dynamically adjusting the meaning to obtain vectorized safety risk data.

[0015] By adopting the above technical solution, the safety risk data slices are vectorized based on the vectorized processing model, and the word meaning is understood based on the vectorized processing model, and the word meaning is dynamically adjusted to obtain vectorized safety risk data. The obtained safety risk data is more suitable for automobile safety risk analysis, which facilitates the automobile safety risk analysis model to output more comprehensive and accurate safety risk analysis results.

[0016] According to another specific implementation of the present application, the implementation of the present application discloses an automobile safety risk analysis method, which determines, based on a product design requirement description file, target safety risk data corresponding to the product design requirement description file from the safety risk data included in a safety risk database, including: inputting the product design requirement description file into a vectorization processing model so that the vectorization processing model performs vectorization processing on the product design requirement description file to obtain vectorized product design requirement description information, and performing similarity calculation between the product design requirement description information and each safety risk data in the safety risk database to obtain the similarity between the product design requirement description information and each safety risk data, and determining the target safety risk data corresponding to the product design requirement description file based on the similarity.

[0017] By adopting the above technical solution, by determining the product design requirement description file corresponding to the automobile business system, the product design requirement description file is vectorized to obtain product design requirement description information, and similarity calculation is performed based on the product design requirement description information and each security risk data in the security risk database to obtain the similarity between the product design requirement description information and each security risk data, and the target security risk data corresponding to the product design requirement description file is determined based on the similarity. In this way, security risk data with high relevance to the product design requirement description file is obtained based on the similarity.

[0018] According to another specific implementation of the present application, the implementation of the present application discloses a vehicle safety risk analysis method, wherein the vehicle safety risk analysis model is an LLaMA model or a ChatGLM model generated based on retrieval enhancement technology.

[0019] By adopting the above technical solution, the LLaMA model or ChatGLM model generated based on the retrieval enhancement technology is used to perform automobile safety risk analysis to obtain automobile safety risk analysis results. In this way, the automobile safety risk can be automatically analyzed, the efficiency of automobile safety risk analysis is accelerated, and the accuracy of automobile safety risk analysis is improved.

[0020] According to another specific implementation of the present application, the implementation of the present application discloses an automobile safety risk analysis method, wherein the product design requirement description document includes business scenario information and business rule information of product functions.

[0021] By adopting the above technical solution, automobile safety risk analysis is performed based on the business scenario information and business rule information of product functions, so that in the early stage of automobile business system development, the safety risk analysis results can be obtained based on the product design requirements document. Existing risk problems can be dealt with in a timely manner, and then the business system design can be carried out, so that the business system design is efficient, low-cost and timely in safety protection, avoiding the existing technology of conducting automobile safety risk analysis in the later stage of automobile business system development. If there are safety risk problems, the existing business system design is low in efficiency, high in cost and safety protection is delayed after the corresponding treatment.

[0022] According to another specific implementation of the present application, the implementation of the present application discloses a vehicle safety risk analysis method, and the vehicle safety risk analysis results include a vehicle safety risk analysis list, vehicle safety risk response measures and protection strategies.

[0023] By adopting the above technical solution, an automobile safety risk analysis list, automobile safety risk response measures and protection strategies can be obtained based on the automobile safety risk analysis model. In this way, a more comprehensive automobile safety risk analysis list can be obtained, and the corresponding automobile safety risk response measures and protection strategies can be directly obtained, which is convenient for engineers to adjust the design of automobile business systems based on automobile safety risk response measures and protection strategies.

[0024] In the second aspect, the implementation method of the present application also discloses an electronic device, including: a processor, and a memory connected to the processor in communication; the memory stores a computer program; the processor executes the computer program stored in the memory, so that the electronic device implements the automobile safety risk analysis method provided by any one of the implementation methods of the first aspect above.

[0025] In a third aspect, the implementation method of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the automobile safety risk analysis method provided by any one of the implementation methods of the first aspect mentioned above.

[0026] In a fourth aspect, an implementation of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the automobile safety risk analysis method provided in any one of the implementations of the first aspect above.

[0027] It can be understood that the beneficial effects of the second to fourth aspects mentioned above can also be found in the relevant description of the first aspect mentioned above, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flow chart of an existing automobile safety risk analysis method provided in an embodiment of the present application;

[0029] Figure 2 This is a schematic diagram of a process for obtaining a security risk database provided by an embodiment of the present application;

[0030] Figure 3 It is a schematic diagram of the principle of vectorized processing by a vectorized processing model provided in an embodiment of the present application;

[0031] Figure 4 This is a schematic diagram of the principle of preprocessing and data segmentation provided in an embodiment of the present application;

[0032] Figure 5 It is a schematic diagram of the principle of vectorization processing by another vectorization processing model provided in an embodiment of the present application;

[0033] Figure 6 It is a schematic diagram of the principle of determining target security risk data provided by an embodiment of the present application;

[0034] Figure 7 It is a flow chart of another automobile safety risk analysis method provided in an embodiment of the present application;

[0035] Figure 8 It is a flow chart of another automobile safety risk analysis method provided in an embodiment of the present application;

[0036] Fig. 9 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application.

