A product digital twin model simulation method and system

Abstract

The present invention relates to a method for representing the digital twin representation of a product. The method comprises: obtaining a product's three-dimensional geometric information, physical property information, assembly topology information, and assembly process information; constructing a product ontology model based on the three-dimensional geometric information, physical property information, and assembly topology information; constructing an assembly process model based on the product ontology model and the assembly process information using a knowledge graph; constructing an assembly performance model based on the assembly process model using a machine learning algorithm; constructing a product function model based on the assembly performance model and the product's application information; and simulating and digitally displaying the product using the product function model. The present invention can achieve a realistic simulation of the product's functional capabilities.

Description

Technical Field

[0001] The present invention relates to the field of high-precision product assembly, and in particular to a product digital twin model simulation method and system. Background Art

[0002] High-precision products are widely used in aerospace, shipbuilding, automotive, and other systems and are core components of these systems. High-precision products require numerous assembly components, complex assembly processes, and high assembly precision. Furthermore, there is a complex nonlinear relationship between input and output. Traditional virtual simulation is primarily used in the assembly design phase. Virtual simulation is performed under an established ideal geometric model to derive assembly paths, assembly sequences, and analyze various assembly performance characteristics. However, the assembly process derived from simulation of an ideal geometric model is often crude and significantly deviates from the actual assembly process parameters, making the simulation data inappropriate for actual assembly operations.

[0003] Digital twin technology has developed rapidly in recent years, finding applications in product assembly simulation design, online quality prediction and process optimization during assembly, and assembly functional testing. Product assembly technology based on digital twins is poised to become a key research area. However, existing literature lacks systematic research on digital twin modeling methods for high-precision products. Summary of the Invention

[0004] The purpose of the present invention is to provide a product digital twin model simulation method and system to realize the use function of the real simulation product.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A product digital twin model simulation method, comprising:

[0007] Obtain product 3D geometry information, physical property information, assembly topology information, and assembly process information;

[0008] Constructing a product ontology model according to the three-dimensional geometric information, the physical property information and the assembly topology information;

[0009] Constructing an assembly process model using a knowledge graph based on the product ontology model and the assembly process information;

[0010] Building an assembly performance model using a machine learning algorithm based on the assembly process model;

[0011] Constructing a product function model based on the assembly performance model and the application information of the product;

[0012] The product functional model is used to simulate and digitally display the product.

[0013] Optionally, constructing an assembly performance model using a machine learning algorithm based on the assembly process model specifically includes:

[0014] Extracting features from the assembly process model using a neural network to obtain multiple feature information;

[0015] Mining the plurality of feature information using a grey correlation analysis method to obtain key assembly feature parameters;

[0016] The assembly process model is filled according to the key assembly feature parameters to obtain an assembly performance model.

[0017] Optionally, the calculation formula of the key assembly feature parameter is:

[0018]

[0019] Among them, f E (x i ) is the key assembly feature parameter, For maximum correlation, For maximum complementarity, Com(x i ,P i ,G) is the minimum redundancy, α and β are the scaling factors of maximum correlation and maximum complementarity, respectively.

[0020] Optionally, the assembly topology information is in triple form.

[0021] A product digital twin model simulation system, comprising:

[0022] Acquisition model, used to obtain the product's three-dimensional geometric information, physical property information, assembly topology information and assembly process information;

[0023] A first construction module is configured to construct a product ontology model based on the three-dimensional geometric information, the physical property information, and the assembly topology information;

[0024] A second construction module is used to construct an assembly process model using a knowledge graph based on the product ontology model and the assembly process information;

[0025] A third building module is used to build an assembly performance model using a machine learning algorithm based on the assembly process model;

[0026] a fourth building module, configured to build a product function model based on the assembly performance model and the application information of the product;

[0027] The simulation and display module is used to simulate and digitally display the product using the product function model.

[0028] Optionally, the third building block specifically includes:

[0029] A feature extraction unit, configured to extract features from the assembly process model using a neural network to obtain a plurality of feature information;

[0030] A mining unit, configured to mine the plurality of feature information using a grey correlation analysis method to obtain key assembly feature parameters;

[0031] A filling unit is used to fill the assembly process model according to the key assembly feature parameters to obtain an assembly performance model.

[0032] Optionally, the calculation formula of the key assembly feature parameter is:

[0033]

[0034] Among them, f E (x i ) is the key assembly feature parameter, For maximum correlation, For maximum complementarity, Com(x i ,P i ,G) is the minimum redundancy, α and β are the scaling factors of maximum correlation and maximum complementarity, respectively.

