Intelligent algorithm integration application platform for typical structural member of airplane

Abstract

The invention belongs to the technical field of structural engineering knowledge management, and relates to an aircraft typical structural member intelligent algorithm integration application platform which is associated with a knowledge base and a database of typical structural members and comprises an application end and a management end. The application end comprises a part type selection module, a positioning element selection and parameter input module, a model preview and installation module and a model evaluation scoring module; the management end comprises a design knowledge and data access module, an algorithm management module and a template management module; a designer can call an algorithm designed by an algorithm designer at any time to complete and verify the model in the process of designing the model, and an algorithm designer can verify and correct the algorithm through a newly designed model, so that dual optimization of the model and the algorithm is realized, and the algorithm design efficiency is improved. The design efficiency of typical structural parts and functional algorithms is effectively improved, resources are saved, and meanwhile integrated management of the functional algorithms and models is achieved.

Description

Technical Field

[0001] The present application belongs to the technical field of structural engineering knowledge management, and in particular, relates to an intelligent algorithm integrated application platform for typical aircraft structural parts. Background Art

[0002] There are millions of typical aircraft structural parts. The design time of typical structural parts in aircraft structures accounts for more than 2 / 3 of the entire structural development cycle. More and more designers are applying the advantages of artificial intelligence technology to solve large-scale, grouped, and personalized aircraft structure designs to improve the design efficiency and quality of typical structural parts.

[0003] In the process of intelligent design of typical aircraft structural parts, designers have built a typical structural parts knowledge base and database around the effective use of historical typical structural parts design data and knowledge, which are used to store the structural form and structural parameters of the structural parts design, and on this basis, they have developed many algorithms and models for the intelligent design of typical structural parts. However, these algorithms and models are relatively isolated and developed for specific structural parts, and their application scenarios are limited. Effective management of many algorithms and models can further improve the application level of assets.

[0004] Therefore, how to effectively manage algorithms and models is a problem that needs to be solved. Summary of the invention

[0005] The purpose of this application is to provide an intelligent algorithm integrated application platform for typical aircraft structural parts to solve the problem of relative independence of algorithms and models developed for typical structural parts knowledge bases and databases in the prior art.

[0006] The technical solution of the present application is: an intelligent algorithm integrated application platform for typical aircraft structural parts, which is associated with a knowledge base and database of typical structural parts, including an application end and a management end, and the application end and the management end are connected through a functional open interface; the application end includes a part type selection module, a positioning element selection and input parameter module, a model preview and installation module and a model evaluation and scoring module; the part type selection module can obtain the design structural parts from the knowledge base of typical structural parts, the positioning element selection and input parameter module is used to obtain the input parameters and structural scheme of the current design structural parts through the management end; the model preview and installation module instantiates the input parameters and structural scheme to form a new model ; The model evaluation and scoring module is used to evaluate and score new models, and automatically store the generated model in the knowledge base and database of typical structural parts after the evaluation is qualified; the management end includes a design knowledge and data access module, an algorithm management module and a template management module; the design knowledge and data access module is used to access the structural form and structural parameters stored in the knowledge base and database of typical structural parts; the algorithm management module is used to select an adaptive algorithm according to the type of typical structural parts, and then select the corresponding typical structural part parametric design template, encapsulate it into a unified format and upload it to the model preview and installation module; the template management module is used to manage the parametric design template of typical structural parts.

[0007] Preferably, the application end further comprises a 3D navigation module, and the 3D navigation module is used to call installation auxiliary elements required for installation from a database of typical structural parts through the model of the selected designed structural part.

[0008] Preferably, the module for selecting positioning elements and inputting parameters first obtains the parts, structural elements, functional elements, constraint elements, instruction elements, function-structure elements, structure-structure elements and constraint-structure elements in the designed structural parts, and then determines the structural data, functional data and constraint data of the designed structural parts, packages them into a universal compressed package, and sends them to the management end. The management end converts the universal compressed package according to the specified format, decompresses and parses it to form data to be matched. The management end calls the corresponding algorithm through the algorithm calling interface, matches and analyzes the data to be matched with the typical structural parts knowledge base and database, and returns the input parameters and structural schemes recommended by the typical structural parts.

[0009] Preferably, the model evaluation and scoring module retrieves the scheme and parameter information corresponding to the new model, and sends a model evaluation request to the management end. After receiving the model evaluation request, the management end retrieves the evaluation algorithm corresponding to the designed structural component through the algorithm calling interface according to the scheme and parameter information therein and returns it to the model evaluation and scoring module. The evaluation algorithm compares and analyzes the input, output and strength information of the designed structural component, and performs evaluation and scoring to determine whether the scoring is qualified.

