Configuration information determination method and device based on AUTOSAR

Abstract

The invention provides a configuration information determination method and device based on AUTOSAR, and relates to the field of automatic software configuration. The method comprises the steps of determining function description information of a configuration item based on configuration demand information and a module dependency relationship between modules in an AUTOSAR, retrieving associated information corresponding to the function description information of the configuration item, determining a configuration result by utilizing the function description information of the configuration item and the associated information, and generating an ARXML file corresponding to the configuration result. When the configuration information is determined, the module dependency relationship between the modules in the AUTOSAR is considered, so that the determined configuration information better conforms to an actual module dependency scene, and the accuracy of the configuration information of the AUTOSAR is improved. And in addition, the associated information corresponding to the function description information of the configuration item can be retrieved, so that the data used when the configuration information is determined is more comprehensive, and the accuracy of the subsequently determined configuration result is ensured.

Description

Technical Field

[0001] The present invention relates to the field of automated software configuration, and more specifically, to a method and device for determining configuration information based on AUTOSAR. Background Art

[0002] At present, AUTOSAR (AUTomotive Open System Architecture) is a standard for automotive electronic software architecture and is widely used in the software development of ECU (Electronic Control Unit) in automobiles.

[0003] During the software development process, the AUTOSAR configuration information used is usually determined manually based on experience, which is prone to errors. Especially when faced with a large number of configuration items and complex hardware architectures, the probability of errors is greater and the accuracy is low. Summary of the invention

[0004] In view of this, the present invention provides a method and device for determining configuration information based on AUTOSAR, so as to solve the problem of low accuracy of AUTOSAR configuration information determined manually based on experience.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A method for determining configuration information based on AUTOSAR, comprising:

[0007] Determine the functional description information of the configuration items based on the configuration requirement information and the module dependencies between modules in the automotive open system architecture AUTOSAR;

[0008] Retrieving association information corresponding to the functional description information of the configuration item; the association information is configuration data having an association relationship with the functional description information of the configuration item;

[0009] Determine a configuration result by using the function description information of the configuration item and the association information, wherein the configuration result includes the configuration content and the configuration reasoning process of the configuration item;

[0010] Generate an ARXML file corresponding to the configuration result.

[0011] Optionally, based on the configuration requirement information and the module dependency relationship between modules in AUTOSAR, functional description information of the configuration item is determined, including:

[0012] Get the module dependencies between modules in AUTOSAR;

[0013] Based on the configuration requirement information and the module dependency relationship, construct a first prompt word;

[0014] The large model is called to determine the functional description information of the configuration item corresponding to the first prompt word.

[0015] Optionally, the generation process of the large model includes:

[0016] Obtain chip manual and AUTOSAR tool configuration manual;

[0017] The pin names in the chip manual are used as questions, and the functional description and configuration are used as answers; and the configuration item names in the AUTOSAR tool configuration manual are used as questions, and the contents corresponding to the configuration item names are used as answers;

[0018] Using questions, answers, and engineering example data, the general model for adjusting the network structure is fine-tuned to obtain a large model.

[0019] Optionally, the general model for adjusting the network structure is fine-tuned using questions, answers, and engineering example data to obtain a large model, including:

[0020] Using the questions and the answers, first fine-tune the general model for adjusting the network structure to obtain a model to be trained;

[0021] The model to be trained is fine-tuned again using the text description in the chip manual and the text description in the AUTOSAR tool configuration manual to obtain a large model.

[0022] Optionally, retrieving the associated information corresponding to the functional description information of the configuration item includes:

[0023] Acquire multiple vector databases; the vector databases store vectorized configuration data; different vector databases have different data sources and data contents;

[0024] The functional description information of the configuration item is used as search information, and a similarity search operation is performed in the multiple vector databases to obtain associated information.

[0025] Optionally, the process of generating the vectorized configuration data includes:

[0026] Performing a data cleaning operation on the configuration data to obtain data to be processed;

[0027] Performing a text segmentation operation on the data to be processed to obtain a text segmentation result;

[0028] A vector embedding operation is performed on the text segmentation result to obtain vectorized configuration data.

[0029] Optionally, determining a configuration result by using the function description information of the configuration item and the association information includes:

[0030] Constructing a second prompt word based on the function description information of the configuration item and the associated information;

[0031] Processing the second prompt word and the configuration reasoning requirement information according to the prompt word framework to obtain a third prompt word;

[0032] The large model is called to determine the configuration result corresponding to the third prompt word.

[0033] Optionally, generating an ARXML file corresponding to the configuration result includes:

[0034] Converting the configuration information into structured data;

[0035] Use a function calling tool to generate an ARXML file corresponding to the structured data.

[0036] A configuration information determination device based on AUTOSAR, comprising:

[0037] An information determination module is used to determine the functional description information of the configuration item based on the configuration requirement information and the module dependency relationship between modules in the automotive open system architecture AUTOSAR;

[0038] An information retrieval module, used to retrieve the associated information corresponding to the functional description information of the configuration item; the associated information is configuration data having an associated relationship with the functional description information of the configuration item;

[0039] A result determination module, used to determine a configuration result by using the function description information of the configuration item and the association information, wherein the configuration result includes the configuration content and the configuration reasoning process of the configuration item;

[0040] The file generation module is used to generate an ARXML file corresponding to the configuration result.

