A domain knowledge graph construction method based on a FBS concept design model

Abstract

The application discloses a kind of based on FBS concept design model's field knowledge graph construction method, it is related to knowledge graph construction technical field.The present application includes: from document backtracking analysis and product concept design relevant design structure, design behavior, design function, and design structure relevant feature attribute, by predefining the relationship type between these design elements, construct design knowledge graph, to solve the problem of multiple entities corresponding to single entity in document, the structure of traditional Casre l model is improved, and attention module is added.According to the mapping relationship between function and structure, the design structure set that can realize the function is retrieved;For two functions that exist interaction, retrieve the structure that exists connection path in the two design structure sets corresponding to two functions, take the union of design structure.The method of the present application can construct the knowledge graph that provides stable and systematic design knowledge, and the improved Casre l model can effectively handle the problem of single entity relationship overlap.

Description

Technical Field

[0001] This invention belongs to the field of knowledge graph construction technology, and in particular relates to a method for constructing a domain knowledge graph based on the FBS conceptual design model. Background Technology

[0002] The product design process is generally divided into three stages: conceptual design, detailed design, and pilot production. Conceptual design relies heavily on the designer's experience and abstract, uncertain concepts such as consumer market demands, thus offering high flexibility. However, the subsequent detailed design and pilot production stages involve the product's specific structure. Changes in one design stage can lead to corresponding changes in other stages, making the product's conceptual design significantly impact the entire production process.

[0003] Surveys show that conceptual design accounts for 70%-80% of the total product manufacturing cost. Product conceptual development and design innovation mainly fall into two categories: one is based on the product's fundamental principles, involving innovative design to create products with different working principles and greater efficiency; the other refers to upgrading and optimizing existing market products to improve their adaptability to demand. Currently, the prevailing practice in manufacturing is to optimize and improve existing products within the market, matching consumer demand under conditions of low cost and mass production.

[0004] Currently, product concept development and design suffers from problems such as low design efficiency, a lack of design solutions, and unsatisfactory design results. Designers lack a complete and systematic design knowledge base during the product development phase, causing them to spend considerable time and effort integrating the relevant design knowledge required for product design. For example, Chinese patent CN118171726A provides a method, system, storage medium, and computer for constructing a knowledge graph for the entire project process. The construction method includes: dividing the entire project process into business stages and designing a knowledge system; constructing a knowledge graph schema layer based on the entity attributes and relationships of the knowledge system; acquiring business-related data from each business stage and preprocessing the data; defining entity types and relationship types for the relevant data based on the knowledge graph schema layer and acquiring knowledge to construct the corresponding knowledge graph; acquiring the entity features of the knowledge graph; performing entity alignment operations on all knowledge graphs based on the entity features to achieve knowledge graph fusion; and completing the fused knowledge graph to construct a complete knowledge graph for the entire project process. Other examples include Chinese patents CN116204660A, CN117668248A, CN115098691A, and so on, all of which provide a method for establishing a knowledge graph.

[0005] However, the current predicament faced by manufacturing enterprises is the rapid changes in the consumer market and the personalization of consumer demands. This poses a greater challenge to enterprises' innovative design capabilities in response to the complex consumer demands of the current market. Models that primarily rely on existing knowledge graphs are clearly struggling to keep pace with market developments. Therefore, this paper proposes a design knowledge graph construction method based on the FBS conceptual design model. Summary of the Invention

[0006] The purpose of this invention is to provide a method for constructing a domain knowledge graph based on the Conceptual Design Model (FBS). This method involves crawling target information, preprocessing data, filtering and classifying data to form a database about the FBS, including a structure database, a behavior database, and a function database. These three databases are sequentially mapped through structure-behavior and behavior-function mappings to form function-structure mapping nodes, thereby constructing a knowledge graph of the design model. This addresses the lack of objectivity in design solutions and the low efficiency of design knowledge reuse in existing traditional conceptual design processes. By collecting and organizing design knowledge, a complete design knowledge base is constructed, which can effectively improve design efficiency, reduce R&D costs, and shorten the product design and development cycle.