[0010] Preferably, the access content of the design knowledge and data access module includes the collection of historical data, the capture of later change data, and the compliance check of the accessed data. Based on the ontology definition of the structural part knowledge graph, it is determined whether the entities, relationships or attributes of the newly added knowledge are consistent with the definitions in the structural ontology, functional ontology, constraint ontology, sub-structural ontology, sub-functional ontology, and sub-constraint ontology. If they are inconsistent, they need to be modified according to the ontology definition.

[0011] Preferably, the management end also includes a knowledge analysis module, which includes a knowledge fusion submodule, a knowledge reasoning submodule and a parameter analysis submodule; the knowledge fusion submodule is used to fuse and align the equivalent knowledge of the corresponding structural elements of typical structural parts with structural similarity in the knowledge base. During fusion, only data with the same part classification and part type can be selected for fusion to obtain a new knowledge graph.

[0012] Preferably, the knowledge reasoning submodule calls the corresponding reasoning algorithm through the algorithm calling interface to perform knowledge reasoning based on the fused knowledge graph results, infers new knowledge or identifies erroneous knowledge based on the existing knowledge in the knowledge graph, and assists in reasoning new facts, new relationships and new rules; the parameter analysis submodule selects the analysis algorithm through the algorithm calling interface to analyze and process the missing or erroneous structural parameters of a single type of typical structural parts in the database.

[0013] Preferably, the determination of the structural similarity of the typical structural parts includes: numerical similarity calculation, semantic similarity calculation and interval similarity calculation.

[0014] Preferably, the specific method of calculating the numerical similarity is: using the numerical similarity algorithm of the Manhattan distance function, when the feature values ​​belong to different magnitudes, based on this, a distance standardization process is introduced after the distance value calculation, and the maximum value function is used to standardize the distance value, and its expression is:

[0015] in It represents the larger value of x and y, d(x,y) is the distance between x and y, and d*(x,y) is the standardized distance between x and y. The calculation formula after distance standardization is:

[0016] in Indicates that a numerical similarity algorithm is used to obtain the similarity between x and y.

[0017] Preferably, the specific method of calculating the semantic similarity is: the semantic similarity calculation is represented by a 0-1 algorithm, that is, when x=y, ,when hour, ; The distance calculation formula is expressed as: .

[0018] Preferably, the interval similarity calculation is specifically as follows: according to the definition of calculus, in the new design requirement characteristic value The similarity calculation is performed between the example design requirement characteristic value y and is set in the interval Any value within The average value of similar results between y and y, then the interval similarity calculation algorithm is: ; in, Indicates that an interval similarity algorithm is used to obtain the similarity between x and y.

[0019] Preferably, the algorithm management module includes an algorithm encapsulation submodule and an algorithm management submodule; the algorithm encapsulation submodule is used to encapsulate, create, manage and visualize a single functional algorithm when a functional algorithm is designed and uploaded to the management end, and to test the encapsulated functional algorithm, and publish it after the test is successful and the review is passed to form an algorithm call interface; the algorithm management submodule obtains the parameter configuration, input and output, test results and associated parts of each functional algorithm, and manages the combination of multiple algorithms related to the same type of parts, configures the algorithm running nodes, algorithm operation status monitoring, and the upstream and downstream relationship running timing of multiple algorithms, and associates with the resources of the template library.

[0020] The intelligent algorithm integrated application platform for typical aircraft structural parts of the present application enables model designers and algorithm designers to work on one platform. During the model design process, the designers can call the algorithms designed by the algorithm designers at any time to improve and verify the model, while the algorithm designers can verify and correct the algorithms through the newly designed models, thereby achieving dual optimization of models and algorithms, effectively improving the design efficiency of typical structural parts and functional algorithms, saving resources, and realizing integrated management of functional algorithms and models. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solution provided by the present application, the following is a brief introduction to the accompanying drawings. Obviously, the accompanying drawings described below are only some embodiments of the present application.

[0022] Figure 1 This is a schematic diagram of the overall structure of this application. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] An intelligent algorithm integration application platform for typical aircraft structural parts rewrites commonly used algorithms into a unified standard format according to different stages of computer-aided typical structural parts design, and integrates and manages them through software.

[0025] like Figure 1 As shown in the figure, the structural analysis of the typical structural parts knowledge base and database is as follows: there are five types of entity labels in the knowledge base of a single typical structural part, which are divided into: part, structural element, functional element, constraint element, and instruction element; there are eight types of relationship labels, which are: first-level structure analysis (part-structural element), second-level structure analysis (structural element-structural element), first-level function analysis (functional element-functional element), second-level function analysis (functional element-functional element), assembly (structural element-structural element), function mapping (structural element-functional element), feature operation (structural element-instruction element), and boundary constraint (structural element-constraint element).