[0035] Optionally, the assembly topology information is in triple form.

[0036] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0037] The present invention provides a product digital twin model simulation method and system, which constructs a product ontology model based on the three-dimensional geometric information, the physical property information and the assembly topology information; constructs an assembly process model based on the product ontology model and the assembly process information using a knowledge graph; constructs an assembly performance model based on the assembly process model using a machine learning algorithm; constructs a product function model based on the assembly performance model and the application information of the product; and uses the product function model to simulate and digitally display the product, thereby realizing a true simulation of the product's usage function. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1A flow chart of a product digital twin model simulation method provided by the present invention;

[0040] Figure 2 This is a schematic diagram of the composition and relationship of the product digital twin model;

[0041] Figure 3 It is a schematic diagram of the product ontology model;

[0042] Figure 4 This is a schematic diagram of the assembly process knowledge graph;

[0043] Figure 5 It is a schematic diagram of the relationship between the product ontology model and the assembly performance model;

[0044] Figure 6 Use functional simulation schematics for product digital twins;

[0045] Figure 7 Schematic diagram of the process of modeling the representability of a product digital twin. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0047] The purpose of the present invention is to provide a product digital twin model simulation method and system to realize the use function of the real simulation product.

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] like Figure 1 As shown, the present invention provides a product digital twin model simulation method, including:

[0050] Step 101: Acquire the product's three-dimensional geometric information, physical property information, assembly topology information, and assembly process information. The assembly topology information is in the form of triples.

[0051] Step 102: Construct a product ontology model based on the three-dimensional geometric information, the physical property information, and the assembly topology information.

[0052] Step 103: Construct an assembly process model using a knowledge graph based on the product ontology model and the assembly process information.

[0053] Step 104: Construct an assembly performance model using a machine learning algorithm based on the assembly process model.

[0054] The step of constructing an assembly performance model using a machine learning algorithm based on the assembly process model specifically includes:

[0055] A neural network is used to extract features from the assembly process model to obtain a plurality of feature information.

[0056] The grey correlation analysis method is used to mine the plurality of feature information to obtain key assembly feature parameters.

[0057] The assembly process model is filled according to the key assembly feature parameters to obtain an assembly performance model. The calculation formula of the key assembly feature parameters is:

[0058]

[0059] Among them, f E (x i ) is the key assembly feature parameter, For maximum correlation, For maximum complementarity, Com(x i ,P i ,G) is the minimum redundancy, α and β are the scaling factors of maximum correlation and maximum complementarity, respectively.

[0060] Step 105: Construct a product function model based on the assembly performance model and the application information of the product.

[0061] Step 106: Use the product function model to simulate and digitally display the product.

[0062] The digital twin model constructed by the method provided by the present invention integrates the product ontology model, assembly process model, assembly performance model and product function model. Different technologies are used to represent each model. The product ontology model includes: three-dimensional geometric information, physical property information, and assembly topology information. The three-dimensional geometric information contains macroscopic geometric dimension information and microscopic surface morphology information. Its three-dimensional model is represented in the form of a GLTF format file. The physical property information is represented in the form of a key-value. The assembly topology information is represented in the form of a triple. The three-dimensional geometric information, physical property information and assembly topology information are stored in an XML file. The assembly process model adopts a representation method based on a knowledge graph, which can quickly retrieve and visualize the part assembly operation information. The assembly performance model uses a machine learning algorithm model to analyze the various quality characteristics exhibited during the product assembly process. The product function model can truly simulate the product's usage function and express it in a digital form. The method provided by the present invention can be applied to the model construction of hydraulic reversing valves and hydraulic proportional valves in a digital twin assembly system.

[0063] Specifically, if Figure 2 As shown, starting from the entire product assembly lifecycle, in a real physical model, the component parts and connector entities of the product are first determined according to the design requirements (represented as the product body model in the digital twin model). The components are then assembled to form an assembly (represented as the assembly process model in the digital twin model). To ensure the quality and functional reliability of the assembly, the product quality and function are further evaluated (represented as the assembly performance model and product function model in the digital twin model). This is constructed using existing 3D modeling software such as SolidWorks.

[0064] The established product digital twin model integrates the product ontology model 1.1, assembly process model 1.2, assembly performance model 1.3 and product function model 1.4, which is described in detail below.