[0037] Specific implementation method

[0038] As mentioned earlier, as the "new four modernizations" of automobiles represented by electrification, intelligence, networking, and sharing are accelerating the reshaping of the automotive industry chain and promoting the automotive industry to move towards digital transformation, the concept of software-defined cars has also become a new engine for traditional car companies to break through. This is a transformation from the traditional manufacturing industry to the travel life and ecological service industry. At the same time, as the data interaction of the intelligent networked car business system involves many parties and the scenarios are complex, the business system faces various security risks. When a security risk occurs in a certain data of the business system, the owner's information is leaked and even has a great impact on driving safety.

[0039] The existing security risk analysis methods for automotive business systems mainly rely on the identification of security risk points of the business system by business system security engineers in the later stage of automotive business system development based on security expertise and traditional technical means such as rule matching and static analysis, and then the security defects of the business system are reinforced and transformed to eliminate the risks and threats of security attacks.

[0040] However, taking the car-cloud business scenario as an example, the automotive business system faces many security risk points, and the means of attack are constantly changing. Relying on security engineers to identify security risk points will require relatively high professionalism from security engineers, and the manual security risk point analysis, security analysis and design cycle will also be very long, resulting in low efficiency and high cost in the research and development and design of business systems.

[0041] In addition, security risk analysis is carried out in the later stages of R&D or even after the product goes online. If there are security risk points, the business system needs to be reinforced, and secondary development and testing of the reinforced business system is required, which brings about secondary development costs. Moreover, security protection in the later stages will also bring about a lag in security protection, which means that the security risks of the business system have not been completely eliminated, and other problems and challenges.

[0042] Based on this, the implementation method of this application provides an automobile safety risk analysis method, which performs intelligent safety risk analysis on the automobile business system in advance. In the early stage of business system development, the safety risk analysis results can be obtained conveniently and accurately based on the automobile safety risk analysis model according to the product design requirement description document (that is, the product design requirement description document) corresponding to the automobile business system and the corresponding target safety risk data. There is no need for manual safety risk analysis, making the safety risk analysis more efficient and comprehensive. It can help business system designers identify safety risks and provide safety protection designs in the research and development design stage, improve the ability of business system safety comprehensive analysis, and at the same time play a role in reducing costs and increasing efficiency in safety analysis and design work in all business scenarios.

[0043] Next, the automobile safety risk analysis method provided by the implementation of the present application is described in detail in conjunction with the accompanying drawings.

[0044] like Figure 1 As shown, in one implementation of the present application, the automobile safety risk analysis method provided by the implementation of the present application specifically includes the following steps.

[0045] S100, determine the product design requirement specification document corresponding to the automobile business system.

[0046] S200, according to the product design requirement specification document, determining target security risk data corresponding to the product design requirement specification document from the security risk data included in the security risk database.

[0047] S300, inputting the product design requirement specification document and the target safety risk data into the automobile safety risk analysis model to obtain the automobile safety risk analysis result.

[0048] The automobile safety risk analysis method provided by the implementation mode of the present application, in the early stage of automobile business system development, by determining the product design requirement description document corresponding to the automobile business system, determining the corresponding target safety risk data according to the product design requirement description document, inputting the product design requirement description document and the target safety risk data into the automobile safety risk analysis model, and obtaining the safety risk analysis result. In this way, in the early stage of development, the automobile safety risk analysis model can obtain the safety risk analysis result based on the product requirement description document. On the one hand, there is no need to manually perform safety risk analysis, making the safety risk analysis more efficient and more comprehensive. On the other hand, the automobile safety risk analysis result can be obtained based on the product design requirement description document in the early stage of automobile business system development, so as to adjust the design of the automobile business system, deal with the existing risk problems in time, and then design the business system, so that the business system design is efficient, low-cost and timely in safety protection, avoiding the existing technology of automobile safety risk analysis in the later stage of automobile business system development, if there is a safety risk problem, after the corresponding treatment, it is necessary to test the existing business system design efficiency, high cost and safety protection. There are lags.

[0049] For step S100, when the research and development of a specific automobile business system enters the upstream product design stage (i.e., the early stage of research and development), the product designers of the business system define each function of the product, describe the business scenarios and business rules of the product functions, and generate a product design requirement specification (i.e., a product design requirement description document) for the design of relevant scenarios of the specific business system.

