SysML model generation method based on large language model in conceptual design

Abstract

The invention provides a SysML model generation method based on a large language model in conceptual design, which comprises the following steps of: guiding LLM by utilizing a thinking chain cue word template, and converting user input into a conceptual design scheme described in a plain text form; utilizing a cue word template I to guide LLM to split the conceptual design scheme into a sentence list containing complete semantics; using a cue word template 2 to guide LLM to extract entities and relationships thereof from each sentence, and outputting a triple; based on the triple, creating a SysML model hierarchical structure according to a predefined rule; and creating an XMI file by following an XMI format specification to obtain a complete SysML model. According to the method, the automation degree and generalization of SysML model generation are remarkably improved, less text input can be effectively processed, a conceptual design scheme model which can be directly used for actual system design and adapts to products is generated, and the method is particularly suitable for high-complexity large-scale system design.

Description

Technical Field

[0001] The present invention relates to the technical field of model-based systems engineering, and in particular, to a method for generating a SysML model based on a large language model in conceptual design. Background Art

[0002] Conceptual design consists of a series of orderly, organized, and goal-oriented design activities from analyzing user requirements to generating a conceptual product. It is manifested as an evolving process from rough to refined, from fuzzy to clear, and from abstract to concrete. As a key stage in the design process, the quality of conceptual design directly affects the innovation and feasibility of the final solution. However, current conceptual design faces many problems. On the one hand, in traditional conceptual design, designers are limited by personal experience and knowledge reserves when exploring solutions, and it is difficult to efficiently process a large amount of fragmented data and cross-domain product conceptual design. On the other hand, in document-based systems engineering design, due to the ambiguity and complexity of natural language descriptions, it is difficult to communicate design solutions among stakeholders, and it is difficult to maintain the document to keep consistency during the modification and iteration of the design solution.

[0003] With the in-depth development of modern systems engineering practice, system models are showing increasingly complex characteristics. For example, highly complex large systems such as aircraft systems, automotive systems, and aerospace systems. These complex systems not only contain a large number of components and subsystems, but also the interaction logic between components is intricate, resulting in more diverse product variability. How to efficiently design complex systems and generate solutions suitable for conceptual design is an urgent problem to be solved.

[0004] Model-based systems engineering uses modeling languages such as SysML to replace documents to improve design efficiency, but its modeling process still highly relies on manual operations, resulting in a large amount of time consumed on repetitive tasks. In summary, current conceptual design and modeling highly rely on manual operations. Automating the conceptual design and modeling process, that is, automatically generating a SysML model, can save a large amount of human resources and time costs, which is of great significance.

[0005] Currently, there are mainly two ways to automatically generate a SysML model: one is based on traditional natural language processing (NLP) technology and automatically generates a SysML model in a rule template manner. The automation degree of this method is insufficient and the requirements for input are high; the other is based on traditional deep learning technology and automatically generates a SysML model in a model plus fine-tuning manner. This method has a high degree of automation, but it relies on a professional domain dataset. Specifically, the current methods for automatically generating a SysML model have the following deficiencies: 1. Additional professional domain datasets need to be prepared. Traditional methods require preparing specific professional domain datasets to fine-tune the model. Different datasets need to be prepared for different domains to achieve good performance in specific domain tasks. Or it is necessary to construct a domain knowledge graph based on the professional dataset and generate a SysML model on the basis of the knowledge graph assistance. The process of preparing datasets and constructing knowledge graphs is very time-consuming in terms of human resources and cost. Without the support of additional professional domain datasets, the generalization of the method is poor.

[0006] 2. The automatically generated modeling results are difficult to be applied in practice. Traditional methods use drawing software to visually express the modeling results, and the generated results cannot be directly applied to actual modeling software, which is not convenient for designers to carry out subsequent design and modification on the initially generated modeling scheme.

[0007] 3. The requirements for input are relatively strict. The input of traditional methods is not natural language text, but structured text with a certain format; or the input text must meet certain rules, templates and specifications; or the input needs to include an existing model as input assistance. It is difficult to handle the situation where the input is only natural language expression, and it is even more impossible to handle the situation where the input text is less, such as only the product name.

[0008] 4. The degree of automation is insufficient. Traditional NLP technology methods use custom heuristic rules or rule algorithms, combined with traditional technical tools, to achieve the automatic generation of SysML models. In the main step of extracting SysML model elements from the conceptual design scheme text, the limited custom rules of this method are difficult to apply to infinite types of natural language texts, the cost of maintaining rules is high, and the generalization of rules is very poor; customizing rules depends on manual operations and requires more professional domain knowledge, and the automation level is low. Traditional deep learning methods also require manual participation in processes such as constructing datasets and fine-tuning models, and the degree of automation needs to be further improved.

[0009] Therefore, there is an urgent need for a SysML model generation method and system with higher automation, stronger generalization, the ability to process less text input, and modeling results that can be applied in practice. With the development of deep learning technology, especially the breakthrough progress made by large language models in the field of natural language processing, it provides new ideas and technical means for solving the above technical problems. Summary of the Invention

[0010] The purpose of the present invention is to provide a method for generating a SysML model based on a large language model in conceptual design. By utilizing the powerful knowledge reserve, semantic understanding and text generation capabilities of the large language model, it realizes the generation of a SysML model with higher automation, stronger generalization, the ability to process less text input, and modeling results that can be applied in practice.

[0011] A method for generating a SysML model based on a large language model in conceptual design provided by an embodiment includes the following steps: Input the user requirements related to conceptual design into the large language model, and use a preset chain-of-thought prompt template to guide the large language model to output a conceptual design solution described in plain text; Use a preset prompt template one to guide the large language model to divide the conceptual design solution described in plain text into sentences, and output a list of sentences containing complete semantics; Use a preset prompt template two to guide the large language model to extract entity relationships for each sentence in the list of sentences containing complete semantics, and output triples containing entities and their relationships; Create a SysML model hierarchical structure according to the triples containing entities and their relationships according to predefined rules; Create an XMI file according to the SysML model hierarchical structure according to the XMI format specification representing SysML model information, and obtain a complete SysML model.

[0012] In one embodiment, the user requirements related to conceptual design are a product name noun in a highly complex large system, or a product name noun and a simple requirement description of the product.

[0013] Further, the highly complex large system includes one or more of: an aircraft system, an automotive system, and an aerospace system.

[0014] In one embodiment, the parameter scale of the large language model is greater than 100 B; the large language model is DeepSeek-V3.

[0015] In one embodiment, the preset chain-of-thought prompt template is a few-shot chain-of-thought prompt template, including two parts: a chain-of-thought sample prompt template and an actual problem prompt template; The chain-of-thought sample prompt template contains 2-10 chain-of-thought samples, and the prompt template for each chain-of-thought sample includes: a problem prompt template for guiding the large language model to generate a product conceptual design solution, a fixed prompt for guiding the large language model to perform chain-of-thought reasoning, sample conceptual design solution chain-of-thought reasoning steps for guiding the large language model, and a fixed prompt for guiding the large language model to summarize and obtain a conceptual design solution; The actual problem prompt template includes: a problem prompt template for guiding the large language model to generate a product conceptual design solution and a fixed prompt for guiding the large language model to perform chain-of-thought reasoning.