[0026] Based on the parameter values ​​and design mechanism knowledge of typical structural parts design modeling, the database extracts the position parameters, constraint information, etc. hierarchically according to the structure tree in the knowledge base, and creates tables hierarchically for the extracted data information and stores them in the database. This will form a structural parameter database with a one-to-one mapping from the knowledge base to the database.

[0027] The whole system includes an application end and a management end, which are connected via a functional open interface.

[0028] First layer application end: The application end is used to realize the model calling / generating function and conducts secondary development based on Catia, including part type selection module, positioning element selection and input parameter module, recommended parameter and structural scheme confirmation module, 3D navigation module, model evaluation and scoring module and model preview and installation module.

[0029] The specific functions of each module are as follows: Part type selection module: used by designers to select the classification and specific form of design structural parts applicable to the current design scenario, where typical structural parts are standard parts and design structural parts are structural parts that designers need to use.

[0030] Select the positioning element and input parameter module: used to obtain the positioning elements and input parameters of the current design structure. The positioning elements are the relationship tags and entity tags of the corresponding design structure. Specifically: first obtain the parts, structural elements, functional elements, constraint elements, instruction elements, function-structure elements, structure-structure elements and constraint-structure elements in the design structure, then determine the structural data, functional data and constraint data of the design structure, package them into a universal compressed package, and send them to the management end. The management end converts the universal compressed package according to the specified format, decompresses and parses it to form the data to be matched. The management end calls the corresponding algorithm through the algorithm calling interface, matches and analyzes the data to be matched with the typical structural parts knowledge base and database, and returns the input parameters and structural solutions recommended by the typical structural parts.

[0031] Preferably, the functional data include: the role of the structure itself in the whole, the load borne by the structure, strength design index, stiffness design index, accuracy requirements, material properties, etc., which are the key to recommending the relevant structural elements that make up the structure.

[0032] As a specific implementation method, when designing the shear surface, the constraint data specifically includes: shape position parameters of the outer surface, frame position parameters, long stringer position parameters, etc. This is the key to further determine the specific position of the recommended overall structure in the overall environment and the geometric characteristics of the structure.

[0033] 3D navigation module: It is used to call the installation auxiliary elements such as points, lines and surfaces required for installation from the database of typical structural parts through the model of the selected design structural parts, and display them on the display screen. It can assist designers in selecting the points, lines, surfaces and other parts installation auxiliary elements required for the installation of the current model in a three-dimensional visual way to facilitate the assembly of subsequent models.

[0034] Recommended parameters and structural scheme confirmation module: used to display the input parameters and structural schemes recommended by typical structural parts returned by the management end algorithm. The recommended scheme is displayed in the form of a structure tree, and the recommended parameters are displayed in the form of <parameter name-recommended parameter value>. Designers can manually adjust the recommended structural parameter values ​​based on design experience, specifically by adjusting the structural parameter values ​​through parametric design templates. The design of parametric design templates is performed on the management end.

[0035] Model preview and installation module: Designers analyze the recommended parameters and structural schemes returned by the management end, select schemes and confirm parameters, and call the corresponding template on the management end for instantiation. Then, they can preview the installation of the model. After confirmation, they can insert the model into the assembly where the typical structural part model needs to be installed to complete the rapid design and assembly of the typical structural parts. If the assembly is successful, the scheme and parameter information of the model, including size, strength, etc., will be obtained.

[0036] Model evaluation and scoring module: After the designer analyzes and adjusts the scheme and parameter information returned by the system background, previews and adjusts the scheme and parameter information, and then sends a model evaluation request to the management end. After receiving the model evaluation request, the management end calls the evaluation algorithm corresponding to the designed structural component through the algorithm calling interface according to the scheme and parameter information and returns it to the model evaluation and scoring module. The evaluation algorithm compares and analyzes the input, output, strength and other information of the designed structural component, and evaluates and scores it to determine whether the score is qualified. The model with qualified score can automatically store its structural scheme and structural parameters in the knowledge base and database; the reason for the unqualified model is mainly that the input and output relationship does not meet the requirements and the strength does not meet the requirements; the modification plan is mainly to adjust the structural size.

[0037] Second layer management end: It includes design knowledge and data access module, knowledge analysis module, algorithm management module, template management module and general service module.