[0050] Among them, the product design requirements description document includes business scenario information and business rule information of product functions.

[0051] Furthermore, in the implementation of this application, before conducting automobile safety risk analysis (that is, when the automobile business system design is offline), it is necessary to build an enterprise private domain security knowledge base to generate security risk data and complete the preparation of the security risk database, which is then applied to the specific automobile business system product development security analysis stage. Figure 2 As shown, for step S200, in one implementation of the present application, the security risk data and the security risk database are obtained by:

[0052] S210, determining target information related to automobile safety risks, the target information including automobile industry safety specification information, automobile safety domain knowledge, automobile safety strategy, and historical specification information for automobile business system design.

[0053] Exemplarily, build an enterprise private domain security knowledge base and obtain data sources, such as collecting and organizing the enterprise's national standards, enterprise standards, regulations and industry ISO26262 safety certification specifications in the automotive field (that is, automotive industry safety specification information), the security field knowledge accumulated from the enterprise's professional knowhow in safety risk analysis on multiple mass-produced vehicle projects (that is, historical specification information for automotive business system design), and enterprise knowledge assets such as security attack and defense strategies and rules (that is, automotive safety strategies) as the data source of the enterprise private domain security knowledge base (as an example of target information).

[0054] S220, pre-processing the target information to obtain a plurality of security risk data blocks.

[0055] For example, due to the data-driven nature, data with a high signal-to-noise ratio is still very important. Therefore, before generating security risk data, it is necessary to preprocess the data source, that is, to optimize the original data.

[0056] Among them, the preprocessing includes at least one of the following processes: data entity parsing processing, document division processing, data enhancement processing, special data processing, and metadata addition processing.

[0057] Exemplarily, data entity resolution processing is to eliminate the ambiguity of entities and terms to achieve consistent references. For example, "LLM", "large language model" and "large model" are standardized as common terms. In other words, different terms with the same meaning and function are unified.

[0058] Document segmentation is to reasonably divide documents of different topics, that is, to bring documents of different topics together to achieve document normalization.

[0059] Data enhancement is to use synonyms, definitions, and even translations into other languages ​​to enhance the diversity of the knowledge base. In other words, the language text is expanded with synonyms, definitions, and translated into other languages.

[0060] Special data processing is for sensitive data such as time. For frequently updated topics, a mechanism is implemented to invalidate or update outdated documents. That is, based on the generation time of the document, documents on the same topic are updated in real time, and invalid documents are deleted.

[0061] Metadata addition is done by adding additional information such as content summary, timestamp, and questions that users may ask to enrich the knowledge base.

[0062] Therefore, in the implementation of the present application, the text content of the target data is first expanded and then divided based on the topics of each document to obtain multiple data blocks of different topics.

[0063] S230, performing data segmentation processing on the security risk data block to obtain a plurality of security risk data slices.

[0064] For example, the amount of data in the enterprise private domain security knowledge base that is usually retrieved will far exceed the input length that the automobile safety risk analysis model can accept. Therefore, reasonable chunking should be done as much as possible without exceeding the input length of the automobile safety risk analysis model to ensure the differences between data blocks and the consistency within the data blocks.

[0065] Therefore, in one implementation of the present application, it is necessary to perform data segmentation processing on each security risk data block generated in S220 to obtain multiple security risk data slices corresponding to each security risk data block.

[0066] Among them, the data segmentation processing includes at least one of the following processing: sentence segmentation processing based on sentence segmentation, recursive segmentation processing based on block rules, semantic segmentation processing based on text similarity, and special structure segmentation processing based on specific structured content.

[0067] Exemplarily, the sentence segmentation processing based on sentence segmentation is called through a mainstream development framework such as langchain, so as to call the sentence segmentation function provided by the Natural Language Processing Toolkit (NLTK) or a natural language processing library (such as the spaCy library), and perform sentence segmentation processing on each security risk data block to obtain multiple security risk data pieces.

[0068] The recursive segmentation process based on the block rule is to pre-formulate the block rule and recursively decompose the text by repeatedly applying the block rule. For example, in langchain, the security risk data block will be segmented by paragraph line breaks (\n\n). Then the size of these segmented security risk data blocks (that is, the security risk data slices after segmentation) is checked. If the size does not exceed the preset threshold, the security risk data block is retained. For the security risk data block that exceeds the preset threshold, a single line break (\n) is used to segment it again until the segmented security risk data slice meets the preset threshold. Among them, the preset threshold can be set according to actual needs.

[0069] Semantic segmentation based on text similarity is to perform semantic segmentation by calculating the similarity of vectorized text content. For example, text content with the same similarity is segmented into the same security risk data block, and text content with different similarities is segmented into different security risk data blocks.