[0041] Optionally, the information determination module includes:

[0042] The relationship acquisition submodule is used to obtain the module dependency relationship between modules in AUTOSAR;

[0043] A prompt word construction submodule, used for constructing a first prompt word based on the configuration requirement information and the module dependency relationship;

[0044] The information determination submodule is used to call the large model to determine the functional description information of the configuration item corresponding to the first prompt word.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The present invention provides a method and device for determining configuration information based on AUTOSAR. When determining configuration information, the present invention takes into account the module dependency relationship between modules in AUTOSAR, so that the determined configuration information is more in line with the actual module dependency scenario, and the accuracy of AUTOSAR configuration information is improved. In addition, in the present invention, the associated information corresponding to the functional description information of the configuration item is also retrieved, so that the data used when determining the configuration information is more comprehensive, and the accuracy of the configuration result determined subsequently is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0048] Figure 1 A flowchart of a method for determining configuration information based on AUTOSAR provided in an embodiment of the present invention;

[0049] Figure 2 A flowchart of a method for determining function description information provided by an embodiment of the present invention;

[0050] Figure 3 A flowchart of a large model training method provided by an embodiment of the present invention;

[0051] Figure 4 A schematic diagram of a scenario of data vectorization and model training provided by an embodiment of the present invention;

[0052] Figure 5 A flowchart of a data retrieval method provided by an embodiment of the present invention;

[0053] Figure 6 A flowchart of another method for determining configuration information based on AUTOSAR provided in an embodiment of the present invention;

[0054] Figure 7 A schematic diagram of the structure of a configuration information determination device based on AUTOSAR provided in an embodiment of the present invention;

[0055] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] At present, AUTOSAR (AUTomotive Open System Architecture) is a standard for automotive electronic software architecture and is widely used in the software development of ECU (Electronic Control Unit) in automobiles.

[0058] During software development, the AUTOSAR configuration information used is usually determined manually based on experience, such as using specialized configuration tools (such as EB tresos (elektrobit tresos), DaVinci, ISOLAR, and self-developed software) for manual configuration. The following problems exist:

[0059] 1) Manual configuration is complex and time-consuming: A large number of configuration items need to be set manually, and engineers need to spend a lot of time and energy on configuration details.

[0060] 2) Error-prone: Manual configuration is prone to errors, especially when dealing with complex hardware and software configurations, which can easily lead to system failures or performance issues.

[0061] In order to improve efficiency and accuracy, AI (Artificial Intelligence) models can be used to determine configuration information.

[0062] However, when existing AI models deal with professional knowledge in vertical fields such as AUTOSAR, it is often difficult to obtain high-quality, industry-specific training data, which makes it impossible to fully understand and adapt to the unique rules and logic in the field when facing complex problems in the vertical field. The accuracy and reliability are insufficient and it is difficult to meet the requirements of AUTOSAR configuration. In addition, the AI ​​model relies on the existing standard template ARXML (Autosar XML) file for configuration. At this time, the standard template ARXML file needs to be configured, which is complicated.

[0063] In order to solve the problems existing in the above-mentioned AI model, in an embodiment of the present invention, a large model is called to determine the configuration information. The large model can be directly trained using the contents in the chip manual and the AUTOSAR tool configuration manual without the need to use more refined data for training, thereby solving the problem of difficulty in obtaining high-quality training data with industry characteristics in vertical fields such as AUTOSAR.

[0064] In addition, in the present invention, a function calling tool is written to automatically generate an ARXML file, and there is no need to use a standard template ARXML file, thereby simplifying the operation.

[0065] Based on the above content, an embodiment of the present invention provides a method for determining configuration information based on AUTOSAR, and the execution subject may be an electronic device such as a processor and a server.

[0066] Reference Figure 1 , a configuration information determination method based on AUTOSAR may include:

[0067] S11. Determine functional description information of the configuration item based on configuration requirement information and module dependencies between modules in the automotive open system architecture AUTOSAR.

[0068] In practical applications, AUTOSAR software is divided into AUTOSAR CP software and AUTOSAR AP software. The present invention mainly determines configuration information for AUTOSAR CP software.

[0069] In this embodiment, when there is an AUTOSAR software configuration requirement, the user can configure the corresponding configuration requirement information, such as the user can input the project configuration requirement by voice or text, and the project configuration requirement can be, for example, to configure the configuration items of AA, BB, CC, etc. in the XX project. When setting the configuration requirement information, the user can write all the configuration items that need to be configured to ensure the comprehensive configuration of all configuration items.

[0070] In addition, users can also input project files, such as hardware schematics, DBC (Database CAN, CAN bus database files) and other information.

[0071] After the user inputs the configuration requirement information including the project configuration requirement and the project file, the user may click a confirmation button, and the electronic device may then receive the processing instruction and begin to execute step S11.

[0072] After receiving the configuration requirement information, the electronic device will obtain the module dependency relationship between the modules in AUTOSAR, such as module 2 depends on module 1, module 5 depends on modules 3 and 4, etc.

[0073] Module dependencies determine the order in which subsequent configuration items are configured. If the module where the configuration item is located needs to depend on other modules, the configuration content of the configuration items in other modules must be determined before configuring the configuration content of the configuration items in this module.

[0074] After obtaining the configuration requirement information and the module dependencies between modules in AUTOSAR, the large model can be used to determine the functional description information of the configuration items.