[0038] The design knowledge and data access module is used to access the structural forms and structural parameters stored in the knowledge base and database of typical structural parts. The access content includes the collection of historical data, the capture of later change data, and the compliance check of the accessed data. Based on the ontology definition of the structural parts knowledge graph, it is determined whether the entities, relationships or attributes of the newly added knowledge are consistent with the definitions in the structural ontology, functional ontology, constraint ontology, sub-structural ontology, sub-functional ontology, and sub-constraint ontology. If they are inconsistent, they need to be modified according to the ontology definition. The accurate knowledge addition and update results are published. After determining the knowledge that needs to be updated, it is necessary to judge whether the data exists. The knowledge and data that meet the requirements are displayed after being associated with the part name, part classification, part type and version to support subsequent algorithm application processing.

[0039] The knowledge analysis module includes a knowledge fusion submodule, a knowledge reasoning submodule and a parameter analysis submodule.

[0040] When the required design structural parts are typical structural parts with similar structures in the knowledge base, the knowledge fusion submodule performs knowledge fusion, and aligns the equivalent knowledge of the structural elements corresponding to the typical structural parts with similar structures in the knowledge base. During fusion, only data of the same part classification and part type can be selected for fusion to improve the integrity of the design structural parts. After knowledge fusion, a fused knowledge graph will be generated with a unique identifier, and it will be marked from which knowledge graphs it is fused. After fusion, the graph can be viewed in real time, and the structure tree, constraint tree, and function tree can be distinguished by different colors. At the same time, a single tree structure can also be displayed. It also provides search, reset and other functions.

[0041] The knowledge reasoning submodule calls the corresponding reasoning algorithm through the algorithm calling interface to perform knowledge reasoning based on the fused knowledge graph results, infers new knowledge or identifies erroneous knowledge based on the existing knowledge in the knowledge graph, and assists in reasoning new facts, new relationships and new rules; the parameter analysis submodule selects the analysis algorithm through the algorithm calling interface to analyze and process the missing or erroneous structural parameters of a single type of typical structural parts in the database.

[0042] Preferably, the determination of the structural similarity of typical structural parts includes: Feature similarity analysis, similar element numerical analysis, quantitative analysis and system similarity analysis. Feature similarity analysis is mainly used to solve the problem of matching the acquired constraints and functions with the same internal knowledge, ensuring that the system can intelligently recommend similar structural parts that meet the input information when there is no stub for the input of the recommended structural parts. The components and features of systems of different types and levels are not exactly the same, but there is a coupling relationship between these systems. The properties and features of some components of these systems are common, forming the similarity of the entire system. There are also similar features or elements between systems, so the necessity of similarity analysis is self-evident.

[0043] For new design requirements, in the actual design process, the way to describe the design problem is not unique, but more about the description object, the designer's familiarity with the design product, the description content, the characteristics and features, and the description forms are varied, including numerical, interval, semantic, etc. According to the actual situation of conceptual design, from the perspective of the description method of new design requirements, there are mainly three types of similarity calculation methods, namely numerical similarity calculation, semantic similarity calculation and interval similarity calculation. For design requirements, since design knowledge comes from the completed design plan, there are few uncertainty problems, so the demand layer information of design knowledge is mostly expressed in semantics. In the actual product design process, the design requirements are often not so certain. This is especially true for conceptual design. Designers obtain design requirements from customers or market research at the beginning of conceptual design, but because customers are not familiar with product field design knowledge, they can often only put forward some fuzzy requirements, such as better sealing and lower use cost. The traditional similarity calculation method cannot process the problem or the processing result has a very large error. In this case, it is necessary to use the fuzzy similarity calculation method to integrate the similarity analysis into the entire intelligent design system to solve the problem of matching the acquired constraints and functions with the same internal knowledge, and ensure that the system can intelligently recommend similar structural parts that meet the input information when there is no stub in the input of the recommended structural parts. It can also improve the coupling and robustness of the system, and improve the robustness and scalability of making intelligent recommendation solutions for new products and new demands.

[0044] The specific method of numerical similarity calculation is: using the numerical similarity algorithm of the Manhattan distance function. When the feature values ​​belong to different magnitudes, the error of the similarity calculation based on the distance function is a bit large. Based on this, a distance standardization process is introduced after the distance value calculation, and the maximum value function is used to standardize the distance value. Its expression is as shown in formula (4.1). It represents the larger value of x and y, d(x,y) is the distance between x and y, and d*(x,y) is the standardized distance between x and y. The calculation after distance standardization is as shown in formula (4.2).