[0070] The similarity calculation is to determine the Euclidean distance and cosine similarity of two text vectors, and calculate the similarity of the two text vectors based on the Euclidean distance and cosine similarity of the two vectors. If the similarity is greater than a preset similarity threshold (e.g., 0.5, 0.7, 0.8, 0.9, etc.), the two text vectors are considered similar, that is, the text contents corresponding to the two text vectors are similar. The preset similarity threshold can also be other thresholds, which are set according to actual needs.

[0071] Special structure segmentation processing based on specific structured content is to develop special segmenters for specific structured content (such as Markdown, LaTex, JSON and other text format content). These segmenters are specially designed to process documents of the corresponding type to ensure that the text structure information is correctly retained and the security risk data slice is obtained.

[0072] S240, input the security risk data slice into the vector processing model for vector processing to obtain security risk data.

[0073] In one implementation of the present application, the vectorization processing model can be an embedding model such as an embedding model, and of course, it can also be other models for data vectorization processing. Therefore, in the present application, it is necessary to perform data vectorization processing (data embedding) on ​​the safety risk data to obtain vectorized data that can be input into the automobile safety risk analysis model and can be recognized by the automobile safety risk analysis model.

[0074] Exemplarily, the text data (ie, the security risk data piece) is vectorized using an embedding model so as to facilitate the use of vector retrieval in the retrieval stage.

[0075] In one implementation of the present application, a security risk data piece is input into a vectorized processing model for vectorized processing to obtain security risk data, including: inputting a security risk data piece into at least one vectorized processing model, understanding the meaning of the security risk data piece through the vectorized processing model, and dynamically adjusting the meaning to obtain vectorized security risk data.

[0076] Exemplarily, the security risk text data of each security risk data piece is input into an embedding model (as an example of a vectorized processing model) using a dynamic embedding method, because dynamic embedding is better able to handle polysemy than static embedding, and the vectors of words with completely different semantics are fixed using static embedding. In contrast, models that introduce a self-attention mechanism, such as a pre-trained language representation model, the BERT (Bidirectional Encoder Representation from Transformers) model (as an example of a vectorized processing model), can provide dynamic word meaning understanding, so that the word meaning can be dynamically adjusted according to the context, so that the same word has different vector representations in different contexts, thus obtaining vectorized security risk data.

[0077] Furthermore, the security risk text data of each security risk data piece can be input into the embedding model by fine-tuning the embedding method. Because most embedding models are trained on general corpus (that is, text semantic analysis without limiting the scenario), in order to allow the embedding model to have a better understanding of the vocabulary in the vertical field and then generate vectorized security risk data, the embedding model is fine-tuned based on the usage scenario so that the embedding model is a model suitable for the current security risk analysis scenario, and then the security risk data piece is vectorized based on the fine-tuned embedding model. In this way, the general model is adjusted based on the usage scenario, and the security risk data piece is input into the adjusted embedding model to obtain vectorized security risk data.

[0078] Furthermore, a hybrid embedding method can be used to input the safety risk text data of each safety risk data piece into the embedding model. For example, according to the user's question, the embedding model is specified as a model for automobile safety risk analysis, and the knowledge base text (that is, the safety risk data piece) is vectorized using different embedding models to obtain vectorized safety risk data.

[0079] S250: storing the security risk data in a security risk database.

[0080] That is, in this application, after the target data is preprocessed, data segmented, and data embedding vectorized based on the embedding model, the security risk data in the enterprise security knowledge base is segmented and converted into a high-dimensional vector representation, so as to facilitate the subsequent automobile safety risk analysis model to perform model processing and matching. Finally, the vectorized security risk data is stored in a vector database (as an example of a security risk database).

[0081] Moreover, in another implementation of the present application, target data in the enterprise security knowledge base can be continuously added and iteratively updated to continuously maintain the knowledge base in the enterprise security field and keep the enterprise's knowledge in the field of automotive safety at the latest status of industry security.

[0082] Therefore, the implementation method of this application builds an enterprise security knowledge base, first of all, based on the cloud, vector processing is carried out on security specifications, security professional experience, and security power amplifier strategies and rules such as automotive industry safety specification information, automotive safety field knowledge, automotive safety strategies, and historical specification information for automotive business system design.

[0083] For example, a loader is used to convert the target information of an enterprise knowledge base into separate documents, and then a splitter is used to convert it into data blocks (i.e., security risk data blocks) or fragments (i.e., security risk data slices), which are passed to the embedding model. The embedding model converts the target information into vectors that can be semantically searched, and then these security risk data vectors and their text fragments are saved in a vector database.