[0016] In one embodiment, the sample conceptual design solution for guiding the large language model to perform chain of thought reasoning includes four mapping reasoning steps, specifically including: product-requirement mapping, requirement-function mapping, function-behavior mapping, and behavior-structure mapping. A corresponding mapping prompt word template is set for each mapping reasoning step.

[0017] In one embodiment, the conceptual design solution described in pure text form includes the chain of thought reasoning process and the structural information of the SysML model, where the structural information of the SysML model includes module composite relationships and component attribute connector relationships.

[0018] In one embodiment, the preset prompt word template one includes: large language model role definition, the sentence segmentation task goal to be achieved, examples of solving the sentence segmentation task, and specific descriptions.

[0019] In one embodiment, the preset prompt word template two includes: large language model role definition, the entity relationship extraction task goal to be achieved, examples of solving the entity relationship extraction task, and specific descriptions.

[0020] In one embodiment, the triple of the entity and its relationship includes the triple of the module and its composite relationship and the triple of the component attribute and its connector relationship.

[0021] In one embodiment, the predefined rules include: rules for triple connection, rules for triple deduplication, and rules for enhancing the SysML model hierarchical structure; For the rules of triple connection, the product name is used as the top node of the SysML model hierarchical structure. Triples are connected according to the entity name and added to the SysML model hierarchical structure one by one; if the triple is a triple of the module and its composite relationship, the tail entity is used as the child node and added to the next level of the head entity node of the SysML model hierarchical structure. If the head entity node does not appear in the SysML model hierarchical structure, the head entity node is newly created; if the triple is a triple of the component attribute and its connector relationship, the tail entity in the triple is added to the next level of the head entity node of the SysML model hierarchical structure, and a connector label is added to identify the connector relationship. If the head entity node does not appear in the SysML model hierarchical structure, the head entity node is newly created; The rules for triple deduplication include: for the triple of the module and its composite relationship, when the head entity, relationship, and tail entity of the triple are the same, it is judged as duplicate; for the triple of the component attribute and its connector relationship, when the relationship of the triple is the same and the name sets of the head entity and the tail entity are the same, it is judged as duplicate; The rules for enhancing the SysML model hierarchy structure include: after all triples are connected, connect the top-level nodes of multiple hierarchy structures to the unique top-level node of the SysML model hierarchy structure; for all component attribute and their connector relationship triples, create a module corresponding to the component attribute, and if the module does not appear in the SysML model hierarchy structure, add it to the SysML model hierarchy structure and connect it to the top-level node, otherwise ignore the module.

[0022] In one embodiment, creating an XMI file includes: mapping SysML model elements to XMI tags, adding SysML model graphical information, and an algorithm for creating XMI tags according to the SysML model hierarchy structure; The mapping of the SysML model elements to XMI tags includes: the mapping of modules, composite relationships, component attributes, and connector relationships in the SysML model to the corresponding tags in XMI; The adding of SysML model graphical information includes: the name of the frame, the module it belongs to, the display position, the size of the frame, the display positions, explicit sizes, and display styles of the module and component attributes, and the display positions, display lengths, and display styles of the connection lines of the composite relationships and connector relationships; The algorithm for creating XMI tags according to the SysML model hierarchy structure includes: creating a module definition diagram, all the modules in the module definition diagram, and the composite relationships between the modules according to the SysML model hierarchy structure, then creating the component attributes under the module according to the composite relationship information of the module, then creating an internal module diagram and connector relationships according to the component attributes and their connector relationships, and finally creating the corresponding extended graphical nodes under the XMI root node according to each existing module definition diagram and internal module diagram.

[0023] Compared with the prior art, the beneficial effects of the present invention at least include: (1) Reducing the dependence on professional field data sets. The present invention uses the rich knowledge base reserve of the large language model to generate concept design schemes and automatically models to generate a SysML model based on this. The whole process does not require manual construction of additional professional field data sets, and the large language model can be used to generate concept design schemes for product names in different fields.

[0024] (2) The automatically generated SysML model can be imported into modeling software, enhancing its practicability. The present invention uses a general XMI format specification file to represent the SysML model, which can be imported into common practical modeling software that supports the XMI format, facilitating designers to perform subsequent further modeling and modification, and being more practical.

[0025] (3) Few requirements for input, and can process single-noun input. The input of the present invention is the product name described in natural language or the product name and its simple requirement description, and it can generate a relatively rich and complete SysML model starting from a single noun; there is no restriction on the input, and even if there is only one product name, it can automatically generate the SysML model corresponding to the conceptual design solution of the product name.

[0026] (4) Significantly improve the degree of automation. The main steps of the whole process of the present invention use the method of large language model plus prompt template, and subsequently use rule-based methods and algorithms for auxiliary processing, which can be embedded in practical modeling software, and can realize the one-step generation of the SysML model corresponding to the conceptual design solution from the product name. The whole process does not require manual participation, and there is no need for additional steps such as manually constructing data sets and knowledge graphs, and the degree of automation has been further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0028] Figure 1 It is a schematic flowchart of the method for generating a SysML model based on a large language model in the conceptual design provided by the embodiment.

[0029] Figure 2 It is a schematic structural diagram of the preset chain of thought prompt template provided by the embodiment.

[0030] Figure 3 It is a schematic structural diagram of the preset prompt template 1 provided by the embodiment.

[0031] Figure 4 It is a schematic structural diagram of the preset prompt template 2 provided by the embodiment.

[0032] Figure 5 It is a schematic structural diagram of the predefined rules provided by the embodiment.

[0033] Figure 6 It is a schematic structural diagram of the XMI format specification provided by the embodiment.

[0034] Figure 7 It is a module definition diagram of the SysML model provided by the embodiment.

[0035] Figure 8 It is an internal module diagram of the SysML model provided by the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] To make the objectives, technical solutions, and advantages of the present invention more clear and understandable, the following further details the present invention in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely used to explain the present invention and do not limit the protection scope of the present invention.

[0037] The solution of the embodiment of the present invention is as Figure 1 shown. A method for generating a SysML model based on a large language model in conceptual design, as Figure 1 shown, includes the following steps: S1. Input the user requirements related to conceptual design into the large language model, and use a preset chain-of-thought prompt template to guide the large language model to output a conceptual design solution described in pure text form.

[0038] In the embodiment, the input of the user requirements related to conceptual design is a simple product name noun, such as: sonar, drone, etc., highly complex large systems, such as: product name nouns of aircraft systems, automotive systems, or aerospace systems; or a product name noun and its simple requirement description, such as: sonar needs to be able to detect the position and distance of underwater objects, that is, detailed requirement descriptions are not given in the input.

[0039] Traditional methods all carry out subsequent work based on detailed requirement texts, and requirement analysis is the work content of designers. The input of the present invention is a single product name noun, which can further liberate the hands of designers and improve the degree of automation. If there is a very detailed requirement description text in the input, this step can be omitted and directly start from step S2.