[0075] Among them, the large model can be a model using the Transformer architecture. The large model can output functional description information of the configuration item. The functional description information of the configuration item can be, for example: what function is the configuration item AA mainly used to achieve.

[0076] S12: Retrieve the associated information corresponding to the functional description information of the configuration item.

[0077] The association information is configuration data that is associated with the functional description information of the configuration item.

[0078] In practical applications, in order to improve the accuracy of the final configuration content, the present invention will perform a database search operation based on the functional description information of the configuration item to obtain associated information related to the functional description information of the configuration item. The associated information can come from the chip manual and the AUTOSAR tool configuration manual, and can participate in the reasoning of the large model as part of the context to improve the accuracy of the output results of the large model.

[0079] S13: Determine a configuration result by using the function description information of the configuration item and the association information.

[0080] The configuration result includes the configuration content of the configuration item and the configuration reasoning process.

[0081] The configuration content of a configuration item refers to the specific configuration data of the configuration item, which can be represented by letters, numbers, etc.

[0082] In order to provide developers with reference and decide whether additional requirements need to be added, in an embodiment of the present invention, the big model can also output a configuration reasoning process, which is the specific reasoning process by which the big model derives the configuration content of the configuration item, and can be displayed in the form of steps.

[0083] S14. Generate an ARXML file corresponding to the configuration result.

[0084] In this embodiment, the configuration result output by the large model is unstructured data and cannot be recognized by subsequent programs. Therefore, in this embodiment, the configuration information needs to be converted into structured data. In the specific implementation, the unstructured data output by the large model can be converted into structured data through the output parser for subsequent configuration reasoning and file generation. The optional technologies of the output parser include the use of JSON Schema validation, data normalization processing, and rule-based structured data generation to ensure the accuracy and standardization of the data.

[0085] After the structured data is obtained, a function call tool may be used to generate an ARXML file corresponding to the structured data.

[0086] Specifically, the tool calls the Function Call technology to call the pre-written code and generates an ARXML file corresponding to the structured data. The configuration reasoning process description in the ARXML file can be used as a reference for developers to decide whether additional requirements need to be added.

[0087] The optional technologies of Function Call in this embodiment include: automated code testing and verification based on CICD (Continuous Integration Continuous Delivery) to ensure the correctness and availability of the generated ARXML file. Through Function Call technology, a complete ARXML file can be automatically generated without using existing templates to fill in the gaps, and comprehensive configuration can be achieved.

[0088] After obtaining the ARXML file, you only need to import the generated ARXML file into the corresponding AUTOSAR CP tool to implement the configuration of the AUTOSAR CP tool.

[0089] In this embodiment, when determining the configuration information, the module dependency relationship between the modules in AUTOSAR is taken into account, so that the determined configuration information is more in line with the actual module dependency scenario, and the accuracy of the AUTOSAR configuration information is improved. In addition, in the present invention, the associated information corresponding to the functional description information of the configuration item is also retrieved, so that the data used when determining the configuration information is more comprehensive, and the accuracy of the configuration result determined subsequently is guaranteed.

[0090] In addition, in the present invention, a large model is called to determine the configuration information. The large model can be trained directly using the contents in the chip manual and the AUTOSAR tool configuration manual without the need to use more sophisticated data for training. This can solve the problem of difficulty in obtaining high-quality, industry-specific training data in vertical fields such as AUTOSAR.

[0091] In addition, in the present invention, a function calling tool is written to automatically generate an ARXML file, and there is no need to use a standard template ARXML file, thereby simplifying the operation.

[0092] In addition, the present invention constructs a complete application chain of the AUTOSAR industrial big model, which has an inspiring effect on the industrial application and practice of the big model in other fields.

[0093] In addition, the present invention directly uses the large model to obtain output without setting corresponding stop conditions, thereby avoiding the situation where the reasoning stops due to setting the stop conditions but the configuration content of the configuration item is not obtained.

[0094] In addition, in the present invention, the specific reasoning process of outputting the configuration content can enable engineers to check and understand the configuration logic, which is convenient for subsequent demand supplementation.

[0095] In another implementation of the present invention, a specific implementation of "determining functional description information of configuration items based on configuration requirement information and module dependencies between modules in AUTOSAR" is provided, referring to Figure 2 , which may include:

[0096] S21. Obtain module dependencies between modules in AUTOSAR.

[0097] In practical applications, the module dependencies between modules in AUTOSAR can be pre-configured in the database.

[0098] When determining the module dependencies between modules in AUTOSAR, the contents in the AUTOSAR tool configuration manual can be semantically understood and logically analyzed. Taking the AUTOSAR tool configuration manual as an example, an example of a configuration tool data set in the AUTOSAR tool configuration manual is as follows:

[0099] question:

[0100] What are the specific descriptions of the YYY configuration items?

[0101] Answer:

[0102] Name: YYY

[0103] Description: Parameters used to configure YYY.

[0104] Multiplicity: 1..1

[0105] Type: Enumeration parameter definition

[0106] ·Value range: X; Y; Z

[0107] Default value: X

[0108] Configuration category: Runtime configuration

[0109] Source: AUTOSAR Configuration

[0110] Dependency: ZZZ

[0111] The last item in the answer is the dependency, that is, YYY depends on ZZZ. The dependency between the configuration items can be obtained through semantic understanding and logical analysis of the contents in the AUTOSAR tool configuration manual. Since the relationship between the module and the configuration item is known, the module dependency between the modules can be determined through the dependency between the configuration items.