[0084] The key technology used in this application is to embed the model to perform data vectorization processing. Embedding refers to the translator from human text language to artificial intelligence digital language. Figure 3 As shown, Embedding Machine can convert human language into digital vector fragments (AI "Language"). Human language is, for example, "Hello, How are you?", and the converted digital vector fragments are, for example, [3, 11, 42, ..., 19, 68]; human language is, for example, "Hello, How's it going?", and the converted digital vector fragments are, for example, [3, 12, 42, ..., 18, 65]; human language is, for example, "Good morning!", and the converted digital vector fragments are, for example, [5, 17, 37, ..., 22, 60]; human language is, for example, "I like cupcakes.", and the converted digital vector fragments are, for example, [28, 91, 6, ..., 42, 30].

[0085] Furthermore, it also involves knowledge base indexing (knowledge base Index), which involves loading and splitting. Loading is to obtain the target content information (that is, target information) in the knowledge base from the usual storage location, and splitting is to split the target information such as knowledge into blocks of a size suitable for embedded search and save it in the vector database.

[0086] That is, if Figure 4 As shown, the target information (Knowledge base) is loaded (Loader) to obtain a document (Document), that is, to obtain a preprocessed security risk data block, and the text content of the document is split (Splitter) to obtain data slices (that is, security risk data slices) (Document Snippets).

[0087] Furthermore, each knowledge fragment (i.e., security risk data fragment) is passed through the Embedding model, and the embedded representation of the text (i.e., security risk data vector) is returned, and the security risk data vector is saved in the vector database to embed the vector database, which is optimized for digital vectors. Figure 5 As shown, the security risk data pieces (Document Snippets, also known as Knowledge Snippets) are vectorized (Embedding Machine) to obtain security risk data vectors (Embeddings), and the security risk data pieces and security risk data vectors are stored in a vector database (Vector Database).

[0088] Furthermore, in the implementation method of the present application, after building the enterprise private domain security knowledge base, when conducting automobile safety risk analysis (that is, when the automobile business system design is online), the target security risk data corresponding to the product design requirement description document is determined based on the security risk data included in the product design requirement description document and the security risk database.

[0089] Therefore, in one implementation of the present application, for step S200, based on the product design requirement description file, target security risk data corresponding to the product design requirement description file is determined from the security risk data included in the security risk database, including: inputting the product design requirement description file into the vectorization processing model, so that the vectorization processing model vectorizes the product design requirement description file to obtain vectorized product design requirement description information, and performing similarity calculation based on the product design requirement description information and each security risk data in the security risk database to obtain the similarity between the product design requirement description information and each security risk data, and determining the target security risk data corresponding to the product design requirement description file based on the similarity.

[0090] Exemplarily, after using the Embedding model to vectorize the product design requirement description document, vectorized product design requirement description information is obtained, which is retrieved and calculated through the enterprise private domain security knowledge base to obtain the vector result of the target security risk data of the semantic data matching the knowledge base and returned to the big model.

[0091] like Figure 6As shown in FIG. 1 , the retrieval principle is to load the original data (i.e., the target data), load, segment, and vectorize the target data to obtain security risk data (e.g., -3.7, -1.2, 0.5, …, 1.2, -4.4, 0.9; 2.8, 4.5, -1.7, …, 0.8, 3.6, -4.8; -0.6, 2.1, 4.0, …, -2.3, 3.4, -1.0) and store them in the vector database, vectorize the user questions (i.e., the product design requirement description documents) to obtain vectorized product design requirement description information (e.g., 4.9, -0.4, 2.3, -1.6, 3.0, -2.8, 1.1, -3.2, …), and query the vector database for the high similarity (e.g., scores of 0.887, 0.763, 0.51, etc.). 2)'s security risk data vectors (for example, vectors 4.6, -0.2, 1.8, -9.9, 3.0, -2.8, 1.2, -2.3, ...) and corresponding documents ((for example, vectors 4.6, -0.2, 1.8, -9.9, 3.0, -2.8, 1.2, -2.3, ...) and corresponding documents (xxxxxxx); (for example, vectors 3.9, -1.4, 5.3, 2.6, 5.0, -7.8, -0.1, 2.2, ...) and corresponding documents (xxxxxxx); (for example, vectors 4.·, -5.4, -1.3, -8.6, 9.0, -3.3, 1.5, -2.2, ...) and corresponding documents (xxxxxxx)) to obtain the security risk data document with the highest similarity as the target security risk data. In this way, questions raised by users are sent and passed to the vector database through the same embedding model, and the closest matching fragment (that is, the target security risk data) is determined, and the target security risk data will be used to answer the user's questions.

[0092] That is, in the implementation method of the present application, vector data retrieval will be performed based on the product design requirement description file. First, the user question will be vectorized (that is, the product design requirement description file will be vectorized) to obtain vector data (that is, the product design requirement description information), and then the Embedding model will be used to retrieve from the vector database, and the similarity between the product design requirement description information and the security risk data will be calculated to obtain the n target security risk data with the highest similarity.