[0007] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0008] This invention presents a method for constructing a domain knowledge graph based on the Freeform Baseline (FBS) conceptual design model. This method systematically describes design knowledge about products. Entities and entity relationships are the fundamental components of the knowledge graph. The ontology mentioned in this invention refers to the category representation and generalization of entities; an entity is a concrete representation of the ontology concept; and entity relationships describe the connections between entities. In subsequent tasks of this invention, structure, behavior, and function are listed as the ontology of related knowledge entities. Knowledge entities are expressed using natural language, and different entity relationship concepts are defined to describe design knowledge. The method for constructing a domain knowledge graph based on the FBS conceptual design model includes the following steps:

[0009] Step 1: Construct a design knowledge carrier acquisition module. Use web crawlers to obtain relevant design documents, preprocess the document data, and establish a top-level design knowledge carrier database. This involves sorting and analyzing design documents related to the product and extracting the text to store in the product design text library.

[0010] Step Two: Utilize a neural network model to extract structural entities of products from relevant texts, classify product structures, extract relevant design structure knowledge, and store it in a product design structure database. This involves constructing a design knowledge representation module and initially establishing a design knowledge graph. The specific process is as follows:

[0011] Step S021: Based on the FBS design model, perform a top-down hierarchical analysis of the design product's design functions, design behaviors, and design structure. The behavior serves as a constraint on the relationship between the design function representation and the structure.

[0012] Step S022: The SRO-based expression structure takes the form of triples, namely (subject, relation, object), as the knowledge representation framework, where both the subject and object are knowledge ontologies;

[0013] Step S023: Based on the knowledge ontology representation framework, define the ontology type and relation type of knowledge in the design knowledge graph. The ontology type is the type based on FBS, and the relation type is the inclusion relationship and connection relationship between design structures, and the input and output relationship of behavior flow with respect to the design structure.

[0014] Step 3: Construct a knowledge extraction module to extract design knowledge entities from relevant product documents and map them to corresponding ontologies. Classify the entities according to their knowledge representation attributes, and then save the classified entities as triples to a graph database. The specific process is as follows:

[0015] Step S331: Perform named entity recognition on the patent text. The extracted design knowledge entities include structural entities in the text, as well as the corresponding attribute information and behavioral information of the structural entities.

[0016] Step S332: Perform relation judgment on all identified structural entities. Based on the relation type defined in Step 2, determine all structural entities corresponding to the subject and object under a specified relation type, and extract the corresponding triples.

[0017] Step S333: Based on the similarity of character vectors, fuse the same structural entities, vectorize the text content and attribute information of the structural entities, and then calculate the cosine similarity between the two. If the similarity value is higher than a certain threshold, they are regarded as the same structural entities.

[0018] Step S334: Based on the triples, the nodes of the knowledge graph are used as structural entities, and the edges are used as the relation types of the structural entities, and are saved to the graph database;

[0019] Step 4: Construct a knowledge generation module to generate functional descriptions of the components included in the product from the relevant product descriptions, thus obtaining functional entities. The specific process is as follows:

[0020] Step S41: Using the encoder-decoder model, receive the document text and output the functional descriptions of each component to obtain the function-structure relationship triplet;

[0021] Step S42: Merge the functional entities and structural entities separately;

[0022] Step S43: Retrieve the corresponding structural entities in the graph database and save the function-structure relationship triples to the graph database;

[0023] Step 5: Construct a knowledge application module to realize knowledge retrieval and reuse during the product concept design process. The specific process is as follows:

[0024] Step S51: Retrieve functional entities and find the required functions;

[0025] Step S52: Based on the mapping relationship between function and structure, retrieve the set of design structures that can realize the function;

[0026] Step S53: For two interactive functions, retrieve the structures with connection paths in the corresponding two design structure sets, and take the union of the design structures.