[0040] To make the generation effect of the present invention optimal, the parameter scale of the large language model used in this step and subsequent steps S2 and S3 should be greater than 100 B. In the embodiment, the DeepSeek-V3 large language model can be used. The DeepSeek-V3 large language model has a total of 671 B parameters and has efficient reasoning ability. Its performance is better than other open-source models and achieves performance comparable to leading closed-source models. A large language model with a parameter scale of more than 100 B can better use prompts to induce its logical reasoning ability, especially the chain-of-thought method used in this step is more suitable for language models with a large number of parameters. Using a model with a parameter scale greater than 100 B, the knowledge reserve, semantic understanding ability, and text generation ability of the model are at a relatively high level, can effectively handle single product name input and cross-domain product concept design situations, can ensure the quality of the generated conceptual design solution, and solve the problems of being limited by designers' experience and knowledge reserve when exploring solutions, and designers' difficulty in efficiently processing massive fragmented data and cross-domain product concept design.

[0041] AsFigure 2 As shown, in this step, the preset thought chain prompt template is a few-shot thought chain prompt template, which includes two parts: a thought chain sample prompt template and an actual problem prompt template.

[0042] The thought chain sample prompt template contains 2-10 samples, and the prompt template for each sample includes: a question prompt template for guiding the large language model to generate a product concept design solution, a fixed prompt for guiding the large language model to perform thought chain reasoning, the thought chain reasoning steps of the sample concept design solution for guiding the large language model to perform thought chain reasoning, and a fixed prompt for guiding the large language model to summarize and obtain the concept design solution.

[0043] The actual problem prompt template includes: a question prompt template for guiding the large language model to generate a product concept design solution, and a fixed prompt for guiding the large language model to perform thought chain reasoning.

[0044] The thought chain reasoning steps of the sample concept design solution described above include four mapping reasoning steps, including: product-requirement mapping, requirement-function mapping, function-behavior mapping, and behavior-structure mapping. A corresponding mapping prompt template is set for each step.

[0045] Specifically, a specific example of the preset thought chain prompt template for this step is as follows: "Question: Generate a concept design solution for ×××.\nAnswer: Let's think step by step:\nFirst step: Clearly define the requirements for this product:....\nSecond step: To meet the requirements of the first step, the following functions need to be implemented:....\nThird step: Provide a more detailed description of the various functions in the second step:....\nFourth step: To implement the various functions described in detail in the second and third steps, this product requires the following structure:....\nTherefore, the answer is:....".

[0046] In this prompt template, "Question: Generate a concept design solution for ×××" is the question prompt template for guiding the large language model to generate a product concept design solution, "Answer: Let's think step by step:" is the fixed prompt for guiding the large language model to perform thought chain reasoning, "First step:...., Fourth step:...." are the thought chain reasoning steps of the sample concept design solution for guiding the large language model to perform thought chain reasoning, and "Therefore, the answer is:" is the fixed prompt for guiding the large language model to summarize and obtain the concept design solution.

[0047] To make the conceptual design solution more concise, only including modules, composite relationships, component attributes, and connector relationships in the SysML model, you can add "The final answer only needs to tell me which components are required and the connection relationships between these components" after the problem prompt template to guide the large language model to generate more expected results. In the step of chain-of-thought reasoning, "The first step:..." is the product-requirement mapping, "The second step:..." is the requirement-function mapping, "The third step:..." is the function-behavior mapping, and "The fourth step:..." is the behavior-structure mapping.

[0048] This prompt template is a sample conceptual design solution prompt template. The content of the specific sample conceptual design solution needs to be designed manually. You can create 2-10 samples according to your needs using this template, which is the few-shot chain-of-thought prompt. Using this few-shot chain-of-thought prompt word can reduce manual operations as much as possible while ensuring the quality of the design solution. When using the large language model to generate the conceptual design solution of a new product, only need to use the prompt word "Question: Generate the conceptual design solution of ×××.\nAnswer: Let's think step by step:" to guide the large language model to generate an answer similar to the sample, including the chain-of-thought reasoning steps and the summary of the conceptual design solution.

[0049] The conceptual design solution described in pure text obtained in this step includes the chain-of-thought reasoning process and the conceptual design solution summarized to include modules, composite relationships, component attributes, and connector relationships in the SysML model. The complete chain-of-thought reasoning process can improve the interpretability of the generated conceptual design solution, facilitate designers to adjust the generated results by adjusting the reasoning process, and bring certain design inspiration to designers. The summarized conceptual design solution contains the basic structural information of the SysML model, that is, the composite relationships between modules in the SysML model and the connection relationships between component attributes. The main part used in the subsequent steps of the present invention in the result of this step is the summarized conceptual design solution part.

[0050] S2. Use the preset prompt word template one to guide the large language model to divide the conceptual design solution described in pure text into sentences and output a list of sentences containing complete semantics.

[0051] In the embodiment, as Figure 3 shown, the preset prompt word template one in this step includes: the definition of the large language model role, the task goal of the sentence division task to be achieved, and the example and specific description of the solution to the sentence division task.

[0052] Specifically, a specific example of the prompt template is as follows: "You are a senior SysML modeler. Now you need to divide the input text into relatively complete sentences and output a list of sentences with complete semantics.\nEach sentence should contain as much relatively complete SysML modeling information as possible to facilitate subsequent SysML modeling based on each sentence.\nFor example: After dividing to get the sentence 'A computer includes a monitor, a keyboard, and a host', subsequent modules such as 'computer','monitor', 'keyboard', and 'host' and their composite relationships can be created according to this sentence; after dividing to get the sentence 'The host is connected to the monitor and the keyboard', subsequent component attributes such as 'host' and'monitor', 'host' and 'keyboard' and their connector relationships can be created according to this sentence.\nThe sentences obtained by division may contain more than one sentence." The specific examples, specific descriptions, and other special requirements in this prompt template can be freely adjusted according to needs.

[0053] Using a large language model for sentence division can make full use of the knowledge reserve and semantic understanding ability of the large language model to achieve an effect similar to manual sentence division. Only the detailed requirements for sentence division need to be completely input to the large language model to guide the generation of the expected results. It has better effect, better generalization, and higher efficiency compared with traditional methods of sentence division.

[0054] S3. Use the preset prompt template two to guide the large language model to extract entity relationships for each sentence in the list of sentences with complete semantics, and output triples containing entities and their relationships.

[0055] In the embodiment, as Figure 4 shown, the preset prompt template two in this step includes: the definition of the large language model role, the task objective of the entity relationship extraction to be achieved, the solution example and specific description of the entity relationship extraction task. The triples of entities and their relationships in the present invention include two types of triples, namely, the module and its composite relationship triples, and the component attribute and its connector relationship triples. The prompt is mainly used to extract the two types of triples contained in the sentence.

[0056] Specifically, a specific example of the prompt template is as follows: "You are a senior SysML modeler. Now you need to extract entity relationships for each sentence in the list of complete semantic sentences and output a list of triples containing entities and their relationships.\nThe entity relationship triples to be extracted include the modules in the SysML model and their composite relationship triples, and the component attributes and their connector relationship triples.\nFor example, for the sentence 'A computer includes a monitor, a keyboard, and a host', you need to extract the triples'module: computer, relationship: composite relationship, module: monitor','module: computer, relationship: composite relationship, module: keyboard', and'module: computer, relationship: composite relationship, module: host', which are three triples of modules and their composite relationships. For the sentence 'The host is connected to the monitor and the keyboard', you need to extract the triples 'component attribute: host, relationship: connector relationship, component attribute: monitor' and 'component attribute: host, relationship: connector relationship, component attribute: keyboard', which are two triples of component attributes and their connector relationships.\nPlease output in the format of the given examples, and indicate the SysML model elements corresponding to the entities and relationships for each entity relationship triple requirement." The specific examples, instructions, and other special requirements in this prompt template can be freely adjusted according to needs.