[0093] In this way, if the user's question hits the vector database, the cached answer information (that is, the target security risk data) will be obtained from the vector database.

[0094] For step S300, an advanced open source big model is used as the big model engine for core analysis. Based on the big model and vectorized target security risk data, the big model is equipped with the ability to identify and analyze security risks, so as to realize business function security analysis based on the big model.

[0095] In one implementation of the present application, the automobile safety risk analysis result includes an automobile safety risk analysis list, automobile safety risk response measures and protection strategies.

[0096] Among them, the big model can detect potential security risks and provide corresponding response strategy suggestions based on the target security risk data and other vectorized data that match the business scenario input included in the product design requirement description document. That is, according to the semantics of the business scenario understanding description included in the product design requirement description document, and the matching security risk items, security strategy vectorized results and other target security risk data are retrieved from the enterprise private domain security knowledge base, the product design requirement description document and the target security risk data are input into the big model, and through the generation capability of the big model, a specific security risk analysis list, response measures and defense plans are output to the business system architecture designer.

[0097] Among them, the automobile safety risk analysis model is an LLM model such as an LLaMA model or a ChatGLM model generated based on retrieval-augmented generation (RAG) technology.

[0098] Therefore, if Figure 7As shown, the automobile safety risk analysis method provided by the implementation of this application is actually a method for reducing costs and increasing efficiency in the safety risk analysis and design of automobile business systems based on large-model RAG technology. The safety specifications, safety professional experience, and security attack and defense strategies and rules (i.e., target data) obtained from the cloud are loaded and preprocessed in the enterprise's private domain security knowledge base to obtain data blocks, and then data segmentation processing is performed to obtain data slices, which are then input into the Embedding model for vectorization processing to obtain the safety risk data vectorized data and stored in the vector database. Then, based on the RAG retrieval enhancement generation technology, the product design requirement description document (i.e., prompt) is input into the Embedding model for vectorization processing to obtain the product design requirement description information, and data retrieval is performed to obtain the target safety risk information (i.e., retrieval matching results), and then the product design requirement description document and the target safety risk information are input into the LLM large model to output the automobile safety risk analysis results (i.e., output). In this way, by introducing big models and RAG technology, combined with the professional knowledge base (i.e., target data) such as security specifications, security professional experience, security attack and defense strategies and rules related to security risk scenarios and security attack and defense strategies of business systems in the automotive field obtained from the cloud, after vector processing, an intelligent security risk analysis system is built in combination with a big language model. In actual application, business description documents such as specific product design requirements description documents are used as inputs of the intelligent analysis system, and then the security risk list and security protection design solutions corresponding to the automotive business system are efficiently, accurately and comprehensively detected, analyzed and responded to, enabling the security function design and R&D process of the business system more efficiently. It solves various problems brought about by the long analysis cycle and inconsistent safety standards caused by the traditional analysis method that relies on manual work of security engineers, reduces the threshold and cost of security design of automotive business systems, and improves the overall energy efficiency of enterprises in the automotive business R&D process system.

[0099] Take the automotive business system at the design node as an example. Figure 8As shown, the automobile safety risk analysis method provided by the implementation of this application first builds the enterprise private domain security knowledge base in advance, performs data preprocessing, data segmentation, and vectorization processing based on the Embedding model on the target information such as automobile industry safety specifications, domain knowledge, and security strategies to obtain security risk data, which is stored in the vector database. When the research and development of a specific automobile business system enters the upstream product design stage, the product designer will carry out product project establishment, define each function of the product, and describe the business scenarios and business rules of the product functions to form a product design requirement specification file, which is submitted to the business system intelligent safety analysis system as input in turn. The product design requirement description information is obtained by vectorizing the product design requirement description file based on the Embedding model, and the target security risk information is retrieved from the vector database. The product design requirement description file and the target security risk information are input into the large model for intelligent security analysis, and the security risk list related to the product design requirements and the security reinforcement measures (that is, security analysis results and protection plans) required to meet the security specifications are output. The business system designer completes the system security architecture design based on the output security risk results. As a result, the automobile safety risk analysis process is completely replaced by manual work by a large model. Not only is the analysis comprehensive and accurate, ensuring that the automotive-grade safety requirements are met, it greatly improves the R&D efficiency of the business system and reduces the cost of special R&D and communication. In addition, the safety risk analysis and security protection are quickly completed in the upstream stage of the business system R&D cycle, and the automobile safety risk analysis process is safely shifted from the downstream R&D stage (i.e., development, testing, deployment, operation and maintenance) to the upstream R&D stage, reducing the R&D testing costs caused by the traditional security reinforcement transformation in the downstream R&D through security vulnerability scanning and testing.