[0112] Since the configuration process of configuration items is complex and easily affected by the dependencies between modules, module dependencies are very important data and require high accuracy and reliability. After automatically obtaining the module dependencies, manual calibration operations can be performed to correct the incorrect content. In addition, the determination of the entire module dependency can also be performed manually.

[0113] S22: construct a first prompt word based on the configuration requirement information and the module dependency relationship.

[0114] In practical applications, since a large model is needed to determine the configuration content of a configuration item, it is necessary to construct the input of the large model. The input of the large model is generally a prompt word Prompt. In order to distinguish it from subsequent prompt words, Prompt in this embodiment is the first prompt word.

[0115] When the first prompt word is constructed based on the configuration requirement information and the module dependency relationship, prompt word engineering (Prompt Engineering) may be used to implement it, and the configuration requirement information and the module dependency relationship are analyzed by Prompt Engineering to obtain a corresponding Prompt.

[0116] S23: Call the large model to determine the functional description information of the configuration item corresponding to the first prompt word.

[0117] Specifically, the Prompt is input into the big model, and the big model can perform functional reasoning analysis on the Prompt to obtain the functional description information of the corresponding configuration item.

[0118] The functional description information of the configuration item may be, for example, what function the configuration item AA is mainly used to implement.

[0119] In another implementation of the present invention, in order to determine whether the reasoning process of the large model is accurate and whether other requirements need to be added manually, in this embodiment, in addition to outputting the functional description information of the configuration item corresponding to the first prompt word, the large model can also output the reasoning process, which is the reasoning and analysis process of how to obtain the functional description information of the configuration item from the Prompt.

[0120] Specifically, you can add configuration reasoning requirements in the input prompt, that is, tell the big model that it needs to output the reasoning process simultaneously with the output results, and the big model can output the corresponding content as required.

[0121] In this embodiment, since the large model is pre-trained, the accuracy and precision of the large model are high, and the accuracy of the functional description information of the output configuration items is high, thereby providing more accurate data input for the determination of subsequent configuration content.

[0122] In another implementation of the present invention, referring to Figure 3 , the generation process of the large model mainly includes:

[0123] S31. Obtain the chip manual and AUTOSAR tool configuration manual.

[0124] In this embodiment, two authoritative manuals in the field of AUTOSAR CP software, namely, the chip manual and the AUTOSAR tool configuration manual, are selected as training data.

[0125] The tool data set in the AUTOSAR CP tool configuration manual includes configuration item information, and the configuration item information may include configuration item name, type, description, value range, etc., and specific examples thereof refer to the above corresponding description.

[0126] The chip manual includes a pin configuration table, which includes the pin name, function description and configuration.

[0127] An example of a chip manual dataset could be:

[0128] question:

[0129] What are the configuration options for a certain pin P of chip Y?

[0130] Answer:

[0131] A pin P of chip Y has the following functions:

[0132] Port number: P, symbol: function A, control: Y1, function description: multiplex output

[0133] Port number: P, symbol: function B, control: Y2, function description: Ethernet data transmission output node

[0134] Port number: P, symbol: function C, control: Y3, function description: CAN data transmission

[0135] Port number: P, Symbol: Function D, Control: Y4, Function description: General input / output

[0136] In addition to the aforementioned chip manual and AUTOSAR tool configuration manual, since engineering examples have been configured in the AUTOSAR CP software in the past, previous engineering examples can be obtained as part of the training data.

[0137] Examples of engineering case datasets may be:

[0138] question:

[0139] If you want to meet {condition 1}, {condition 2}, etc., how do you configure the YYY function of a pin of chip Y to meet the requirements?

[0140] Answer:

[0141] According to {Condition 1}, {Condition 2}, etc., the function of a pin YYY of chip Y needs to be configured as Z to meet the specific functions {Function A}, {Function B}, etc. in {Condition 1}, {Condition 2}, etc.

[0142] Subsequently, the configured engineering instances will be used as data sets for supervised training to enhance the reasoning capabilities of large models in vertical fields.

[0143] S32. Use the pin names in the chip manual as questions, and use the function description and configuration as answers; and use the configuration item names in the AUTOSAR tool configuration manual as questions, and use the contents corresponding to the configuration item names as answers.

[0144] In this embodiment, in order to enable the large model to have knowledge related to AUTOSAR configuration, the contents of the chip manual and the AUTOSAR tool configuration manual are fed to the large model, so that the large model can learn knowledge related to AUTOSAR configuration and thus have the ability to determine the AUTOSAR configuration content.

[0145] When feeding the contents of the chip manual and the AUTOSAR tool configuration manual to the big model, they need to be processed in some way. Since the big model uses the question-and-answer method when performing reasoning, the contents of the chip manual and the AUTOSAR tool configuration manual also need to be processed into a question-and-answer mode. The above example is an example that has been processed.

[0146] In the specific processing, the pin configuration table is extracted from the chip manual, and the pin name is used as the question, and its function description and configuration are used as the answer. At the same time, the configuration item information is extracted from the configuration manual of the AUTOSAR CP tool, and the configuration item name is used as the question, and the content corresponding to the configuration item name, such as type, description, value range, etc., is used as the answer.

