A Method for Constructing a Digital Twin Model for Assembly Shops

Abstract

This invention discloses a method for constructing a digital twin model for an assembly workshop, relating to the field of digital twins. First, this invention employs a multi-scale element representation method, representing all elements within the workshop at four scales: elements, processes, production lines, and the workshop itself, thus constructing a multi-scale digital twin model of the assembly workshop. Then, starting from a multi-dimensional concept, it relies on the Model-Based Definition (MBD) method to construct digital twin models at four levels: geometric, physical, behavioral, and rule-based. Specifically, it creates data node models, represents the data evolution process based on these models, and achieves the fusion of multi-scale and multi-dimensional models on the basis of the multi-dimensional data model.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and more specifically to a method for constructing a digital twin model for an assembly workshop. Background Technology

[0002] The digital twin system for assembly workshops takes into account the characteristics of discrete manufacturing and improves the existing digital twin technology framework. It takes a multi-scale and multi-dimensional approach and incorporates the features of Model Based Definition (MBD) technology to build a digital twin model of the assembly workshop, providing strong support for discrete manufacturing enterprises to utilize digital twin technology.

[0003] It should be noted that digital twin technology makes production simulation in manufacturing workshops more convenient and intelligent. Simultaneously, digital twin systems enable assembly workshops to utilize efficiently collected real-time data for production scheduling simulation, significantly reducing the time and economic costs for enterprises to conduct mid-term scheduling when facing order changes and resource fluctuations. By incorporating MBD (Model-Based Design) technology, digital twin models can be lightweight, consuming fewer computing resources for simulation and control.

[0004] Current analyses of digital twin manufacturing processes and business interactions generally focus on the autonomous or collaborative interactions of multiple stakeholders, including people, machines, materials, laws, and environments. The classification of stakeholders follows traditional discrete workshop entity research practices, emphasizing entity attributes and lateral relationships. This leads to unclear characteristics of the corresponding stakeholders within the overall workshop structure, hindering the establishment of a proper correspondence between physical entities and information space models and services. Furthermore, existing models lack a deep understanding of workshop data collection and evolution patterns, resulting in a disconnect between the logic and rules within the digital twin model and real-world data deduction.

[0005] Currently, in the research on digital twin modeling methods, Wu Pengxing et al., in their study on the visualization and real-time monitoring method of discrete workshops, proposed a "Visual Real-Time Monitoring Method for Discrete Manufacturing Workshops Based on Digital Twins." This method employs data modeling based on Automation ML and OPC UA, establishing a data model from the perspectives of resources, processes, and products. However, it lacks sufficient characterization of the vertical management structure of the workshop and cannot reflect the inheritance relationships between different production levels within the entire workshop. Ding Kai et al., combining the definition of intelligent manufacturing space, proposed "Multi-dimensional and Multi-scale Intelligent Manufacturing Space Based on Digital Twins and Its Modeling Method," characterizing the digital twin model from a multi-dimensional and multi-scale perspective. However, it lacks sufficient consideration of all elements of the workshop, only modeling from generalized features, and lacks practical comparative effects. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a method for constructing digital twin models for assembly workshops. This method enables the construction of digital twin models for assembly workshops, reducing the workload of corresponding digital twin modeling and development.

[0007] To achieve the above objectives, the technical solution adopted by the present invention to solve its technical problem is as follows:

[0008] A method for constructing a digital twin model for an assembly workshop includes the construction of a multi-scale model, the construction of a multi-dimensional model, and corresponding model fusion and model lightweighting methods.

[0009] Step 1: Construction of a multi-scale digital twin model:

[0010] Multi-scale refers to the four basic industrial scales of elements, processes, production lines, and workshops. The hierarchical modeling method is used to model from the smallest element scale to the largest scale.

[0011] Step 1.1: The elements are divided into two main categories: entity class and space class. The entity class includes four elements: collection equipment, logistics equipment, assembly equipment, and operator. The space class includes three elements: storage space, production space, and transportation space. Analyze and establish a formal expression that includes the types, attributes, and interrelationships of elements.

[0012] Step 1.2: Construct an element-based process mapping model, and divide the processes into four categories based on actual production: preparation, inspection, operation and cleaning. Different combinations of elements reflect different process functions and attributes.

[0013] Step 1.3: Based on processes and elements, construct a formal representation of the production line; organize the processes using the process flow model, and then supplement them with different independent elements. Use different process organization forms to determine the organizational form of the production line, and determine the attributes and interrelationships of the production line.