[0100] The automobile safety risk analysis method provided by the implementation method of this application is a method for realizing intelligent safety risk analysis and security left shift of automobile business systems based on large model RAG technology, which puts security intelligent analysis and protection in the upstream design stage of the system development process, and uses AI large model RAG technology to vectorize target information such as cloud security specifications, security professional experience, and security attack and defense strategies and rules. Business system designers only need to use product requirement description documents as inputs to the intelligent security analysis system to quickly and accurately obtain a list of security risk analysis of business scenarios and a design plan for security protection. Based on this, system designers can timely supplement and improve the security architecture design of relevant parts of the system, thereby realizing the implementation of security protection for all business scenarios of automobiles, and improving the research and development efficiency of security risk analysis and security design of automobile business systems. The business system enabled by the intelligent security analysis of this application has fully met the safety requirements of automotive standards, greatly reducing the security risks of security vulnerability scanning and analysis in the later stages of development or after going online, and reducing the research and development costs of business system security transformation.

[0101] The automobile safety risk analysis method provided by the implementation method of this application realizes the intelligent and professional design of safety risk analysis of automobile long business links and multiple scenarios based on the support of large model RAG technology, which can significantly improve the comprehensiveness and accuracy of automobile business system safety risk analysis, and meet the safety requirements of automobile business systems in laws, national standards, and industry specifications. And it can significantly improve the response speed and R&D efficiency of safety risk analysis for automobile business systems, and the analysis work cycle is reduced from 3 to 5 days to minutes. Further, it reduces the threshold of professional safety analysis, reduces the manpower investment in safety analysis, and reduces the communication and confirmation link of understanding business needs, which can greatly reduce the R&D manpower and communication costs of safety analysis and design. And through the intelligent safety analysis system (that is, the automobile safety risk analysis system) and the R&D process, the left shift of safety analysis and design is realized, and the safety risk analysis and safety protection design are quickly completed in the upstream stage of the business system. Compared with the traditional method of security vulnerability scanning analysis and security architecture transformation in the downstream of R&D, it greatly reduces the additional transformation and testing costs caused by security reinforcement, so that the impact range of security transformation on the business system can be estimated to reduce the R&D cost of several man-months. Moreover, based on the open API interface of the intelligent security analysis system, the system has strong adaptability and scalability. By connecting this system with the R&D process through the API interface, the automation of the intelligent security analysis of the R&D process can be further realized.

[0102] In one implementation of the present application, the automobile safety risk analysis method of the present application can be applied to the cloud, that is, an automobile safety risk analysis system is generated in the cloud based on the enterprise's private domain security knowledge base and a large model to perform automobile safety risk analysis.

[0103] In another implementation of the present application, the automobile safety risk analysis method of the present application can be applied to electronic devices.

[0104] Among them, the automobile safety risk analysis method can be applied to electronic devices such as computers. Fig. 9 , Fig. 9 The figure shows a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig. 9 As shown, the electronic device may include: a transceiver 121 , a processor 122 , and a memory 123 .

[0105] The processor 122 executes the computer execution instructions stored in the memory, so that the processor 122 executes part of the technical solutions of the automobile safety risk analysis method in the above embodiment. The processor 122 can be a general-purpose processor, including a central processing unit CPU, a network processor (NP), etc.; it can also be a digital data processor DSP, an application-specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gates or transistor logic devices, and discrete hardware components.

[0106] The memory 123 is connected to the processor 122 via a system bus and completes communication between them. The memory 123 is used to store computer program instructions.

[0107] By way of example and not limitation, the memory 123 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 123 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 123 may be inside or outside the integrated gateway device. In a particular embodiment, the memory 123 is a non-volatile solid-state memory. In a particular embodiment, the memory 123 includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or a flash memory, or a combination of two or more of these.

[0108] The transceiver 121 may be used to obtain tasks to be executed and configuration information of the tasks to be executed.

[0109] The system bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The transceiver is used to realize the communication between the database access device and other computers (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM) and may also include non-volatile memory.

[0110] An embodiment of the present application also provides a chip for running instructions, which is used to execute the technical solution of the automobile safety risk analysis method in the above embodiment.

[0111] An embodiment of the present application also provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed on a processor of an electronic device, the processor of the electronic device executes the technical solution of the automobile safety risk analysis method of the above embodiment.

[0112] In some possible implementations, various aspects of the method provided in the present application can also be implemented in the form of a program product, which includes program code. When the program product runs on a processor of an electronic device, the program code is used to enable the processor of the electronic device to execute the steps of the method according to various exemplary implementations of the present application described above in this specification. For example, the electronic device can execute the automobile safety risk analysis method recorded in the embodiments of the present application.