[0147] When extracting questions and answers, you can manually select the extraction range and then call the data extraction program to extract the corresponding data from the chip manual and the configuration manual of the AUTOSAR CP tool.

[0148] S33. Using the questions, the answers and the engineering example data, fine-tune the general model for adjusting the network structure to obtain a large model.

[0149] In this embodiment, in order to improve the training efficiency of the model, the general model can be fine-tuned. Since the general model has been pre-trained and has a certain reasoning ability, it only needs to be fine-tuned using the training data in the AUTOSAR CP field to enable it to have the reasoning ability in the AUTOSAR CP field and reduce the amount of sample data.

[0150] Before fine-tuning the general model, the network structure of the general model can also be adjusted according to the task conditions and data sets in the AUTOSAR CP field to optimize the model structure of the general model so that it can be better adapted to the AUTOSAR CP field.

[0151] Specifically, in terms of model structure optimization, since different tasks and data sets require different network structures, the original network structure can be modified according to the data characteristics of the chip manual and the relevant CP tool configuration manual. The possible ways to change the network structure include changing the number of neurons, changing the thickness and depth of the layer, changing the activation function of the neuron, changing the attention mechanism, using residual connections, etc., so as to obtain a general model for adjusting the network structure.

[0152] Subsequently, the model parameters of the general model for adjusting the network structure are adjusted. During the adjustment, the prepared data set, i.e., the above-mentioned questions, the answers and the engineering example data, is used to fine-tune the large model to obtain the large model.

[0153] In the specific implementation, LoRA fine-tune (LoRA: Low-Rank Adaptation) technology in supervised learning is used for model training to save time and resources.

[0154] Among them, the model training in this embodiment refers to adjusting the parameters of the model based on the training data set by learning the rules and logic in the data set to make it perform better in the data set. Fine-tuning is a way of model training, which improves the understanding of the model in the vertical field by adjusting only a small part of the model's parameters. Compared with full parameter adjustment, fine-tuning the model is low-cost, fast, and will not have a significant impact on the generalization ability of the model.

[0155] In another implementation of the present invention, the fine-tuning can be implemented in two steps, as follows:

[0156] 1) Using the questions and the answers, a general model for adjusting the network structure is fine-tuned for the first time to obtain a model to be trained.

[0157] The specific implementation of this step refers to the above description.

[0158] 2) Using the text description in the chip manual and the text description in the AUTOSAR tool configuration manual, fine-tune the model to be trained again to obtain a large model.

[0159] Using self-supervised learning, the text descriptions of the chip and the AUTOSARCP tool in the manual are used as training sets to further optimize the model's generalization ability in this field.

[0160] It should be noted that, refer to Figure 4 In addition to using the data in the chip manual and AUTOSAR tool configuration manual as training sets, you can also use other available manuals, such as company development specifications, etc., and use all available manuals and configured project content (such as the above engineering examples) to create data sets and use supervised learning for fine-tuning. Later, you can use the text description summarized in the manual to fine-tune the model again using self-supervised learning to obtain a large model that can be used in the end.

[0161] It should be noted that you can also perform fine-tuning of self-supervised learning first, and then fine-tune supervised learning, or perform both at the same time. In the specific implementation, you can choose the appropriate method according to your needs.

[0162] In this embodiment, the large model is trained by two fine-tuning methods, so that the large model has better reasoning and analysis capabilities in the AUTOSAR field.

[0163] In addition, in the present invention, only questions and answers need to be extracted from the manual to train the large model. When the AI ​​model is trained, the questions and answers need to be further processed to obtain the training data of the AI ​​model. In this embodiment, the use of a large model can simplify the preparation process of training data and reduce the complexity of training.

[0164] After obtaining the functional description information of the configuration item using the large model, the associated information corresponding to the functional description information of the configuration item can also be retrieved to provide data support for subsequent reasoning.

[0165] Specifically, refer to Figure 5 , retrieving the associated information corresponding to the functional description information of the configuration item, which may include:

[0166] S41. Acquire multiple vector databases.

[0167] The vector database stores vectorized configuration data.

[0168] In this embodiment, the manuals are divided into chip manuals, AUTOSAR tool configuration manuals, and other available manuals. Figure 4 , for different manuals, corresponding vector databases can be configured, such as chip vector database, AUTOSAR CP tool vector database and other vector databases. That is, the data sources and data contents in different vector databases are different.

[0169] Taking the chip vector database as an example, the configuration data in the chip manual is obtained. The content of the configuration data is similar to the content in the above chip manual data set, but the format is different. For example, the chip pin and function description information are extracted from the chip manual, the configuration data is vectorized, and the vectorized configuration data is obtained, and the vectorized configuration data is stored in the chip vector database.

[0170] For the AUTOSAR tool configuration manual, the description information related to the configuration items can be extracted from the configuration manual of the AUTOSAR CP tool, and then vectorized operations can be performed.

[0171] Specifically, the process of performing vectorization processing on the configuration data to obtain the vectorized configuration data may be as follows:

[0172] 1) Performing a data cleaning operation on the configuration data to obtain data to be processed.