[0065] Product ontology model 1.1 is the visual representation of the product in the twin space, such as Figure 3As shown, the information it contains is: three-dimensional geometric information 1.1.1, physical property information 1.1.2, and assembly topology information 1.1.3. Among them, the three-dimensional geometric information includes macroscopic geometric dimension information and microscopic surface morphology information; the physical property information includes the material information and quality information of the product; and the assembly topology information includes the connection and interference relationship between the various components of the product. ① Three-dimensional geometric information refers to dimensional information such as length, width, height, and curvature, as well as morphological information such as surface roughness; ② Physical properties refer to specific information about the material, including its density, mass, and stiffness; ③ Assembly topology information refers to the connection constraint relationship between the components (that is, which parts are connected and which parts are not connected).

[0066] Specifically, if Figure 3 As shown in the present invention, the three-dimensional geometric model adopts the GLTF file format, and the three-dimensional geometric information 1.1.1, physical property information 1.1.2 and assembly topology information 1.1.3 are stored in the XML file format.

[0067] Assembly process model 1.2 adopts a representation method based on knowledge graph. The knowledge graph is used to fuse the product ontology model and assembly process information in the form of a graph network for representation, which can quickly retrieve and visualize the parts assembly operation information. Rapid retrieval is achieved by searching for part IDs; at the same time, the "edges" in the knowledge graph are used to represent the connection relationship, that is, the surface relationship. The association relationship between some parts is not always presented on the surface, and there may be some direct or indirect relationships that are not easy to be found, especially between non-adjacent parts. The mining of hidden relationships is equivalent to the process of implicit knowledge discovery in knowledge graph technology. In the present invention, it refers to the use of neural network algorithms to solve potential hidden relationships through existing surface relationships.

[0068] Specifically, if Figure 4 The assembly process knowledge graph shown is composed of node-link-node triples. Specifically, Figure 7 (a), where the nodes (Pi) represent the components of the product body model, and the nodes (Ci-Pj) represent the components of the product component module (Ci), which contain the macroscopic geometric size information, microscopic surface morphology information, and physical property information of the components. Figure 7 As shown in Figures (b) and 7(c), the connections between nodes represent the assembly topology and assembly process information between components. In the assembly process knowledge graph, searching for part IDs allows for rapid retrieval of corresponding part information and adjacent parts assembly information. Converting assembly process information into assembly process knowledge not only establishes connections between existing adjacent parts (surface relationships), but also, through implicit knowledge discovery techniques, deeply explores connections between non-adjacent parts (hidden relationships).

[0069] Assembly Performance Model 1.3 uses machine learning algorithms to analyze the various quality characteristics exhibited during the product assembly process. Grey correlation analysis is used to identify correlations between part attributes and product performance, populating the process model knowledge graph generated in the previous steps. The output of the assembly performance model is the key parameters that influence product performance.

[0070] Specifically, the deep learning algorithm first uses the product assembly process model as input. Specifically, all the data information contained in the process model is used as input, including geometric dimension data, surface morphology data and other physical attribute data, as well as topological structure data and process information data. After neural network calculation and processing, the characteristic information of the product is extracted. Furthermore, an activation function is used to map the performance of the product. The characteristic information includes the geometric and physical characteristics of the body model. This mainly refers to the object characteristics, specifically including fit clearance, contact force, shear force, friction force, and pressure.

[0071] Specifically, if Figure 5 The figure shows how gray correlation analysis is used to determine the correlations between the geometric, physical, electromagnetic, and hydraulic factors of assembly components and various performance indicators. Each of these factors includes multiple parameters. For example, geometric information includes parameters in multiple dimensions, such as length, width, height, curvature, and roughness; physical factors also include material information and parameters in multiple dimensions, such as density. By comprehensively analyzing the different gray correlation sequences obtained using different data normalization methods, the parameters ranked at the top of each gray correlation sequence are identified as the primary geometric parameters influencing the performance indicators. This ultimately determines the key parameters influencing various performance indicators of multidisciplinary coupling and high-precision products. A deep learning algorithm is first used to determine the product's performance indicators. Next, gray correlation analysis is used to identify the influencing factors associated with these performance indicators. Performance is reflected through characteristic information. For example, large clearances manifest as structural instability; high friction manifests as valve switching irregularities; and high shear forces manifest as fragility. The following formula is used to determine the key influencing parameters among these influencing factors.

[0072] Among them, for each parameter i of each factor x, the one with the greatest correlation is screened by comprehensively considering the three correlation relationships Maximum complementarity and minimum redundancy Com(x i ,P i ,G) parameters are used as key assembly feature parameters.