[0057] Using a large language model for entity relationship extraction can make full use of the semantic understanding and text generation capabilities of the large language model. By using specific examples and detailed instructions, the large language model can be guided to fully understand the requirements of SysML modeling and generate the expected results in the given example format.

[0058] S4. Create a SysML model hierarchical structure according to the triples containing entities and their relationships, following predefined rules.

[0059] In the embodiment, as Figure 5 shown, the predefined rules in this step include: rules for triple connection, rules for triple deduplication, and rules for enhancing the SysML model hierarchical structure. The purpose of the predefined rules is to recombine all the scattered triples into a hierarchical structure containing complete SysML model structure information. The finally obtained SysML hierarchical structure mainly contains all the modules in the SysML model, the hierarchical relationships between the modules, and the connection relationships between the modules, etc., and contains the model structure information of the block definition diagram (BDD) and the internal block diagram (IBD) in the finally required SysML model.

[0060] The rules for triple connection use the product name as the top-level node of the SysML model hierarchy. Triple connections are made based on entity names and added one by one to the SysML model hierarchy. If the triple is a module and its composite relationship triple, the tail entity is used as a child node and added to the next level of the head entity node in the SysML model hierarchy. If the head entity node does not appear in the SysML model hierarchy, a new head entity node is created. If the triple is a component attribute and its connector relationship triple, the tail entity in the triple is added to the next level of the head entity node in the SysML model hierarchy, and a connector label is added to identify the connector relationship. If the head entity node does not appear in the SysML model hierarchy, a new head entity node is created.

[0061] The rules for triple deduplication include: for a module and its composite relationship triple, it is judged as duplicate when the head entity, relationship, and tail entity of the triple all correspond and are the same; for a component attribute and its connector relationship triple, it is judged as duplicate when the relationships of the triples are the same and the name sets of the head entity and the tail entity are the same; when constructing the SysML model hierarchy, duplicate triples are ignored.

[0062] The rules for enhancing the SysML model hierarchy include: after all triples are connected, the top-level nodes of multiple hierarchies are connected to the unique top-level node of the SysML model hierarchy; for all component attribute and its connector relationship triples, a module corresponding to the component attribute is created. If the module does not appear in the SysML model hierarchy, it is added to the SysML model hierarchy and connected to the top-level node, otherwise the module is ignored.

[0063] In the embodiment, specifically, an example of the predefined rules is as follows: First, a top-level node in the SysML model hierarchy is created according to the product name; then the triples are read in the order of the list of entity and its relationship triples obtained in step S3; then each triple is applied with the triple connection rules and triple deduplication rules to create the corresponding nodes of the modules and component attributes in the triple one by one, and added to the SysML model hierarchy according to the relationships in the triple; finally, after all triples are read, the SysML model hierarchy enhancement rules are applied to the initially generated SysML model hierarchy to obtain the final enhanced and complete SysML model hierarchy. The numerous and complex triples are converted into a concise SysML model hierarchy, which is convenient for retrieving and utilizing the required elements when creating an XMI file later.

[0064] S5. Create an XMI file according to the SysML model hierarchy in accordance with the XMI format specification representing the SysML model information to obtain a complete SysML model.

[0065] As shown Figure 6 in the figure, the XMI format specification includes these three parts: SysML model structure information, SysML model graphical information, and other necessary tags. First, the SysML model structure elements need to be mapped to XMI tags, and then the graphical information and other necessary tags are added to the XMI file. The steps to create an XMI file in this step include: the mapping of SysML model elements to XMI tags, adding SysML model graphical information, and an algorithm for creating XMI tags according to the SysML model hierarchical structure.

[0066] This step requires parsing the standard XMI format representing SysML model information. According to the mapping between the SysML model and XMI tags in the order of the XMI format, necessary XMI tags are gradually created, and finally a complete and standardized XMI file is obtained. Importing this XMI file into the modeling software can obtain and display the complete SysML model.

[0067] The mapping of SysML model elements to XMI tags includes: the mapping of modules, composite relationships, part attributes, and connector relationships in the SysML model to the corresponding tags in XMI. Specifically, the mapping relationship between the SysML model elements and XMI tags is shown in Table 1 below: Table 1

[0068] For example, for the first mapping in Table 1, when creating an XMI file, every time a module information is read, a "packagedElement" tag with the label attribute "name=block" will be created at the corresponding position in the XMI structure, and "block" is the module name. This tag also contains other tag attributes such as the globally unique "xmi:id" and the "xmi:type" identifying the element type, which are used to store various detailed information of the module.

[0069] Adding SysML model graphical information includes: the name of the frame, the module it belongs to, the display position, the size of the frame, the display position, explicit size, and display style of the module and part attributes, and the display position, display length, and display style of the connection lines representing composite relationships and connector relationships. The two diagrams, BDD and IBD, in SysML, in addition to containing SysML model structure information, also need to add graphical information for visualization.

[0070] Specifically, after creating the SysML model structure information in the XMI file, it is necessary to create the corresponding BDD and IBD according to the structure information of the SysML model. For example, for the module and its composite relationships, the BDD belonging to its parent module needs to be created, and for the component properties and their connector relationships, the IBD corresponding to the module of the component properties needs to be created. After creating the BDD and IBD, corresponding extended graphic nodes will be created under the XMI root node to store the graphic information of the BDD and IBD. The BDD node and the BDD extended graphic node, as well as the IBD node and the IBD extended graphic node, are connected through the globally unique "xmi:id". The extended graphic node contains all the graphic information required for drawing the BDD and IBD, such as: the frame information, which specifies the drawing area; the specific coordinates of the module in the figure, the size and color of the icon representing the module; the specific arrow symbol representing the composite relationship; the coordinates at both ends of the arrow, the style and color of the arrow line, etc.

[0071] The algorithm for creating XMI tags according to the SysML model hierarchy structure includes: first creating the BDD of the entire model, all the modules in the BDD, and the composite relationships between the modules according to the SysML model hierarchy structure, then creating the component properties under the module according to the composite relationship information of the module, then creating the IBD and the connector relationship according to the component properties and their connector relationships, and finally creating the corresponding extended graphic nodes under the XMI root node according to each existing BDD and IBD to store the SysML model graphic information for displaying the figure.

[0072] An example of the algorithmic process for creating XMI tags based on the SysML model hierarchy is as follows: The XMI to be created is divided into two parts, namely the model part containing the SysML model structure information and the graphical part containing the visualization information. First, create the root node of the entire model part and the package under this node that contains all the structural information of the model. Then, under this package node, create the content of three parts in sequence, namely the BDD that describes the entire model architecture, all the modules in the BDD, and the composite relationships of these modules. After that, under each module, create the component attributes of the modules owned by this module according to the composite relationship information of this module. If a module has component attributes, create an IBD for it and belong it to this module, otherwise this module has no IBD. Finally, create connections according to the connector relationships between the component attributes. After creating the model part, first, retrieve all the BDDs and IBDs in the model part again. Then, add the corresponding extended graphical nodes under the XMI root node in the order in which the BDDs and IBDs appear. Finally, create the corresponding visualization graphical information according to the structural information in the BDDs and IBDs. Finally, an XMI file containing complete SysML model information is obtained. Throughout the creation process, according to the order in which the modules and their composite relationships, component attributes and their connector relationships appear in the described SysML model hierarchy, create the corresponding XMI tags in sequence. The retrieval of the required elements during the creation process also depends on the SysML model hierarchy to complete.