[0173] In order to ensure the reliability and accuracy of the data, data cleaning operations can be performed on it. Data cleaning refers to the process of re-examining and verifying the data, aiming to discover and correct errors in the data files, including checking data consistency, processing invalid values ​​and missing values. This process is usually completed by computers, not humans, to ensure the accuracy and reliability of the data. The goal of data cleaning is to improve data quality and ensure the uniqueness, integrity, consistency and validity of the data.

[0174] 2) Performing a text segmentation operation on the data to be processed to obtain a text segmentation result.

[0175] Specifically, text segmentation is to divide the data to be processed into shorter sentences for subsequent embedding operations.

[0176] When performing text segmentation, the text segmentation can be performed by word segmentation or character segmentation to obtain a text segmentation result. The text segmentation result includes text blocks, and the text blocks can be words or characters.

[0177] 3) Performing a vector embedding operation on the text segmentation result to obtain vectorized configuration data.

[0178] Specifically, the text blocks in the text segmentation result are vectorized using an embedding model to obtain vectorized configuration data, which is then stored using a corresponding vector database.

[0179] In this embodiment, a vector database (VectorDB) is established according to the configuration manual of the chip and AUTOSAR CP tool to provide accurate reference information during the reasoning process. In addition, additional databases are established using configuration manuals of other tools, company development specifications and other documents to improve the correctness of reasoning and the standardization of configuration.

[0180] S42: Use the function description information of the configuration item as search information, perform a similarity search operation in the multiple vector databases, and obtain associated information.

[0181] Specifically, the enhanced retrieval generation (RAG) link can be used for retrieval. During the retrieval, the functional description information of the above configuration items is used as the retrieval information of RAG, and similarity retrieval is performed in each of the above VectorDBs to obtain accurate information about the relevant chip and AUTOSAR CP configuration items, and output content related to the top K (K is set according to the actual scenario) configuration items. The information output by RAG is the associated information in this embodiment, which can be used as part of the context to participate in the reasoning of the subsequent large model, so that the large model can better understand the problem and improve the accuracy of the answer.

[0182] In an optional implementation, RAG may use multimodal information retrieval to retrieve multimodal data and enrich the types of retrieval results to improve the comprehensiveness and accuracy of information retrieval.

[0183] In this embodiment, the associated information is retrieved based on the RAG, and the associated information is used as part of the context to participate in the reasoning of the subsequent large model, so that the large model can better understand the problem and improve the accuracy of the answer.

[0184] Specifically, when the configuration result is determined by using the function description information of the configuration item and the associated information, the implementation process is as follows:

[0185] 1) Constructing a second prompt word based on the function description information of the configuration item and the associated information.

[0186] The process of constructing the second prompt word is similar to the process of constructing the first prompt word. In this embodiment, the context-aware prompt design method is adopted to construct the prompt of the large model input using the associated information as context information.

[0187] Specifically, in order to further improve the accuracy of the big model's answer, some context information can be added as input when asking questions to the big model to increase the big model's understanding of the question. The source of context includes but is not limited to the relevant information obtained in the vector database using the enhanced retrieval mentioned above.

[0188] 2) Processing the second prompt word and the configuration reasoning requirement information according to the prompt word framework to obtain a third prompt word.

[0189] In this embodiment, in order to ensure the standardization of Prompt generation, a template-based Prompt generation method can be adopted. In addition, in order to enable the large model to generate the configuration reasoning process, the configuration reasoning requirement information also needs to be input into the model, that is, when constructing the Prompt, the above-mentioned second prompt word and configuration reasoning requirement information need to be used.

[0190] In the template-based Prompt generation method, a prompt word framework is designed. Non-general information such as chip pin names and function descriptions can be used as variables to create prompt word templates. Different prompt words can be obtained by using different variables to achieve the purpose of reuse.

[0191] When using the template-based prompt generation method, the second prompt word and the configuration reasoning requirement information are processed according to the prompt word framework to obtain the corresponding third prompt word.

[0192] Specifically, the design of prompt words is mainly the design of prompt word framework. The input, output and constraint conditions are clarified based on the task being performed. The input is the question asked to the large model, including the chip pin name, required function, corresponding configuration items in the AUTOSAR CP tool, etc. The output is to clarify the answer that the large model needs to give, that is, the clear configuration information of the configuration items. Constraints are the constraints of the large model when processing input information, reasoning and deduction, and output results. After clarifying the above information, it is integrated into the system prompt words, and then continuously evaluated and iterated in the future to optimize the prompt words and obtain the final third prompt words.

[0193] In this embodiment, context-aware prompt design and template-based prompt generation are used to improve the adaptability of the model to different inputs and the accuracy of the output.

[0194] 3) Calling the large model to determine the configuration result corresponding to the third prompt word.

[0195] In this embodiment, after designing an appropriate prompt based on prompt engineering, the fine-tuned large model can output the configuration items required to complete a specific function based on the input information and record the reasoning process. In order to make the large model output the reasoning process, it is necessary to require the large model to give the reasoning process when answering questions in prompt engineering.

[0196] It should be noted that when constructing the third prompt word, module dependencies can also be added, so that potential interdependencies between modules can be considered when using the reasoning ability of the large model. Specifically, the configuration manual of the AUTOSAR CP tool describes the connections between the configuration items of each module. These connection information is used as part of the subsequent prompt words, so the model can obtain the final configuration information from a global perspective during reasoning. When the requirements of a certain module change, the large model can be re-reasoned to obtain the final result. The present invention can simplify the configuration process and improve the integrity and accuracy of the configuration.