[0073]

[0074] The key assembly characteristic parameter f refers to the influence of a parameter i of a factor x of a part on assembly performance, determined by a formula. The aforementioned "primary parameters" refer to geometric, physical, electromagnetic, and hydraulic parameters; they are the first few selected from multiple joint assembly characteristic parameters solved by the formula. In other words, the primary parameters include multiple key assembly characteristic parameters. Within a series of solved key assembly characteristic parameters, the influence of multiple parameters for each factor varies. This value is used to screen and determine the key parameters influencing the performance indicators mentioned above.

[0075] Furthermore, while each model, including the assembly performance analysis model and the process analysis model, reflects different aspects of high-precision product performance, they influence and constrain each other. The process model directly influences the coupling relationship between assemblies, and both jointly influence the quality performance after assembly. Simultaneously, product functionality and assembly performance requirements set design requirements for the process model. This paper investigates the correlation between performance indicators through statistical correlation analysis, analyzing the degree of linear fit for each indicator.

[0076] Product function model 1.4 can truly simulate the product's usage functions.

[0077] The performance model represents the specific quality characteristics of a product, while the function is an abstract characteristic of the product. The product function characteristic model established in this invention uses a digital representation. For example, if there is lubrication between the shaft hole, there is sliding kinetic energy; if there is relay transposition, there is switching function.

[0078] like Figure 7 As shown in (d), Figure 7 Based on (c), by assigning the same attribute value to specific nodes (represented by the diagonal filling method in the figure), it means that these nodes with the same mark constitute a functional module. Specifically, Figure 6 As shown, the input and output of high-precision products are converted into digital signals. The output represents the product's function. For example, an output of 1 indicates that the product can be reversed; an output of -1 indicates that the product cannot be reversed; an output of 2 indicates that the product can be manually controlled; an output of -2 indicates that manual control is not possible, and so on. The digital signal is the manifestation of the product's function. For example, for the reversing function of a solenoid valve, the "reversing command" is input. After passing through the functional module, the output changes from forward to reverse. The digital signal, such as the change from "1" to "-1", indicates the function has been achieved.

[0079] For complex functions, a high-dimensional matrix can be used to represent them. (For example, a 4x3x2 matrix can be used to represent a product with four types of functions, each type with three functional modules, and each module with two functions. For inconsistent numbers of functions or modules, zero padding is performed to maintain consistent matrix dimensions.)

[0080] Specifically, Figure 7 As shown in (d), the relationship between the basic feature model of the product and the functional model is established to visualize the changes of each component of the product during the assembly process, completing the change process from attribute components to feature modules and then to functional product modules.

[0081] Ultimately, the established product digital twin model can not only realistically visualize the flow of hydraulic oil, electric current, and electromagnetics, but can also be intuitively presented to the operator in a digital form, facilitating data analysis.

[0082] The present invention provides a product digital twin model simulation system, comprising:

[0083] Acquisition model is used to obtain the product's three-dimensional geometric information, physical property information, assembly topology information and assembly process information.

[0084] The first construction module is used to construct a product ontology model according to the three-dimensional geometric information, the physical property information and the assembly topology information.

[0085] The second construction module is used to construct an assembly process model using a knowledge graph based on the product ontology model and the assembly process information.

[0086] The third building module is used to build an assembly performance model based on the assembly process model using a machine learning algorithm.

[0087] The fourth building module is used to build a product function model according to the assembly performance model and the application information of the product.

[0088] The simulation and display module is used to simulate and digitally display the product using the product function model.

[0089] The third building block specifically includes:

[0090] The feature extraction unit is used to extract features from the assembly process model using a neural network to obtain multiple feature information.

[0091] The mining unit is used to mine the plurality of feature information using a grey correlation analysis method to obtain key assembly feature parameters.

[0092] A filling unit is used to fill the assembly process model according to the key assembly feature parameters to obtain an assembly performance model.

[0093] Among them, the calculation formula of key assembly feature parameters is:

[0094]

[0095] Among them, fE (x i ) is the key assembly feature parameter, For maximum correlation, For maximum complementarity, Com(x i ,P i ,G) is the minimum redundancy, α and β are the scaling factors of maximum correlation and maximum complementarity, respectively.

[0096] The assembly topology information is in the form of triples.