[0073] To clearly demonstrate a method for generating a SysML model based on a large language model in a conceptual design provided by the present invention, corresponding example demonstrations are carried out next.

[0074] ①. Problem: Generate a conceptual design solution for a lunar flight probe. The final answer only needs to tell me which components are required and the connection relationships between these components.

[0075] Answer: Let's think step by step: First step: Clearly define the requirements for this product: This product needs to be able to detect lunar soil composition, water ice distribution, geological structure, and dynamic environment, and adapt to extreme environments such as low lunar gravity, extreme temperature differences, lunar dust, and vacuum. It should support short-distance flight using the rotor mode and long-distance jumping using the thruster mode, and have autonomous navigation and multi-detector collaboration functions.

[0076] Step 2: To meet the requirements of Step 1, the following functions need to be implemented: The core functions include mobility and navigation functions, energy management functions, scientific exploration functions, communication and collaboration functions. Among them, the mobility and navigation functions include short-distance flight function, long-distance jump function, and autonomous obstacle avoidance function. The energy management functions include power generation function, energy storage function, and thermal control function. The scientific exploration functions include sampling function, imaging function, and environmental monitoring function. The communication and collaboration functions include data transmission function and formation control function. Other functions include terrain perception function, power mode switching function, equipment temperature regulation function, data compression and transmission function.

[0077] Step 3: Provide a more detailed description of the various functions in Step 2: The mobility module needs to start rotor flight in the illuminated area (at an altitude of 50 m - 200 m) and switch to a cold gas thruster jump in the shadow area (50 km per jump). The energy module unfolds solar panels for charging during the day and enables RTG power supply during the lunar night, dynamically distributing energy to each subsystem. The exploration module needs to scan the lunar surface in real-time during flight (LiDAR + spectrometer), drill and sample after landing, and analyze the samples. The communication module uploads data to the orbiter through a phased array antenna and synchronizes the status of multiple detectors through a local Mesh network.

[0078] Step 4: To achieve the various functions described in detail in Steps 2 and 3, the product requires the following structure: including a power system, an energy system, a navigation system, a detection system, a control system, and a communication system. Among them, the power system includes a rotor propulsion system, a cold gas thruster, and a fuel storage tank. The rotor propulsion system uses an electric ducted rotor to provide short-range flight lift and connects the power management module (energy input) to the navigation controller (flight instruction receiver). The cold gas thruster uses nitrogen fuel for propulsion to perform long-range jumps and connects the fuel storage tank, the navigation controller (ignition instruction), and the inertial measurement unit (IMU). The energy system includes a foldable solar panel, an RTG nuclear battery, and a storage battery. The foldable solar panel uses a perovskite flexible battery to be responsible for daytime power generation and connects the power management module (charging control) to the storage battery (energy storage). The RTG nuclear battery is responsible for continuous power supply during the lunar night and connects the power management module (priority switching) to the thermal control system (waste heat utilization). The navigation system includes LiDAR, a vision camera, an IMU, and a star sensor. LiDAR + vision camera are responsible for real-time terrain modeling and obstacle avoidance and connect the main control computer (transmission of point cloud data) to the storage unit (map cache). The detection system includes a micro SAR radar, a spectrometer, and a modular payload interface. The micro SAR radar is responsible for detecting the distribution of groundwater ice and connects the main control computer (data preprocessing) to the communication module (transmission of results). The modular payload interface supports extended devices (drillers, seismographs, etc.) and connects the main control computer (control instructions) to the power module (independent power supply). The control system includes a main control computer, a power management module, and a thermal control system. The main control computer is responsible for running autonomous navigation algorithms and mission planning and connects all sensors (LiDAR, camera, IMU), the propulsion system, and the communication module (data flow center). The thermal control system uses phase change materials (PCM) + electric heating films to regulate the temperature and connects the RTG (waste heat collection), the equipment compartment (temperature sensor), and the external heat dissipation plate (radiative heat dissipation). The communication system includes a phased array antenna and a Mesh network relay node. The phased array antenna is responsible for high-speed data transmission and relaying and connects the main control computer (instruction reception) to the deep space network (data uplink).

[0079] Therefore, the answer is: The lunar flight detector includes a power system, an energy system, a navigation system, a detection system, a control system, and a communication system. The power system includes a rotor propulsion system, a cold gas thruster, and a fuel storage tank. The energy system includes a foldable solar panel, an RTG nuclear battery, and an energy storage battery. The navigation system includes a LiDAR, a vision camera, an IMU, and a star sensor. The detection system includes a micro SAR radar, a spectrometer, and a modular payload interface. The control system includes a main control computer, a power management module, and a thermal control system. The communication system includes a phased array antenna and a Mesh network relay node. The rotor propulsion system is connected to the power management module and the navigation controller. The cold gas thruster is connected to the fuel storage tank, the navigation controller (ignition command), and the inertial measurement unit (IMU). The foldable solar panel is connected to the power management module and the energy storage battery. The RTG nuclear battery is connected to the power management module and the thermal control system. The LiDAR and the vision camera are connected to the main control computer and the storage unit. The micro SAR radar is connected to the main control computer and the communication module. The modular payload interface is connected to the main control computer and the power module. The main control computer is connected to the LiDAR, the vision camera, the IMU, the propulsion system, and the communication module. The thermal control system is connected to the RTG, the equipment compartment, and the external heat dissipation plate. The phased array antenna is connected to the main control computer and the deep space network.

[0080] ②. Question: Generate a conceptual design solution for a diesel engine. The final answer only needs to tell me what components are required and the connection relationships between these components.

[0081] Answer: Let's think step by step: Step 1: Clearly define the requirements for this product: Convert diesel chemical energy into mechanical energy and output stable torque and power.

[0082] Step 2: To meet the requirements of Step 1, the following functions need to be achieved: The main functions include efficiently burning diesel to release chemical energy, energy conversion (linear piston motion → crankshaft rotation), and waste heat and exhaust gas management. The sub-functions include precise fuel injection, efficient air compression, mechanical energy transfer, a cooling system, and post-emission treatment.

[0083] Step 3: Describe the various functions in Step 2 in more detail: Through each function, achieve the four-stroke cycle of intake → compression → combustion → exhaust. Energy is transferred through the path of piston → connecting rod → crankshaft → flywheel / output shaft. It is also necessary to adjust the fuel injection volume and boost pressure in real time, atomize the fuel under high pressure, dynamically adjust the intake air volume, and achieve adaptive adjustment of the emission valve according to sensor feedback.