[0113] The program product may employ any combination of one or more readable media. The readable medium may be a readable data medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0114] The implementation method of the present application also provides a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor can read the computer program from the computer-readable storage medium. When at least one processor executes the computer program, it can implement the technical solution of the automobile safety risk analysis method in the above-mentioned embodiment.

[0115] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices and computer program products according to the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable information processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable information processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0116] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable information processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0117] These computer program instructions can also be loaded onto a computer or other programmable information processing device so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0118] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary technical means in the art that are not disclosed in the present application.

[0119] It should be noted that, in addition to the implementation of the present application described in the above-mentioned specific embodiments, those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Although the description of the present application is introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to the implementation. On the contrary, the purpose of introducing the invention in conjunction with the implementation is to cover other options or modifications that may extend based on the claims of the present application. In order to provide an in-depth understanding of the present application, the above description contains many specific details, and the present application may also be implemented without using these details. In addition, in order to avoid confusion or blurring the focus of the present application, some specific details will be omitted in the description. It should be noted that, in the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0120] It should be noted that in this specification, similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0121] Although the present application has been illustrated and described by referring to certain preferred implementations of the present application, it should be understood by those skilled in the art that the above contents are further detailed descriptions of the present application in combination with specific implementations, and it cannot be determined that the specific implementation of the present application is limited to these descriptions. Those skilled in the art may make various changes in form and details, including making several simple deductions or substitutions, without departing from the spirit and scope of the present application.

Claims

1. A method for analyzing automobile safety risks. It is characterized in that The method comprises: Determine the product design requirement specification document corresponding to the automotive business system; According to the product design requirement specification document, determining target security risk data corresponding to the product design requirement specification document from the security risk data included in the security risk database; The product design requirement specification document and the target safety risk data are input into an automobile safety risk analysis model to obtain an automobile safety risk analysis result.

2. The automobile safety risk analysis method according to claim 1, It is characterized in that The method further includes obtaining the security risk data included in the security risk database by: Determine target information related to automobile safety risks, the target information including automobile industry safety specification information, automobile safety domain knowledge, automobile safety strategy, and historical specification information for automobile business system design; Preprocessing the target information to obtain a plurality of security risk data blocks; Performing data segmentation processing on the security risk data block to obtain a plurality of security risk data slices; The security risk data slice is input into a vector processing model for vector processing to obtain the security risk data.

3. The automobile safety risk analysis method according to claim 2, It is characterized in that The pre-processing includes at least one of the following processes: Data entity parsing and processing; Document segmentation processing; Data enhancement processing; Special data processing; Metadata addition processing.

4. The automobile safety risk analysis method according to claim 2 or 3, It is characterized in that The data segmentation process includes at least one of the following processes: Sentence segmentation processing based on sentence segmentation; Recursive segmentation processing based on block rules; Semantic segmentation processing based on text similarity; Special structure segmentation processing based on specific structured content.

5. The automobile safety risk analysis method according to any one of claims 2 to 4, It is characterized in that Inputting the security risk data piece into the vector processing model for vector processing to obtain the security risk data includes: The security risk data piece is input into at least one vectorized processing model, and the vectorized processing model is used to understand the meaning of the security risk data piece and dynamically adjust the meaning of the word to obtain the vectorized security risk data.

6. The automobile safety risk analysis method according to claim 5, It is characterized in that According to the product design requirement specification document, target security risk data corresponding to the product design requirement specification document is determined from the security risk data included in the security risk database, including: Input the product design requirement description file into the vectorization processing model so that the vectorization processing model performs vectorization processing on the product design requirement description file to obtain vectorized product design requirement description information, and perform similarity calculation based on the product design requirement description information and each security risk data in the security risk database to obtain the similarity between the product design requirement description information and each security risk data, and determine the target security risk data corresponding to the product design requirement description file based on the similarity.

7. The automobile safety risk analysis method according to any one of claims 1 to 6, It is characterized in that The automobile safety risk analysis model is an LLaMA model or a ChatGLM model generated based on retrieval enhancement technology.

8. The automobile safety risk analysis method according to any one of claims 1 to 7, It is characterized in that The product design requirement description document includes business scenario information and business rule information of product functions.

9. The automobile safety risk analysis method according to any one of claims 1 to 8, It is characterized in that The automobile safety risk analysis results include an automobile safety risk analysis list, automobile safety risk response measures and protection strategies.

10. An electronic device, It is characterized in that include: A processor, and a memory communicatively connected to the processor; The memory stores a computer program; The processor executes the computer program stored in the memory so that the electronic device implements the automobile safety risk analysis method as described in any one of claims 1-9.