[0197] In addition, the output file contains a description of the reasoning configuration process, which enables engineers to understand the reasoning process and decide whether additional requirements need to be added.

[0198] The entire large model usage process is as follows Figure 6 Specifically, the configuration requirements of the project and the configuration requirement information of the project files and the module dependency are used as the input of the big model, and functional reasoning is performed inside the big model to obtain the reasoning result, which is the functional description information of the above-mentioned configuration items. The functional description information of the configuration items is the configuration requirement that can enable the big model to be reused in the future.

[0199] Then, similarity search is performed in multiple vector databases to obtain related information, the functional description information of the configuration item and the related information (as context information) are integrated, and questions are asked to the large model again to obtain configuration item configuration and RAG configuration item configuration reasoning, wherein the configuration item configuration is the configuration content of the above-mentioned configuration item, and the RAG configuration item configuration reasoning is the above-mentioned configuration reasoning process.

[0200] Subsequently, structured conversion is performed and the structured data is used as the parameter of FunctionCall. The large model calls (FunctionCall) to generate the required ARXML configuration file. The generated ARXML file is imported into the corresponding AUTOSAR CP tool to realize the automatic configuration of the AUTOSAR CP tool.

[0201] In this embodiment, the time and workload of manual configuration are greatly reduced through the use of large models and automated tools, and efficiency and quality are improved. In addition, the automated configuration process reduces human errors and improves the accuracy and reliability of configuration. In addition, through fine-tuning and data set preparation, the reasoning ability of the large model in the field of AUTOSAR CP configuration is improved.

[0202] Based on the above-mentioned embodiment of the method for determining configuration information based on AUTOSAR, another embodiment of the present invention provides a device for determining configuration information based on AUTOSAR, referring to Figure 7 ,include:

[0203] An information determination module 11, used to determine functional description information of a configuration item based on configuration requirement information and module dependencies between modules in an automotive open system architecture AUTOSAR;

[0204] An information retrieval module 12 is used to retrieve the associated information corresponding to the functional description information of the configuration item; the associated information is configuration data having an associated relationship with the functional description information of the configuration item;

[0205] A result determination module 13, configured to determine a configuration result by using the function description information of the configuration item and the association information, wherein the configuration result includes the configuration content and the configuration reasoning process of the configuration item;

[0206] The file generating module 14 is used to generate an ARXML file corresponding to the configuration result.

[0207] In one implementation, the information determination module includes:

[0208] The relationship acquisition submodule is used to obtain the module dependency relationship between modules in AUTOSAR;

[0209] A prompt word construction submodule, used for constructing a first prompt word based on the configuration requirement information and the module dependency relationship;

[0210] The information determination submodule is used to call the large model to determine the functional description information of the configuration item corresponding to the first prompt word.

[0211] In one implementation, a model generation module is further included, and the model generation module includes:

[0212] Manual acquisition submodule, used to obtain chip manuals and AUTOSAR tool configuration manuals;

[0213] A data determination submodule, used to use the pin name in the chip manual as a question, and the function description and configuration as an answer, and to use the configuration item name in the AUTOSAR tool configuration manual as a question, and the content corresponding to the configuration item name as an answer;

[0214] The model fine-tuning submodule is used to fine-tune the general model of adjusting the network structure using questions, answers and engineering example data to obtain a large model.

[0215] In one implementation, the model fine-tuning submodule includes:

[0216] A first fine-tuning unit, used to perform first fine-tuning on the general model for adjusting the network structure using the question and the answer to obtain a model to be trained;

[0217] The re-fine-tuning unit is used to use the text description in the chip manual and the text description in the AUTOSAR tool configuration manual to re-fine-tune the model to be trained to obtain a large model.

[0218] In one implementation, the information retrieval module 12 includes:

[0219] A database acquisition submodule is used to acquire multiple vector databases; the vector databases store vectorized configuration data; the data sources and data contents in different vector databases are different;

[0220] The retrieval submodule is used to use the functional description information of the configuration item as retrieval information, perform a similarity retrieval operation in the multiple vector databases, and obtain associated information.

[0221] In one implementation, a data configuration module is also included, and the data configuration module includes:

[0222] A cleaning submodule, used for performing a data cleaning operation on the configuration data to obtain data to be processed;

[0223] The slicing submodule is used to perform a text slicing operation on the data to be processed to obtain a text slicing result;

[0224] The vector embedding submodule is used to perform a vector embedding operation on the text segmentation result to obtain vectorized configuration data.

[0225] In one implementation, the result determination module 13 includes:

[0226] A first prompt word processing submodule, configured to construct a second prompt word based on the function description information of the configuration item and the association information;

[0227] A second prompt word processing submodule, used for processing the second prompt word and the configuration reasoning requirement information according to the prompt word framework to obtain a third prompt word;

[0228] The result determination submodule is used to call the large model to determine the configuration result corresponding to the third prompt word.

[0229] In one implementation, the file generation module 14 is specifically used for:

[0230] The configuration information is converted into structured data, and an ARXML file corresponding to the structured data is generated using a function calling tool.