[0084] Step 4: To achieve the various functions described in detail in Step 2 and Step 3, the product requires the following structure: The core components include a combustion system, an intake and exhaust system, a power transmission system, a control system, an after-treatment system for emissions, and a lubrication and cooling system. The combustion system includes a cylinder block, a piston, and an injector. The cylinder block houses the piston movement to form a combustion chamber; the piston is used to convert the combustion pressure into a linear motion; the injector achieves precise atomization of fuel and injects it into the combustion chamber. The intake and exhaust system includes a turbocharger, an intercooler, and an EGR valve. The turbocharger is used to compress the intake air and increase the oxygen density; the intercooler is used to cool the high-temperature compressed air to improve the combustion efficiency; the EGR valve is used to return part of the exhaust gas to the intake end to reduce NOx emissions. The power transmission system includes a crankshaft, a connecting rod, and a flywheel. The crankshaft converts the linear motion of the piston into a rotational motion; the connecting rod connects the piston and the crankshaft to transmit kinetic energy; the flywheel stores rotational inertia and balances the output torque fluctuation. The control system includes an ECU and a sensor network. The ECU receives sensor signals and controls fuel injection, supercharging, and EGR; the sensor network monitors the engine status in real time. The after-treatment system for emissions includes a diesel oxidation catalyst (DOC) and a particulate filter (DPF). The diesel oxidation catalyst (DOC) is used to oxidize CO; the particulate filter (DPF) is used to filter PM particles in the exhaust gas. The lubrication and cooling system includes a variable-displacement oil pump and a split water pump. The variable-displacement oil pump supplies lubricating oil to friction components such as the crankshaft and piston as needed; the split water pump is used to independently control the coolant flow rate of the cylinder block and cylinder head. Among them, the piston, connecting rod, crankshaft, and flywheel are mechanically connected to form a linear-rotary energy transfer chain; the turbocharger, intercooler, intake pipe, and cylinder are mechanically connected to achieve the input of compressed air into the combustion chamber; the exhaust pipe, turbocharger, EGR valve, DOC, and DPF are mechanically connected to form an exhaust gas treatment and energy recovery path. The fuel tank, high-pressure oil pump, common rail pipe, and injector are fluid-connected to form a fuel supply path; the oil pump, main oil gallery, crankshaft bearing / piston pin are fluid-connected to form a lubricating oil circulation path; the water pump, cylinder block water jacket, cylinder head water jacket, and radiator are fluid-connected to form a coolant circulation path. The sensor network and the ECU are connected by control signals to achieve data input; the ECU is respectively connected to the injector, EGR valve, turbocharger vane angle, and oil pump displacement by control signals for performing control output.

[0085] Therefore, the answer is: A diesel engine includes a combustion system, an intake and exhaust system, a power transmission system, a control system, an after-treatment system for emissions, and a lubrication and cooling system. The combustion system includes a cylinder block, a piston, and an injector. The intake and exhaust system includes a turbocharger, an intercooler, and an EGR valve. The power transmission system includes a crankshaft, a connecting rod, and a flywheel. The control system includes an ECU and a sensor network. The after-treatment system for emissions includes a diesel oxidation catalyst (DOC) and a diesel particulate filter (DPF). The lubrication and cooling system includes a variable displacement oil pump and a split water pump. The piston, connecting rod, crankshaft, and flywheel are mechanically connected. The turbocharger, intercooler, intake pipe, and cylinder are mechanically connected. The exhaust pipe, turbocharger, EGR valve, DOC, and DPF are mechanically connected. The fuel tank, high-pressure fuel pump, common rail pipe, and injector are fluidly connected. The oil pump, main oil gallery, crankshaft bearings / piston pins are fluidly connected. The water pump, cylinder block water jacket, cylinder head water jacket, and radiator are fluidly connected. The sensor network and the ECU are connected by control signals. The ECU is respectively connected to the injector, EGR valve, turbocharger vane angle, and oil pump displacement by control signals.

[0086] ③. Question: Generate a conceptual design solution for a sonar. The final answer only needs to tell me what components are required and the connection relationships between these components.

[0087] Answer: Let's think step by step: Step 1: Define the requirements for this product: The sonar system needs to be able to detect information such as the position, distance, and speed of underwater targets through sound waves, adapt to complex underwater environments, and have high precision and anti-interference capabilities.

[0088] Step 2: To meet the requirements of Step 1, the following functions need to be implemented: The core functions include a sound wave emission and reception function, a signal processing function, a target recognition and positioning function, and a data output function. The sound wave emission and reception function includes sound wave emission, sound wave reception, and signal amplification. The signal processing function includes noise filtering, signal enhancement, and target feature extraction. The target recognition and positioning function includes target distance calculation, target speed calculation, and target azimuth calculation. The data output function includes data visualization and data transmission.

[0089] Step 3: Describe each function in Step 2 in more detail: The sound wave emission module needs to generate sound wave signals of specific frequencies and emit them into the water through a transducer; the sound wave reception module receives the reflected sound wave signals through a hydrophone and amplifies them. The signal processing module needs to filter, enhance, and extract target features from the received signals. The target recognition and positioning module calculates the distance, speed, and azimuth of the target through the processed signals. The data output module displays the processed data through a visualization interface or transmits it to other systems.

[0090] Step 4: To implement the various functions described in detail in Step 2 and Step 3, the product requires the following structure: including an acoustic wave emission system, an acoustic wave reception system, a signal processing system, a target recognition and positioning system, and a data output system. The acoustic wave emission system includes a signal generator, a power amplifier, and a transducer. The signal generator generates acoustic wave signals of specific frequencies, connecting the power amplifier (for signal amplification) and the transducer (for acoustic wave emission). The acoustic wave reception system includes a hydrophone and a preamplifier. The hydrophone receives the reflected acoustic wave signals, connecting the preamplifier (for signal amplification) and the signal processing system (for signal input). The signal processing system includes a filter and a signal processor. The filter is used for noise filtering, connecting the preamplifier (for signal input) and the signal processor (for signal enhancement and feature extraction). The target recognition and positioning system includes a distance calculation module, a speed calculation module, and a bearing calculation module. The distance calculation module connects the signal processor (for signal input) and the data output system (for distance data output). The speed calculation module connects the signal processor (for signal input) and the data output system (for speed data output). The bearing calculation module connects the signal processor (for signal input) and the data output system (for bearing data output). The data output system includes a data visualization module and a data transmission module. The data visualization module connects the target recognition and positioning system (for data input) and the display device (for data output). The data transmission module connects the target recognition and positioning system (for data input) and the external system (for data transmission).

[0091] Therefore, the answer is: The sonar system includes an acoustic wave emission system, an acoustic wave reception system, a signal processing system, a target recognition and positioning system, and a data output system. The acoustic wave emission system includes a signal generator, a power amplifier, and a transducer. The acoustic wave reception system includes a hydrophone and a preamplifier. The signal processing system includes a filter and a signal processor. The target recognition and positioning system includes a distance calculation module, a speed calculation module, and a bearing calculation module. The data output system includes a data visualization module and a data transmission module. The signal generator connects the power amplifier and the transducer. The hydrophone connects the preamplifier and the signal processing system. The filter connects the preamplifier and the signal processor. The distance calculation module, the speed calculation module, and the bearing calculation module respectively connect the signal processor and the data output system. The data visualization module and the data transmission module respectively connect the target recognition and positioning system and the display device or the external system. The answer here is for subsequent steps.

[0092] Next, you are a senior SysML modeler. Now you need to divide the input text into relatively complete sentences and output a list of sentences with complete semantics.