[0027] Furthermore, the ontology types defined in step two include design function ontology, design behavior ontology, and design structure ontology. The relation type definitions include inclusion relations and constraint relations between design structures, as well as input and output relations between design structures and behavior flows. Among them, behavior flow is a general term for material flow, energy flow, and signal flow. That is, if there is functional interaction between two design structures, then there will be input and output relations of behavior flows between them.

[0028] Furthermore, in step three, the named entity recognition part is based on a BERT model pre-trained word vector encoder, uses Bi LSTM for encoding long texts, and uses a head and tail pointer network for decoding.

[0029] Furthermore, the relation judgment part in step three is based on the Cairel model. The subject in the triple is extracted in advance, and then a multi-layer head and tail pointer network is constructed. Each layer corresponds to a predefined relation type, and the object corresponding to the subject is decoded under each head and tail pointer network.

[0030] Furthermore, the graph database in step three is Neo4j.

[0031] Furthermore, the encoder-decoder model in step four is a Transformer.

[0032] The present invention has the following beneficial effects:

[0033] This invention utilizes a concept design model based on FBS to retrospectively analyze the design structure, design behavior, design function, and design structure-related characteristic attributes from design documents. By pre-defining the relationship types between these design elements, a design knowledge graph is constructed to provide stable and systematic design knowledge support for the product development process, thereby improving product development efficiency and reducing product development costs.

[0034] This invention improves the CARE L model by constructing a multi-layer head and tail pointer network and adding a self-attention mechanism to fuse head entity information and sequence element information. This can solve the problem of a single entity corresponding to multiple entities in a document and improve the quality of knowledge graph generation.

[0035] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0036] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart of the design knowledge graph construction method based on the FBS conceptual design model of the present invention;

[0038] Figure 2 This is the type graph of the behavior chain in the design knowledge graph construction method based on the FBS conceptual design model of this invention;

[0039] Figure 3 This is a structural diagram of the entity recognition model of the knowledge extraction module in the knowledge graph construction method based on the FBS conceptual design model of this invention;

[0040] Figure 4 This is a structural diagram of the entity relationship extraction model in the knowledge extraction module of the knowledge graph construction method based on the FBS conceptual design model of this invention.

[0041] Figure 5 This is a model structure diagram generated by the design function description of the knowledge generation module in the design knowledge graph construction method based on the FBS conceptual design model of this invention;

[0042] Figure 6 This is an example diagram illustrating the division of functional modules in the knowledge application module of the knowledge graph construction method based on the FBS conceptual design model of this invention.

[0043] Figure 7 This is an application flowchart of the knowledge application module in the knowledge graph construction method based on the FBS conceptual design model of this invention. Detailed Implementation

[0044] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0045] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0046] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0047] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0048] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0049] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0050] Example 1:

[0051] Please see Figure 1As shown, this invention is a method for constructing a domain knowledge graph based on the FBS conceptual design model, comprising the following steps:

[0052] S1: Organize and analyze relevant product design documents, extract the text and store it in the product design text library;

[0053] S2: Use a neural network model to extract the structural entities of products from relevant texts, classify product structures, extract relevant design structure knowledge, and store it in a product design structure database.

[0054] S3: Based on the product design structure database established in step S2, establish the behavior chain between product structures and store the product behavior flow into the product design behavior database.

[0055] S4: Based on the product design structure database and product design behavior database established in steps S2 and S3, use structure-behavior and behavior-function mapping to generate design functions according to the behavior chain between product structures and store them in the product design function database.

[0056] S5: Retrieve the design function database and extract the structural solution set of the corresponding design function using relevant mappings;

[0057] S6: For design functions with coupling, solve the relevant structures using the behavior flow constraints in the behavior database. If there is a behavior chain between the corresponding structures, extract the corresponding design structure.