[0045] (4.1) (4.2) in Indicates that a numerical similarity algorithm is used to obtain the similarity between x and y.

[0046] The specific method of semantic similarity calculation is: Semantic similarity calculation is represented by 0-1 algorithm, that is, when x=y, ,when hour, According to the definition of Manhattan distance function, semantic similarity calculation is to some extent a special case of numerical similarity algorithm, and its distance calculation formula is expressed as formula (4.3).

[0047] (4.3) The specific method for calculating interval similarity is: When the new design requirement value is expressed as When , the interval similarity calculation algorithm is used. According to the definition of calculus, the characteristic value of the new design requirement is The similarity calculation between the example design requirement feature value y can be considered to be in the interval Any value within The average value of similar results between y and y. Therefore, the interval similarity calculation algorithm can be expressed as formula (4.4).

[0048] (4.4) in, Indicates that an interval similarity algorithm is used to obtain the similarity between x and y.

[0049] The knowledge reasoning submodule calls the corresponding reasoning algorithm through the algorithm calling interface to perform knowledge reasoning based on the fused knowledge graph results. It infers new knowledge or identifies erroneous knowledge based on the existing knowledge in the knowledge graph, and assists in reasoning new facts, new relationships, and new rules. The reasoning algorithm is an existing algorithm. The results of reasoning need to be processed as valid or invalid to facilitate subsequent knowledge recommendation. Only one valid reasoning data is allowed for the same part classification and part type. After reasoning, a real-time viewing function is provided, and the structure tree, constraint tree, and function tree are distinguished by different colors. At the same time, the structure of a single tree can also be displayed. It also includes functions such as search and reset.

[0050] The parameter analysis submodule selects the analysis algorithm through the algorithm call interface to analyze and process the missing or incorrect structural parameters of a single type of typical structural parts in the database. During the analysis, the missing structural parameter data of a certain type of structural parts is first selected, and then the system selects the parameter analysis algorithm according to the task type, and further determines whether the current algorithm is available according to the current algorithm input requirements. If available, the data analysis is performed based on the parameter analysis algorithm. After the data analysis is successfully completed, the analysis results will be automatically generated, and the parameter information will be updated after review for subsequent recommendation. If it is not available, switch the algorithm and try again. It also includes functions such as search, reset, create analysis, view diagrams, view analysis result data, and download.

[0051] The algorithm management module is used to select the appropriate algorithm according to the typical structural part type, and then select the corresponding typical structural part parametric design template to encapsulate it into a unified format and upload it to the model preview and installation module. The algorithm management module includes an algorithm encapsulation submodule and an algorithm management submodule. Algorithms include analysis algorithms, evaluation algorithms, derivation algorithms, etc. When designers design a functional algorithm and upload it to the management end, they need to set the functional algorithm to facilitate the recognition of this platform. For example, if it is marked as an evaluation algorithm during parameter configuration, it will be selected as one of the candidates when the evaluation algorithm is called.

[0052] The algorithm encapsulation submodule is used to realize the functions of encapsulation, creation, management and visualization of single-function algorithms. A wizard-style approach is used to guide designers to complete the four steps of packaging-parameter configuration-parameter testing-algorithm review. The unified format of the artificial intelligence algorithm is encapsulated through parameter configuration. In the process of algorithm creation, the three methods of initialization, execution and release of resources after calling must be implemented, and an abnormal capture and processing mechanism is designed. At the same time, the algorithm master file, resource file and algorithm model file are placed in the same directory and packaged in ZIP format. The algorithm application link, algorithm language, algorithm master file and algorithm description are marked before uploading the algorithm.

[0053] After the algorithm is packaged, it is uploaded. After uploading, the next step is to describe the input and output parameters of the algorithm and the algorithm test data. The input and output parameters need to specify the parameter name and type, and then go to the next step of algorithm testing. Success and failure will be prompted during the test. If the test fails, the system will return the error information thrown by the algorithm. During the test and after the test fails, the algorithm may be upgraded, modified, deleted, etc. After the test is successful, the algorithm will be submitted for review. After the review is passed, the algorithm can be published to form an algorithm call interface and published. After the release, the algorithm can be upgraded, modified, deleted, retrieved, browsed, and unpublished, so that it can be called by the application end.

[0054] The algorithm management submodule obtains the parameter configuration, input and output, test results and related parts of each functional algorithm, manages multiple algorithms related to the same type of parts in combination, configures the algorithm running nodes (such as input), algorithm operation status monitoring (such as output), and the running timing of multiple algorithms' upstream and downstream relationships, and associates them with the resources of the template library, so that the corresponding algorithm can be quickly called when performing algorithm analysis on this type of parts in the future. After definition, it can be published, and after publication, it has management functions such as editing, deleting, and upgrading the algorithm model.