[0093] The input text is as follows: The sonar system includes a sound wave emission system, a sound wave reception system, a signal processing system, a target recognition and positioning system, and a data output system. The sound wave emission system includes a signal generator, a power amplifier, and a transducer. The sound wave reception system includes a hydrophone and a preamplifier. The signal processing system includes a filter and a signal processor. The target recognition and positioning system includes a distance calculation module, a speed calculation module, and a bearing calculation module. The data output system includes a data visualization module and a data transmission module. The signal generator is connected to the power amplifier and the transducer. The hydrophone is connected to the preamplifier and the signal processing system. The filter is connected to the preamplifier and the signal processor. The distance calculation module, the speed calculation module, and the bearing calculation module are respectively connected to the signal processor and the data output system. The data visualization module and the data transmission module are respectively connected to the target recognition and positioning system and the display device or external system.

[0094] The output is that the sonar system includes a sound wave emission system, a sound wave reception system, a signal processing system, a target recognition and positioning system, and a data output system.

[0095] Among them, the sound wave emission system includes a signal generator, a power amplifier, and a transducer.

[0096] The sound wave reception system includes a hydrophone and a preamplifier.

[0097] The signal processing system includes a filter and a signal processor.

[0098] The target recognition and positioning system includes a distance calculation module, a speed calculation module, and a bearing calculation module.

[0099] The data output system includes a data visualization module and a data transmission module.

[0100] The signal generator is connected to the power amplifier and the transducer.

[0101] The hydrophone is connected to the preamplifier and the signal processing system.

[0102] The filter is connected to the preamplifier and the signal processor.

[0103] The distance calculation module, the speed calculation module, and the bearing calculation module are respectively connected to the signal processor and the data output system.

[0104] The data visualization module and the data transmission module are respectively connected to the target recognition and positioning system and the display device or external system.

[0105] Next, you are a senior SysML modeler. Now you need to perform entity relationship extraction on each sentence in the complete semantic sentence list and output a list of triples containing entities and their relationships.

[0106] List of input sentences: The sonar system includes a sound wave emission system, a sound wave reception system, a signal processing system, a target recognition and positioning system, and a data output system.

[0107] Among them, the sound wave emission system includes a signal generator, a power amplifier, and a transducer.

[0108] The sound wave reception system includes a hydrophone and a preamplifier.

[0109] The signal processing system includes a filter and a signal processor.

[0110] The target recognition and positioning system includes a distance calculation module, a speed calculation module, and a bearing calculation module.

[0111] The data output system includes a data visualization module and a data transmission module.

[0112] The signal generator is connected to the power amplifier and the transducer.

[0113] The hydrophone is connected to the preamplifier and the signal processing system.

[0114] The filter is connected to the preamplifier and the signal processor.

[0115] The distance calculation module, the speed calculation module, and the bearing calculation module are respectively connected to the signal processor and the data output system.

[0116] The data visualization module and the data transmission module are respectively connected to the target recognition and positioning system and the display device or external system.

[0117] The list of triples of the output entities and their relationships is specifically: Module: Sonar, Relationship: Composite Relationship, Module: Sound Wave Emission System Module: Sonar, Relationship: Composite Relationship, Module: Sound Wave Reception System Module: Sonar, Relationship: Composite Relationship, Module: Signal Processing System Module: Sonar, Relationship: Composite Relationship, Module: Target Recognition Module: Sonar, Relationship: Composite Relationship, Module: Positioning System Module: Sonar, Relationship: Composite Relationship, Module: Data Output System Module: Sound Wave Emission System, Relationship: Composite Relationship, Module: Signal Generator Module: Sound Wave Emission System, Relationship: Composite Relationship, Module: Power Amplifier Module: Sound Wave Emission System, Relationship: Composite Relationship, Module: Transducer Module: Sound Wave Reception System, Relationship: Composite Relationship, Module: Hydrophone Module: Acoustic Wave Receiving System, Relationship: Composite Relationship, Module: Preamplifier Module: Signal Processing System, Relationship: Composite Relationship, Module: Filter Module: Signal Processing System, Relationship: Composite Relationship, Module: Signal Processor Module: Positioning System, Relationship: Composite Relationship, Module: Distance Calculation Module Module: Positioning System, Relationship: Composite Relationship, Module: Speed Calculation Module Module: Positioning System, Relationship: Composite Relationship, Module: Azimuth Calculation Module Module: Data Output System, Relationship: Composite Relationship, Module: Data Visualization Module Module: Data Output System, Relationship: Composite Relationship, Module: Data Transmission Module Component Attribute: Signal Generator, Relationship: Connector Relationship, Component Attribute: Power Amplifier Component Attribute: Signal Generator, Relationship: Connector Relationship, Component Attribute: Transducer Component Attribute: Hydrophone, Relationship: Connector Relationship, Component Attribute: Preamplifier Component Attribute: Hydrophone, Relationship: Connector Relationship, Component Attribute: Signal Processing System Component Attribute: Filter, Relationship: Connector Relationship, Component Attribute: Preamplifier Component Attribute: Filter, Relationship: Connector Relationship, Component Attribute: Signal Processor Component Attribute: Distance Calculation Module, Relationship: Connector Relationship, Component Attribute: Signal Processor Component Attribute: Speed Calculation Module, Relationship: Connector Relationship, Component Attribute: Signal Processor Component Attribute: Azimuth Calculation Module, Relationship: Connector Relationship, Component Attribute: Signal Processor Component Attribute: Distance Calculation Module, Relationship: Connector Relationship, Component Attribute: Data Output System Component Attribute: Speed Calculation Module, Relationship: Connector Relationship, Component Attribute: Data Output System Component Attribute: Azimuth Calculation Module, Relationship: Connector Relationship, Component Attribute: Data Output System Component Attribute: Data Visualization Module, Relationship: Connector Relationship, Component Attribute: Target Recognition Component Attribute: Data Visualization Module, Relationship: Connector Relationship, Component Attribute: Positioning System Component Attribute: Data Visualization Module, Relationship: Connector Relationship, Component Attribute: Display Device Component Attribute: Data Visualization Module, Relationship: Connector Relationship, Component Attribute: External System Component Attribute: Data Transmission Module, Relationship: Connector Relationship, Component Attribute: Target Recognition Component Attribute: Data Transmission Module, Relationship: Connector Relationship, Component Attribute: Positioning System Component Attribute: Data Transmission Module, Relationship: Connector Relationship, Component Attribute: Display Device Component Attribute: Data Transmission Module, Relationship: Connector Relationship, Component Attribute: External System Next, using the above triples containing entities and their relationships as input, create a SysML model hierarchical structure, and the specific SysML model hierarchical structure is as follows: - Sonar - Acoustic Wave Emission System - Signal Generator - to Power Amplifier - to Transducer - Power Amplifier - Transducer - Acoustic Wave Reception System - Hydrophone - to Preamplifier - Preamplifier - Signal Processing System - Filter - to Signal Processor - Signal Processor - Positioning System - Distance Calculation Module - Speed Calculation Module - Azimuth Calculation Module - Data Output System - Data Visualization Module - Data Transmission Module - Display Device - Target Recognition - External System Finally, using the above SysML model hierarchical structure as input, generate an XMI file, import the XMI file into the modeling software, and obtain a visual SysML model, such as Figure 7 The BDD schematic diagram generated for the sonar instance, Figure 8 The IBD schematic diagram of the SysML model generated for the sonar instance, where Figure 7 The small icons under "Sonar", "Acoustic Wave Emission System", "Acoustic Wave Reception System", and "Signal Processing System" respectively correspond to Figure 8 4 IBD diagrams, which play a role in navigation. Figure 8 In the internal module diagram of "Sonar", the small icons under "Acoustic Wave Emission System", "Acoustic Wave Reception System", and "Signal Processing System" respectively correspond to the other 3 IBD diagrams, which also play a role in quick navigation.