[0097] The product digital twin representable model constructed by the present invention integrates the product ontology model, assembly process model, assembly performance model and product function model, and can accurately represent the real assembly process of the product. The product ontology model includes: three-dimensional geometric information, physical property information, and assembly topology information; the assembly process model adopts a knowledge graph-based representation method to quickly retrieve and visualize product assembly operation process information in the form of a graph; the assembly performance model adopts a data-driven machine learning algorithm to analyze the various quality characteristics exhibited during the product assembly process; the product function model is used to truly simulate the product's usage function. The product digital twin model constructed by adopting the method of the present invention can accurately represent the product assembly status, quality characteristics and usage functions in the virtual space, and is used to guide the model construction of high-precision products in the digital twin assembly system. It has the following advantages:

[0098] (1) The constructed high-precision product digital twin model has high fidelity and truly depicts the structure, behavior, and function of the physical product.

[0099] (2) High-precision product digital twin models can be applied to digital twin assembly systems to achieve quality prediction and process optimization of the assembly process.

[0100] (3) The three-dimensional geometric model established based on GLTF is lightweight compared to other file formats, which effectively reduces the burden on computer operation and improves the smoothness of computer operation.

[0101] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0102] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A product digital twin model simulation method, characterized in that: include: Obtain product 3D geometry information, physical property information, assembly topology information, and assembly process information; Constructing a product ontology model according to the three-dimensional geometric information, the physical property information and the assembly topology information; Constructing an assembly process model using a knowledge graph based on the product ontology model and the assembly process information; Building an assembly performance model using a machine learning algorithm based on the assembly process model; Constructing a product function model based on the assembly performance model and the application information of the product; assigning the same attribute value to specific nodes, indicating that nodes with the same mark constitute a function module; and establishing an association relationship between the product basic feature model and the product function model; The product function model is used to simulate and digitally display the product. The product function model converts the input and output of the product into digital signals, and the digital signals are the manifestation of the product function; the input of the product is the functional instructions of the product; the output of the product is the function of the product or the realization of the function.

2. The product digital twin model simulation method according to claim 1, characterized in that: The step of constructing an assembly performance model using a machine learning algorithm according to the assembly process model specifically includes: Extracting features from the assembly process model using a neural network to obtain multiple feature information; Mining the plurality of feature information using a grey correlation analysis method to obtain key assembly feature parameters; The assembly process model is filled according to the key assembly feature parameters to obtain an assembly performance model.

3. The product digital twin model simulation method according to claim 2, characterized in that: The calculation formula of the key assembly feature parameters is: Among them, f E (x i ) is the key assembly feature parameter, For maximum correlation, For maximum complementarity, Com(x i ,P i ,G) is the minimum redundancy, α and β are the scaling factors of maximum correlation and maximum complementarity, respectively.

4. The product digital twin model simulation method according to claim 1, characterized in that: The assembly topology information is in the form of triples.

5. A product digital twin model simulation system, characterized by: include: Acquisition model, used to obtain the product's three-dimensional geometric information, physical property information, assembly topology information and assembly process information; A first construction module is configured to construct a product ontology model based on the three-dimensional geometric information, the physical property information, and the assembly topology information; A second construction module is used to construct an assembly process model using a knowledge graph based on the product ontology model and the assembly process information; A third building module is used to build an assembly performance model using a machine learning algorithm based on the assembly process model; A fourth construction module is configured to construct a product function model based on the assembly performance model and the application information of the product; assign the same attribute value to specific nodes, indicating that nodes with the same label constitute a function module; and establish an association relationship between the product basic feature model and the product function model; The simulation and display module is used to simulate and digitally display the product using the product function model. The product function model converts the input and output of the product into digital signals, and the digital signals are the manifestation of the product function; the input of the product is the functional instructions of the product; and the output of the product is the function of the product or the realization of the function.

6. The product digital twin model simulation system according to claim 5, characterized in that: The third building block specifically includes: A feature extraction unit, configured to extract features from the assembly process model using a neural network to obtain a plurality of feature information; A mining unit, configured to mine the plurality of feature information using a grey correlation analysis method to obtain key assembly feature parameters; A filling unit is used to fill the assembly process model according to the key assembly feature parameters to obtain an assembly performance model.

7. The product digital twin model simulation system according to claim 6, characterized in that: The calculation formula of the key assembly feature parameters is: Among them, f E (x i ) is the key assembly feature parameter, For maximum correlation, For maximum complementarity, Com(x i ,P i ,G) is the minimum redundancy, α and β are the scaling factors of maximum correlation and maximum complementarity, respectively.

8. The product digital twin model simulation system according to claim 5, characterized in that: The assembly topology information is in the form of triples.

Images