[0055] The template management module is used to manage the parametric design templates of typical structural parts. For a certain type of parts, the designed parametric design templates are managed uniformly. Specifically, fill in the template file name, part classification, and part type in the system to bind the applicable structural parts of the template, and fill in the input parameters, input reference parameters, and designer parameter classification to determine the parameters required to instantiate the template. Each parameter should specify the parameter name and type, and submit it for review after completion. After the review is passed, the template becomes published and can be used. It also provides functions such as editing and deleting parametric templates.

[0056] The general service module is based on the personalized permission control model of designer-role-function, which realizes the flexible display of functional interface according to the different division of labor of designers in the system. The system provides maintenance for designer information. It mainly includes functions such as adding, modifying, deleting, enabling, disabling, resetting passwords, assigning roles, and querying. The system provides maintenance for role information, mainly including functions such as adding, modifying, deleting, permission setting, and querying. The system provides maintenance for organizational information, mainly including functions such as adding, modifying, deleting, and querying.

[0057] As a specific implementation method, a specific example is used to illustrate: The application side specifically uses CAA secondary development technology to implement application side interface design, interface elements, function event template creation, input and output parameter determination, 3D navigation and other functions, and interacts with the management side through RESTful API (an architecture style of application program interface (API) that uses HTTP requests to access or use data). The following takes the design of electrical corner plate structure as an example to illustrate the specific implementation method.

[0058] 1) Part type and form selection: Design in the Catia part design module based on the CatiaCAA method, and use the toolbar to select various types of part-level typical structural parts and different forms of this type supported by the system.

[0059] 2) Select positioning elements and input parameters: The Catia end obtains the model's empty axis, positioning points and installation surfaces by selecting the skin, then checks whether the input conditions are complete, calculates the intermediate elements, and returns the features that need to be instantiated. Then, the HTTPGet request method is used to package this information into Json format as the parameter of the recommendation algorithm to call the management end recommendation algorithm interface request path. The recommendation algorithm obtains the structural scheme and structural parameter information of the part through a series of algorithm analysis and returns it.

[0060] 3) Call the CATpart model file of the part: After obtaining the parameter information through the recommendation algorithm, the designer uses HTTPPost to reference the corresponding CATPart template file from the template library according to the part classification and part type, and uses the structural parameter information of the recommendation result to instantiate the template to generate a new part model.

[0061] 4) Call the evaluation algorithm: The solutions given by the recommendation algorithm are evaluated, and the recommended structural parameter information is packaged in Json format using the HTTPGet request method as the parameter of the evaluation algorithm to call the management-side evaluation algorithm interface request path. The evaluation algorithm obtains the evaluation result information of the generated parts through a series of algorithm analysis, including performance analysis and evaluation, processability analysis and evaluation, three-dimensional analysis and evaluation, and weight analysis and evaluation.

[0062] 5) Save the evaluation results: The evaluation results are saved according to the part type in the reasoning knowledge information with the reasoning status as "valid".

[0063] The management side is developed through Javaweb technology to achieve data and knowledge management, algorithm packaging and management, model packaging and management, and designer management.

[0064] 1) Design knowledge and data access module: Specifically, the part-level change data is captured by calling the built-in process of Neo4j graph database CALL through a timed patrol method, and the nodes and relationships of the changed parts are obtained through the knowledge graph language Cypher query. First, the structure tree is traversed according to the structural parsing level in a breadth-first manner, and then each node of the structure tree is checked to see whether there are function nodes and constraint nodes of the corresponding level. If so, it is represented by 1, otherwise it is represented by 0. After the traversal, if there is 0 completeness missing, the part is not imported. Otherwise, the parameter information associated with each node is queried in the typical structural parts database and knowledge base (without integrity judgment). If it exists, the part is imported. Then, the change data is pulled to the current system for subsequent operations.

[0065] 2) Knowledge analysis module: Knowledge fusion submodule: This algorithm is called when two or more data are selected for knowledge fusion. This algorithm is called when the fusion result after knowledge fusion is used for knowledge reasoning. The system connects to the neo4j account username, String type, the password for connecting to neo4j, String type, the name of the knowledge graph to be fused, the collection and segmentation of String type (names cannot be repeated), the name of the fused knowledge graph, String type (names cannot be repeated), and after the algorithm completes the fusion, the result is stored in the specified knowledge graph. The part classification and part type of the selected data must be consistent during fusion; after fusion, the graph and data of the fusion result can be viewed and uniquely identified in the form of Fus+ time. The constructed knowledge graph is stored in the neo4j graph database, and the D3.js third-party library is used to visualize the fusion results of the neo4j graph database.