[0058] Example 2:

[0059] As an embodiment of the present invention, preferably addressing the problems of insufficient objectivity of design schemes, low efficiency of design knowledge reuse, and inability to uncover implicit design knowledge in the field of product concept design, the present invention provides a design knowledge graph construction method based on the FBS concept design model, specifically implemented for the patent documents of AGV vehicles, including the following steps:

[0060] Step 1: Construct a design knowledge carrier acquisition module, which obtains relevant design documents through web crawlers, preprocesses the design document data, and establishes a top-level design knowledge carrier database; as an embodiment of the present invention, preferably, Step 1 specifically includes:

[0061] Using the Scrapy web crawler framework to crawl patent documents from major patent websites.

[0062] Batch extract text content from PDF patent documents and convert the data to be processed into text type;

[0063] Remove stop words from text based on the stop word list from Harbin Institute of Technology to eliminate the influence of noise;

[0064] Step Two: Construct a design knowledge representation module, and establish a design knowledge graph based on the FBS design model and the SRO representation structure; as an embodiment provided by the present invention, preferably, Step Two specifically includes:

[0065] Regarding knowledge definition, design functions, design behaviors, and design structures are three types of ontology;

[0066] Define the inclusion and connection relationships between design structures;

[0067] Defining the behavioral flow of the design structure based on the black-box model;

[0068] Construct behavioral chains between design structures based on the transmission of behavioral flow, such as Figure 2 As shown, the behavior chain is divided into sequential behavior chain, branching behavior chain, sinking behavior chain, and selection behavior chain;

[0069] Store the above definitions in a graph database;

[0070] Step 3: Extract design knowledge entities from the design document and map them to the corresponding ontology. Perform attribute classification for the knowledge representation of the entities, and save the classified entities as triples in the graph database. As an embodiment of this invention, preferably, Step 3 specifically includes:

[0071] The product-related documents are pre-trained using BERT to obtain word vector codes.

[0072] The product-related documents are encoded using Bi LSTM for long text feature vectors. The model structure is as follows: Figure 3 As shown;

[0073] The obtained word vector encoding and feature vector encoding are decoded based on head and tail pointers to extract relation triples from the text. The model structure is as follows: Figure 4 As shown, the obtained structural entity triples are saved to the graph database;

[0074] Step 4: Construct a knowledge generation module to generate functional descriptions of the components included in the product from the relevant product descriptions. As an embodiment of the present invention, preferably, Step 4 specifically includes:

[0075] The knowledge generation module uses the Transformer model, and its structure is as follows: Figure 5 As shown;

[0076] Step 5: Construct a knowledge application module to enable knowledge retrieval and reuse during the product concept design process;

[0077] Step five specifically includes:

[0078] Divide the product into functional modules, such as Figure 6As shown, the functional modules of the AGV are divided based on the behavior flow;

[0079] according to Figure 7 The flowchart shown retrieves the required design structure from feasible design solutions.

[0080] Step 5: Build a knowledge application module to enable knowledge retrieval and reuse during the product concept design process.

[0081] As an embodiment of the present invention, preferably, the specific process of constructing the design knowledge expression module and establishing the design knowledge graph in step two is as follows:

[0082] Based on the FBS design model, a top-down hierarchical analysis of the design product's design functions, design behaviors, and design structure is performed.

[0083] The SRO-based representation structure uses triples as the knowledge representation framework, where both the subject and the object are knowledge ontologies. The triples are: subject, relation, and object.

[0084] The knowledge ontology-based representation framework defines ontology types and relation types in the design of knowledge graphs, i.e., FBS-based types.

[0085] As an embodiment of the present invention, preferably, in step two, the defined ontology types include: design function ontology, design behavior ontology, and design structure ontology, and the relationship type definitions include: inclusion relationship and constraint relationship between design structures, and input and output relationship between design structure and behavior flow.