[0231] In this embodiment, when determining the configuration information, the module dependency relationship between the modules in AUTOSAR is taken into account, so that the determined configuration information is more in line with the actual module dependency scenario, and the accuracy of the AUTOSAR configuration information is improved. In addition, in the present invention, the associated information corresponding to the functional description information of the configuration item is also retrieved, so that the data used when determining the configuration information is more comprehensive, and the accuracy of the configuration result determined subsequently is guaranteed.

[0232] It should be noted that, for the working process of each module, sub-module and unit in the embodiment of the present invention, please refer to the corresponding description in the above embodiment, which will not be repeated here.

[0233] The present application also provides an electronic device in an embodiment. Figure 8 As shown, it shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiment of the present application. The electronic device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 8 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0234] like Figure 8As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 to a random access memory (RAM) 603 to implement the above-mentioned AUTOSAR-based configuration information determination method. When the electronic device is powered on, various programs and data required for the operation of the electronic device are also stored in the RAM 603. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0235] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a memory card, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 8 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0236] An embodiment of the present application also provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any AUTOSAR-based configuration information determination method provided in the embodiment of the present application.

[0237] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any AUTOSAR-based configuration information determination method provided in an embodiment of the present application.

[0238] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining configuration information based on AUTOSAR, characterized in that: include: Determine the functional description information of the configuration items based on the configuration requirement information and the module dependencies between modules in the automotive open system architecture AUTOSAR; Retrieving association information corresponding to the functional description information of the configuration item; the association information is configuration data having an association relationship with the functional description information of the configuration item; Determine a configuration result by using the function description information of the configuration item and the association information, wherein the configuration result includes the configuration content and the configuration reasoning process of the configuration item; Generate an ARXML file corresponding to the configuration result.

2. The configuration information determination method based on AUTOSAR according to claim 1, characterized in that: Based on the configuration requirement information and the module dependencies between modules in AUTOSAR, determine the functional description information of the configuration item, including: Get the module dependencies between modules in AUTOSAR; Based on the configuration requirement information and the module dependency relationship, construct a first prompt word; The large model is called to determine the functional description information of the configuration item corresponding to the first prompt word.

3. The configuration information determination method based on AUTOSAR according to claim 2, characterized in that: The generation process of the large model includes: Obtain chip manual and AUTOSAR tool configuration manual; The pin names in the chip manual are used as questions, and the functional description and configuration are used as answers; and the configuration item names in the AUTOSAR tool configuration manual are used as questions, and the contents corresponding to the configuration item names are used as answers; Using questions, answers, and engineering example data, the general model for adjusting the network structure is fine-tuned to obtain a large model.

4. The configuration information determination method based on AUTOSAR according to claim 3, characterized in that: Using questions, answers, and engineering example data, we fine-tune the general model for adjusting the network structure to obtain a large model, including: Using the questions and the answers, first fine-tune the general model for adjusting the network structure to obtain a model to be trained; The model to be trained is fine-tuned again using the text description in the chip manual and the text description in the AUTOSAR tool configuration manual to obtain a large model.

5. The configuration information determination method based on AUTOSAR according to claim 1, characterized in that: Retrieving the associated information corresponding to the functional description information of the configuration item includes: Acquire multiple vector databases; the vector databases store vectorized configuration data; different vector databases have different data sources and data contents; The functional description information of the configuration item is used as search information, and a similarity search operation is performed in the multiple vector databases to obtain associated information.

6. The method for determining configuration information based on AUTOSAR according to claim 5, characterized in that: The process of generating the vectorized configuration data includes: Performing a data cleaning operation on the configuration data to obtain data to be processed; Performing a text segmentation operation on the data to be processed to obtain a text segmentation result; A vector embedding operation is performed on the text segmentation result to obtain vectorized configuration data.

7. The configuration information determination method based on AUTOSAR according to claim 1, characterized in that: Determining a configuration result by using the function description information of the configuration item and the association information includes: Constructing a second prompt word based on the function description information of the configuration item and the associated information; Processing the second prompt word and the configuration reasoning requirement information according to the prompt word framework to obtain a third prompt word; The large model is called to determine the configuration result corresponding to the third prompt word.

8. The configuration information determination method based on AUTOSAR according to claim 1, characterized in that: Generate an ARXML file corresponding to the configuration result, including: Converting the configuration information into structured data; Use a function calling tool to generate an ARXML file corresponding to the structured data.

9. A configuration information determination device based on AUTOSAR, characterized in that: include: An information determination module is used to determine the functional description information of the configuration item based on the configuration requirement information and the module dependency relationship between modules in the automotive open system architecture AUTOSAR; An information retrieval module, used to retrieve the associated information corresponding to the functional description information of the configuration item; the associated information is configuration data having an associated relationship with the functional description information of the configuration item; A result determination module, used to determine a configuration result by using the function description information of the configuration item and the association information, wherein the configuration result includes the configuration content and the configuration reasoning process of the configuration item; The file generation module is used to generate an ARXML file corresponding to the configuration result.

10. The AUTOSAR-based configuration information determination device according to claim 9, characterized in that: The information determination module comprises: The relationship acquisition submodule is used to obtain the module dependency relationship between modules in AUTOSAR; A prompt word construction submodule, used for constructing a first prompt word based on the configuration requirement information and the module dependency relationship; The information determination submodule is used to call the large model to determine the functional description information of the configuration item corresponding to the first prompt word.

Images