[0066] Knowledge reasoning submodule: For the fusion result after knowledge fusion, the algorithm is called during the reasoning. The system uses the username of the account connected to neo4j, which is of String type, the password of the password connected to neo4j, which is of String type, the name of the knowledge graph to be inferred, which is a collection of String type, divided by (names cannot be repeated), and the name of the knowledge graph after reasoning, which is of String type (names cannot be repeated). After the algorithm completes the reasoning, the result is stored in the specified knowledge graph. Reasoning is only performed in the knowledge graph that is valid after fusion; after reasoning, the graph and data of the fusion result can be viewed and uniquely identified in the form of Rea+time. The constructed knowledge graph is stored in the neo4j graph database, and the D3.js third-party library is used to visualize the reasoning results of the neo4j graph database.

[0067] Parameter analysis submodule: The system selects data of a specific structural part type from the accessed data and performs data analysis based on the parameter analysis algorithm. The system selects a typical structural part as the object, constructs the model input and model output parameters in a string in Json format, selects the analysis algorithm, and displays the results in text form after the algorithm is completed. The algorithm maintenance personnel determines the path for the file storage based on the results. After analysis, the analysis data can be viewed and uniquely identified in the form of Ana+time.

[0068] 3) Algorithm encapsulation module: The algorithm encapsulation is implemented in Python programming language. During the creation process, the three methods of initialize(), execute() and finalize() must be implemented, and an exception capture and processing mechanism must be designed. After the algorithm is created, it is packaged into a Zip file format, and the algorithm metadata information including the algorithm name, algorithm type, algorithm file path, main algorithm file name and a description of the algorithm execution process are filled in the system for upload. The system automatically parses and stores it in the specified file path. After the upload is successful, the algorithm status defaults to "working". The algorithm name must be unique and no duplication is allowed; the algorithm types include fusion, reasoning, parameter analysis, recommendation, evaluation and other types; the main algorithm file name must exist in the algorithm file; the algorithm status is divided into two types: "working" and "published".

[0069] Algorithm management creates intelligent design algorithms for parts by building names, part classifications, and part types, and selects relevant algorithms in the nodes of structural part intelligent design fusion, reasoning, parameter analysis, recommendation, and evaluation in turn, configures the algorithms, and selects template files for automatic generation of parts.

[0070] 4) Template management module: Use Powercopy or Udf in the Catia knowledge engineering module to create parametric design templates for various typical structural parts.

[0071] 5) General service module: The system has the roles of super administrator, system administrator, data construction personnel and algorithm maintenance personnel. The super administrator has the applicable authority for all system functions, the system administrator has all functions except the general module, and the data construction personnel has the authority to design the data and knowledge management modules. The algorithm maintenance personnel has the functional authority of their own algorithm management and template management modules.

[0072] Finally, it should be noted that: the drawings of the embodiments disclosed in the present invention only involve structures related to the embodiments disclosed in the present invention, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other; Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent algorithm integrated application platform for typical aircraft structural parts, characterized by: It is associated with the knowledge base and database of typical structural parts, including an application end and a management end, and the application end and the management end are connected through a functional open interface; the application end includes a part type selection module, a positioning element selection and input parameter module, a model preview and installation module, and a model evaluation and scoring module; the part type selection module can obtain the design structural parts from the knowledge base of typical structural parts, and the positioning element selection and input parameter module is used to obtain the input parameters and structural scheme of the current design structural parts through the management end; the model preview and installation module instantiates the input parameters and structural scheme to form a new model; the model evaluation and scoring module is used to evaluate and score the new model, and automatically store the generated model in the knowledge base and database of typical structural parts after the evaluation is qualified; the management end includes a design knowledge and data access module, an algorithm management module, and a template management module; The design knowledge and data access module is used to access the knowledge base of typical structural parts and the structural forms and structural parameters stored in the database; The algorithm management module is used to select an adaptive algorithm according to the type of typical structural parts, and then select the corresponding typical structural parts parametric design template, encapsulate it into a unified format and upload it to the model preview and installation module; The template management module is used to manage the parametric design templates of typical structural parts.

2. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 1, characterized in that: The application end also includes a 3D navigation module, which is used to call installation auxiliary elements required for installation from a database of typical structural parts through the model of the selected designed structural part.

3. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 1, characterized in that: The module for selecting positioning elements and inputting parameters first obtains the parts, structural elements, functional elements, constraint elements, instruction elements, function-structure elements, structure-structure elements and constraint-structure elements in the designed structural parts, and then determines the structural data, functional data and constraint data of the designed structural parts, packages them into a universal compressed package, and sends them to the management end. The management end converts the universal compressed package according to the specified format, decompresses and parses it to form data to be matched. The management end calls the corresponding algorithm through the algorithm calling interface, matches and analyzes the data to be matched with the typical structural parts knowledge base and database, and returns the input parameters and structural schemes recommended by the typical structural parts.

4. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 1, characterized in that: The model evaluation and scoring module retrieves the scheme and parameter information corresponding to the new model, and sends a model evaluation request to the management end. After receiving the model evaluation request, the management end retrieves the evaluation algorithm corresponding to the designed structural component through the algorithm calling interface according to the scheme and parameter information therein and returns it to the model evaluation and scoring module. The evaluation algorithm compares and analyzes the input, output and strength information of the designed structural component, and evaluates and scores it to determine whether the score is qualified.

5. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 1, characterized in that: The access content of the design knowledge and data access module includes the collection of historical data, the capture of later change data, and the compliance check of the accessed data. Based on the ontology definition of the structural part knowledge graph, it is determined whether the entities, relationships or attributes of the newly added knowledge are consistent with the definitions in the structural ontology, functional ontology, constraint ontology, sub-structural ontology, sub-functional ontology, and sub-constraint ontology. If they are inconsistent, they need to be modified according to the ontology definition.

6. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 1, characterized in that: The management end also includes a knowledge analysis module, which includes a knowledge fusion submodule, a knowledge reasoning submodule and a parameter analysis submodule; The knowledge fusion submodule is used to fuse and align the equivalent knowledge of the structural elements corresponding to typical structural parts with structural similarity in the knowledge base. During fusion, only data with the same part classification and part type can be selected for fusion to obtain a new knowledge graph.

7. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 6, characterized in that: The knowledge reasoning submodule performs knowledge reasoning by calling the corresponding reasoning algorithm through the algorithm calling interface based on the fused knowledge graph results, infers new knowledge or identifies erroneous knowledge based on the existing knowledge in the knowledge graph, and assists in reasoning new facts, new relationships and new rules; the parameter analysis submodule selects the analysis algorithm through the algorithm calling interface to analyze and process the missing or erroneous structural parameters of a single type of typical structural parts in the database.

8. The intelligent algorithm integrated application platform for typical aircraft structures as claimed in claim 6, characterized in that: The judgment of the structural similarity of the typical structural parts includes: numerical similarity calculation, semantic similarity calculation and interval similarity calculation.

9. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 8, characterized in that: The specific method of numerical similarity calculation is: using the numerical similarity algorithm of the Manhattan distance function, when the feature values ​​belong to different magnitudes, based on this, a distance standardization process is introduced after the distance value calculation, and the maximum value function is used to standardize the distance value, and its expression is: ； in It represents the larger value of x and y, d(x,y) is the distance between x and y, and d*(x,y) is the standardized distance between x and y. The calculation formula after distance standardization is: ； in Indicates that a numerical similarity algorithm is used to obtain the similarity between x and y.

10. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 9, characterized in that: The specific method of semantic similarity calculation is: semantic similarity calculation is represented by 0-1 algorithm, that is, when x=y, ,when hour, ; The distance calculation formula is expressed as: 。 11. The intelligent algorithm integrated application platform for typical aircraft structures according to claim 8, characterized in that: The interval similarity calculation is specifically as follows: according to the definition of calculus, in the new design requirement characteristic value The similarity calculation is performed between the example design requirement characteristic value y and is set in the interval Any value within The average value of similar results between y and y, then the interval similarity calculation algorithm is: ； in, Indicates that an interval similarity algorithm is used to obtain the similarity between x and y.

12. The intelligent algorithm integrated application platform for typical aircraft structural parts according to claim 1, characterized in that: The algorithm management module includes an algorithm encapsulation submodule and an algorithm management submodule; the algorithm encapsulation submodule is used to encapsulate, create, manage and visualize a single functional algorithm when a functional algorithm is designed and uploaded to the management terminal, and to test the encapsulated functional algorithm. After the test is successful and the review is passed, it will be released to form an algorithm call interface; The algorithm management submodule obtains the parameter configuration, input and output, test results and associated parts of each functional algorithm, manages the combination of multiple algorithms related to the same type of parts, configures the algorithm running nodes, algorithm operation status monitoring, the upstream and downstream relationship running timing of multiple algorithms, and associates with the resources of the template library.

Images