[0086] As an embodiment of the present invention, preferably, the specific steps of saving the classified entities in the form of triples in step three are as follows:

[0087] Step S31: Perform named entity recognition on the design document. Use the BERT model, Bi LSTM model, and head and tail pointer network to pass through the feature encoding layer, sequence encoding layer, and label decoding layer respectively to identify the component names, corresponding attribute information, and behavioral information contained in the design document.

[0088] Step S32: Perform relation judgment on all identified entities. Using the CAREL model based on predefined relation types, determine all entities corresponding to the subject and object under a specified relation type, thereby extracting the corresponding triples.

[0089] Step S33: Based on the similarity of character vectors, fuse the same entities, vectorize the text content and attribute information of the entities, and then calculate the cosine similarity between the two. If the similarity value is higher than the threshold, they are regarded as the same entities.

[0090] Step S34: Based on the triples, represent knowledge entities with nodes of the knowledge graph and relationship types with edges, and save them to the graph database.

[0091] As an embodiment of the present invention, preferably, in step three, the named entity recognition method is as follows: a word vector encoder pre-trained based on the BERT model is used to encode long texts using BiLSTM, and a head and tail pointer network is used for decoding.

[0092] As an embodiment of the present invention, preferably, in step three, the BERT model is a pre-trained model used to extract text entities. The BERT model is composed of stacked Enocder modules from multiple Transformer model structures; BiLSTM is a bidirectional long short-term memory network, consisting of a forget gate, an input gate, and an output gate, and its calculation formula is:

[0093] Step S301: Calculate the forget gate by multiplying the input text vector by the weight matrix and adding it to the bias vector. Then, process the result using the sigmoid function to calculate a vector representing the probability.

[0094] p t =sig(W i g[l t-1 ,x t ]+a i )

[0095] In the formula: x t L represents the input at time t. t-1 W represents the hidden state vector at time t-1, where sig is the sigmoid function; i and a i This represents the weight matrix and bias vector of the forget gate, vector p. t Representing probability, vector p t It can be 0 or 1, where 0 means to forget everything and 1 means to retain everything;

[0096] Step S302: Calculate the input gate. First, receive the current input text vector and the hidden state information from the previous time step. Process this information using the sigmoid function to obtain a vector representing probability values. A probability value of 0 means no information is updated, while a probability value of 1 means information is updated. Second, concatenate the received current input text vector with the hidden state information received from the previous time step. Process this information using the tanh function to obtain short-term memory units. Then, jointly update the short-term and long-term memory units to obtain new memory units. The calculation formula is as follows:

[0097] d t =sig(W i g[l t-1 ,xt ]+a n )

[0098]

[0099] In the formula: x t W represents the input at time t. n and a n Represents the weight matrix and bias vector of the input gate; S represents the short-term memory unit at time t; t represents the long-term memory unit at time t, and th represents the tanh function;

[0100] Step S303: Calculate the output gate. The output gate receives the current input text vector and the hidden state information from the previous time step. After processing with the sigmoid function, it obtains a vector representing probability values, indicating which information in the long-term memory unit needs to be output. Then, the long-term memory unit is processed using the tanh function, and finally, it is summed with O. t The dot product outputs a final result, calculated using the following formula:

[0101] G t =sig(W O g[l t-1 ,x t ]+a f )

[0102] h t =G t gth(C t )

[0103] In the formula: W O and a f This represents the weight matrix and bias vector of the output gate; l t This represents the hidden state information at time t.