[0118] Based on the example results, in the concept design provided by the present invention, the method for generating a SysML model based on a large language model uses few-shot chain-of-thought prompts to minimize manual operations while ensuring the quality of the design solution. In addition, the rich knowledge base of the large language model is utilized to generate a concept design solution, and based on this, a SysML model is automatically modeled. The entire process does not require manual construction of additional professional domain datasets, and the large language model can be used to generate concept design solutions for product names in different fields; there are no restrictions on the input, and even if there is only one product name, a SysML model corresponding to the concept design solution of the product name can be automatically generated; the SysML model is represented using a common XMI format specification file, which can be imported into common practical modeling software that supports the XMI format, facilitating subsequent further modeling and modification by designers, being more practical and having further improved automation.

[0119] The above-described specific embodiments have elaborated in detail the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, supplements, equivalent replacements, etc. made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for generating a SysML model based on a large language model in conceptual design, characterized in that It includes the following steps: Input the user requirements related to conceptual design into the large language model, and use the preset chain-of-thought prompt template to guide the large language model to output a conceptual design solution described in plain text; Use the preset prompt template one to guide the large language model to divide the conceptual design solution described in plain text into sentences, and output a list of sentences containing complete semantics; Use the preset prompt template two to guide the large language model to extract entity relationships for each sentence in the list of sentences containing complete semantics, and output triples containing entities and their relationships; Create a SysML model hierarchical structure according to the triples containing entities and their relationships according to predefined rules; Create an XMI file according to the SysML model hierarchical structure according to the XMI format specification representing SysML model information, and obtain a complete SysML model.

2. The method for generating a SysML model based on a large language model in the conceptual design according to claim 1, characterized in that, The user requirements related to conceptual design are a product name noun in a highly complex large system, or a product name noun and a simple requirement description of the product.

3. The method for generating a SysML model based on a large language model in the conceptual design according to claim 1, wherein The preset chain-of-thought prompt template is a few-shot chain-of-thought prompt template, which includes two parts: a chain-of-thought sample prompt template and an actual problem prompt template; The chain-of-thought sample prompt template contains 2-10 chain-of-thought samples, and the prompt template for each chain-of-thought sample includes: a problem prompt template for guiding the large language model to generate a product conceptual design solution, a fixed prompt for guiding the large language model to perform chain-of-thought reasoning, a chain-of-thought reasoning step of the sample conceptual design solution for guiding the large language model to perform chain-of-thought reasoning, and a fixed prompt for guiding the large language model to summarize and obtain the conceptual design solution; The actual problem prompt template includes: a problem prompt template for guiding the large language model to generate a product conceptual design solution and a fixed prompt for guiding the large language model to perform chain-of-thought reasoning.

4. The method for generating a SysML model based on a large language model in the conceptual design according to claim 3, characterized in that, The chain-of-thought reasoning steps of the sample conceptual design solution for guiding the large language model to perform chain-of-thought reasoning include four mapping reasoning steps, specifically including: product-requirement mapping, requirement-function mapping, function-behavior mapping, and behavior-structure mapping. Corresponding mapping prompt templates are set for each mapping reasoning step.

5. The method for generating a SysML model based on a large language model in the conceptual design according to claim 1, wherein The conceptual design solution described in plain text includes the chain-of-thought reasoning process and the structure information of the SysML model, where the structure information of the SysML model includes modules and their composite relationships, and component attributes and their connector relationships.

6. The method for generating a SysML model based on a large language model in the conceptual design according to claim 1, characterized in that, The preset prompt template one includes: large language model role definition, the task goal of sentence division to be achieved, a sentence division task solution example, and specific descriptions.

7. The method for generating a SysML model based on a large language model in the conceptual design according to claim 1, characterized in that, The preset prompt template two includes: large language model role definition, the task goal of entity relationship extraction to be achieved, an entity relationship extraction task solution example, and specific descriptions.

8. The method for generating a SysML model based on a large language model in the conceptual design according to claim 5, wherein The triples containing entities and their relationships include triples of modules and their composite relationships and triples of component attributes and their connector relationships.

9. The method for generating a SysML model based on a large language model in the conceptual design according to claim 8, characterized in that, The predefined rules include: rules for triple connection, rules for triple deduplication, and rules for enhancing the SysML model hierarchical structure; The rules for triple connection use the product name as the top-level node of the SysML model hierarchy. Connect triples according to the entity names and add them to the SysML model hierarchy one by one. If the triple is a module and its composite relationship triple, use the tail entity as the child node and add it to the next level of the head entity node in the SysML model hierarchy. If the head entity node does not appear in the SysML model hierarchy, create the head entity node. If the triple is a component attribute and its connector relationship triple, add the tail entity in the triple to the next level of the head entity node in the SysML model hierarchy and add a connector label to identify the connector relationship. If the head entity node does not appear in the SysML model hierarchy, create the head entity node. The rules for triple deduplication include: for a module and its composite relationship triple, it is judged as duplicate when the head entity, relationship, and tail entity of the triple are the same; for a component attribute and its connector relationship triple, it is judged as duplicate when the relationships are the same and the name sets of the head entity and tail entity are the same. The rules for enhancing the SysML model hierarchy include: after all triples are connected, connect the top-level nodes of multiple hierarchies to the unique top-level node of the SysML model hierarchy; for all component attribute and its connector relationship triples, create a module corresponding to the component attribute. If the module does not appear in the SysML model hierarchy, add it to the SysML model hierarchy and connect it to the top-level node, otherwise ignore the module.

10. The method for generating a SysML model based on a large language model in the conceptual design according to claim 9, wherein, Creating an XMI file includes: the mapping of SysML model elements to XMI tags, adding SysML model graphic information, and an algorithm for creating XMI tags based on the SysML model hierarchy. The mapping of SysML model elements to XMI tags includes: the mapping of modules, composite relationships, component attributes, and connector relationships in the SysML model to the corresponding tags in XMI. Adding SysML model graphic information includes: the name of the drawing frame, the module it belongs to, the display position, the size of the drawing frame, the display positions, explicit sizes, and display styles of modules and component attributes, and the display positions, display lengths, and display styles of the connecting lines of composite relationships and connector relationships. The algorithm for creating XMI tags based on the SysML model hierarchy includes: creating a module definition diagram, all modules in the module definition diagram, and the composite relationships between modules according to the SysML model hierarchy, then creating the component attributes under the module according to the composite relationship information of the module, then creating an internal module diagram and connector relationships according to the component attributes and their connector relationships, and finally creating the corresponding extended graphic nodes under the XMI root node according to each existing module definition diagram and internal module diagram.

Images