[0104] As an embodiment of the present invention, preferably, the method for determining the relationship of all identified entities in step three is as follows:

[0105] Step S3001: Based on the Casrel model, construct a multi-layer head and tail pointer network, add a self-attention mechanism to fuse head entity information and sequence element information. The vector representing the head entity is mapped to two different vector spaces, generating M and N. The sequence element information is mapped to the query vector space, generating O. Then, a scaling dot product attention model is used to calculate the relevance coefficient between O and M. The calculation formula is as follows:

[0106]

[0107] In the formula: This represents the t-th head entity identified. This represents the t-th element in the sequence information;

[0108] Step S3002: Process vectors O and M using the distribution function y to obtain the energy score b. Use the calculated energy score to characterize the correlation between sequence elements and the head entity. The calculation expression is as follows:

[0109]

[0110] In the formula: Indicates the dimensional size of O and M;

[0111] Step S3003: Multiply the energy fraction b by the vector representing the head entity, and then concatenate it with the vector representing the sequence elements to generate a new vector. The column elements are represented, and then... The projection transformation is expressed as follows:

[0112] w = soft(b)

[0113]

[0114] In the formula: w represents the attention weight between the sequence element and the head entity, N sb It is a vector representing the head entity. This represents the new head entity representation vector after the head entity vector is processed and concatenated with the sequence element vector. P represents the head entity representation vector... Mapping to a vector space of relational dimension, where F represents the probability value of the correlation between the sequence element vector and the head entity vector in the relational dimension.

[0115] As an embodiment of the present invention, preferably, in step three, the CARE I model is used to determine all entities corresponding to the subject and object under a specified relation type based on a predefined relation type, and after extracting the corresponding triples, the Neo4j graph database is used to store the triples.

[0116] As an embodiment of the present invention, preferably, the specific process of generating functional descriptions of the components included in the product from the relevant product description in step four is as follows:

[0117] The encoder-decoder model receives the design document extracted in step one and outputs the functional descriptions of each component, thus obtaining the triplet of the relationship between function and structure.

[0118] The functional entities and structural entities are merged separately;

[0119] Retrieve the corresponding structural entities from the graph database and save the functional and structural relation triples to the graph database.

[0120] As an embodiment of the present invention, preferably, the specific process of knowledge retrieval and reuse in step five of the product concept design process is as follows:

[0121] Search for functional entities to find the desired function;

[0122] Based on the mapping relationship between function and structure, retrieve the set of design structures that can achieve the corresponding function;

[0123] For two interactive functions, retrieve the structures with connection paths in the corresponding two design structure sets, and take the union of the design structures.

[0124] A domain knowledge graph construction method based on the FBS conceptual design model is proposed. This method utilizes the FBS conceptual design model to back-analyze design documents, identifying design structures, design behaviors, design functions, and related feature attributes. By pre-defining the relationship types between these design elements, a design knowledge graph is constructed, providing stable and systematic design knowledge support for the product development process, thereby improving product development efficiency and reducing costs. Furthermore, this invention improves the CARE I model by constructing a multi-layer head-and-tail pointer network and adding a self-attention mechanism to fuse head entity information and sequence element information. This addresses the problem of a single entity corresponding to multiple entities in a document, improving the quality of knowledge graph generation.

[0125] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0126] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for constructing a domain knowledge graph based on a FBS concept design model, characterized in that, Includes the following steps: Step 1: Construct a design knowledge carrier acquisition module, obtain relevant design documents through web crawlers, preprocess the design document data, and establish a top-level design knowledge carrier database; Step 2: Construct a design knowledge representation module and build a design knowledge graph based on the FBS design model and the SRO representation structure; Step 3: Extract design knowledge entities from the design document and map them to the corresponding ontology. Classify the knowledge representation attributes of the entities and save the classified entities as triples to the graph database. Step 4: Construct a knowledge generation module to generate functional descriptions of the components included in the product from the relevant product descriptions: Step 5: Construct a knowledge application module to enable knowledge retrieval and reuse during the product concept design process; In step two, the specific process of constructing the design knowledge expression module and establishing the design knowledge graph is as follows: Based on the FBS design model, a top-down hierarchical analysis of the design product's design functions, design behaviors, and design structure is performed. The SRO-based representation structure uses triples as the knowledge representation framework, where both the subject and the object are knowledge ontologies. The triples are: subject, relation, and object. The knowledge ontology-based representation framework defines ontology types and relation types in the design of knowledge graphs, i.e., FBS-based types; In step three, the specific steps for saving the categorized entities as triples are as follows: Step S31: Perform named entity recognition on the design document. Use the BERT model, BiLSTM model, and head and tail pointer network through the feature encoding layer, sequence encoding layer, and label decoding layer respectively to identify the component names, corresponding attribute information, and behavioral information contained in the design document. Step S32: Perform relation judgment on all identified entities. Using the Casrel model based on predefined relation types, determine all entities corresponding to the subject and object under a specified relation type, thereby extracting the corresponding triples. Step S33: Based on the similarity of character vectors, fuse the same entities, vectorize the text content and attribute information of the entities, and then calculate the cosine similarity between the two. If the similarity value is higher than the threshold, they are regarded as the same entities. Step S34: Based on the triples, represent knowledge entities with nodes of the knowledge graph and relationship types with edges, and save them to the graph database. 2.The method of claim 1, wherein, In step two, the defined ontology types include: design function ontology, design behavior ontology, and design structure ontology. The relationship type definitions include: inclusion relationship and constraint relationship between design structures, and input and output relationship between design structures and behavior flow. 3.The method of claim 1, wherein, In step three, the named entity recognition method is as follows: a word vector encoder pre-trained based on the BERT model is used to encode long texts using BiLSTM, and a head and tail pointer network is used for decoding. 4.The method of claim 1, wherein, In step three, the method for determining the relationships of all identified entities is as follows: Step S3001: Based on the Casrel model, construct a multi-layer head and tail pointer network, add a self-attention mechanism to fuse head entity information and sequence element information, representing that the vector of the head entity is mapped to two different vector spaces, generating... The sequence element information is mapped to the query vector space to generate... Then, a scaled dot product attention model is used to calculate... The correlation coefficient is calculated using the following formula: ; In the formula: The first one to be identified Physical entity, Indicates the first in the sequence information One element; Step S3002: using the distribution function processing vector , obtaining the energy score , using the calculated energy score to represent the correlation between the sequence element and the head entity, and the calculation expression is as follows: In the formula: represents the dimension size of ; Step S3003: Energy Fraction Multiply the vector representing the head entity by the vector, then concatenate it with the vector representing the sequence elements to generate a new vector. The column elements are represented, and then... The projection transformation is expressed as follows: ; In the formula: This represents the attention weights between sequence elements and the head entity. It is a vector representing the head entity. The vector representing the head entity is processed and then concatenated with the sequence element vector to form a new head entity representation vector. This indicates that the head entity representation vector will be used. Mapping to a vector space with relational dimensions. This represents the probability value of the correlation between the sequence element vector and the head entity vector in the relation dimension.

5. The method of claim 4, wherein the method further comprises: In step three, the Casrel model is used to determine all entities corresponding to the subject and object under a specified relation type based on a predefined relation type. After extracting the corresponding triples, the Neo4j graph database is used to store the triples. 6.The method of claim 1, wherein, In step four, the specific process of generating functional descriptions of the components included in the product from the relevant product description is as follows: The encoder-decoder model receives the design document extracted in step one and outputs the functional descriptions of each component, thus obtaining the triplet of the relationship between function and structure. The functional entities and structural entities are merged separately; Retrieve the corresponding structural entities from the graph database and save the functional and structural relation triples to the graph database.

7. The method of claim 1, wherein the method further comprises: In step five, the specific process for realizing knowledge retrieval and reuse in the product concept design process is as follows: Search for functional entities to find the desired function; Based on the mapping relationship between function and structure, retrieve the set of design structures that can achieve the corresponding function; For two interactive functions, retrieve the structures with connection paths in the corresponding two design structure sets, and take the union of the design structures.

Images