[0014] Step 1.4: Combine the three scale models of elements, processes, and production lines to construct an overall framework at the workshop scale, ensuring the consistency of internal logic and the independence of functions within the workshop;

[0015] Step 2: Construction of a multi-dimensional digital twin model;

[0016] Modeling is conducted at four levels: geometric, physical, behavioral, and rule-based. Model-Based Definition (MBD) technology uses the geometric model of mechanical equipment as a foundation, attaching process information, physical information, and numbering attribute information to the geometric model to achieve lightweight and efficient modeling. Drawing inspiration from the MBD method, this paper proposes a digital twin modeling technique based on Bill of Process (BOP). Using the geometric model as a foundation, information involving physical, behavioral, and rule models, such as process information and product physical information, is attached to the geometric model. The formal description of this virtual mapping is as follows:

[0017] V = M + P + B + I;

[0018] P Δ B Δ I→M;

[0019] Based on the models of different granularities established in the multi-scale modeling stage, data nodes can be formed based on data of different granularities. The establishment of data nodes first requires determining the data and information that need to be covered by nodes at different levels, and then completing the establishment of node flows between different nodes, clarifying the composition logic and compositional relationships between nodes;

[0020] Step 2.1: Establish data nodes based on the multi-scale model. This mainly involves establishing basic data nodes based on element and process models, identifying data nodes that match different types of elements and processes, and data nodes with independent control functions.

[0021] Step 2.2: Based on the characteristics of the constructed data nodes, the evolution process of the data nodes is analogized to the growth process of a tree to determine the evolution logic of the data nodes;

[0022] Step 2.3: Based on the theory of multi-source information fusion, a three-level data fusion model is established according to the general physical information fusion framework. Combined with the multi-scale model, a three-level data fusion model is formed, which realizes metadata fusion at the process level, feature-level data fusion at the production line level, and decision-level data fusion at the workshop level, thus completing the fusion of multi-scale and multi-dimensional models.

[0023] Furthermore, the specific method of step 1.1 is as follows:

[0024] The data acquisition equipment includes video surveillance equipment at the assembly site, RFID data acquisition equipment, and heterogeneous sensors. The data acquisition equipment uses three state variables—on, off, and abnormal—to reflect its own state; the characteristics are as follows:

[0025] SEC j ={ETC id EC type CEC st}

[0026] EC type ={1,2,3,4}

[0027] CEC st ={-1,0,1}

[0028] In the formula SEC j Let j be the j-th data collection device, ETC id This indicates the unique serial number of the data acquisition device within the production workshop, EC. type This indicates the specific equipment category of the data acquisition device, where 1 represents video surveillance equipment, 2 represents RFID data acquisition equipment, 3 represents heterogeneous sensors, and 4 represents other data acquisition devices; CEC st This indicates the current operating status of the data acquisition device, where -1 indicates an abnormal state, 0 indicates a shut-down state, and 1 indicates a startup state.

[0029] Logistics equipment mainly includes AGVs and conveyor belts on the production site. Their primary function is to transfer and transport materials. Logistics equipment needs to transfer materials, semi-finished products, and finished products according to a certain rhythm, as shown below.

[0030] C l ={C AGV C con C el}

[0031] Where C AGV Indicates AGV,C con Indicates a conveyor belt, C el Indicates other logistics equipment;

[0032] An AGV (Automated Guided Vehicle) encompasses translational and rotational movements on a plane, and also possesses various transfer functions on its upper surface. Its data includes the AGV's spatial position data, transfer action signals, AGV status signals, and conveyor belt start / stop signals. Formal representation at the element level is as follows:

[0033] SEW j ={ETW id EW type ,CEW st ,SW}

[0034] EW type ={1,2,3}

[0035] CEW st ={-1,0,1,2}

[0036] Wherein, SEW j Let ETW represent the j-th logistics device. idThis indicates the unique serial number of the logistics equipment within the production workshop, EW type This indicates the specific equipment category of the logistics equipment, where 1 represents AGV, 2 represents conveyor belt, and 3 represents other logistics equipment; CEW st This indicates the current operating status of the logistics equipment, where -1 indicates an abnormal state, 0 indicates a closed state, 1 indicates a working state, and 2 indicates a waiting state; SW represents the location information set of the AGV and the conveyor belt.

[0037] Assembly equipment includes ion guns, dust removal guns, and crack spotlights used on the production floor, characterized as follows:

[0038] SEZ j ={ETZ id ,CEZ st ,FZ}

[0039] CEZ st ={-1,0,1}

[0040] Wherein, SEZ j Let ETZ represent the j-th assembly equipment. id This indicates the unique serial number of the equipment within the production workshop, CEZ. st This indicates the current operating status of the equipment, where -1 indicates an abnormal state, 0 indicates a closed state, and 1 indicates a occupied state; FZ represents the functional information set of the AGV and the conveyor belt;

[0041] The relationships between devices are defined in four ways: competitive relationship, no relationship, and merger relationship.

[0042] R SE (m,n)∈{-1,0,1}

[0043] R SE (m,n) represents the relationship between the m-th and n-th devices, where R SE (m,n) = -1 indicates that m and n are in a competitive relationship, R SE (m,n) = 0 indicates that there is no relationship between m and n, R SE (m,n) = 1 indicates that m and n are in a merge relationship;

[0044] Warehousing space, assembly space, and logistics space are all subsets of the space category set. They have similar basic characteristics, but their functional characteristics differ. The basic characteristics of a space include its status, function, and location information. The status includes idle status, occupied status, and abnormal status. The function is defined as three functions: warehousing, production, and transportation. The location information is determined by three-dimensional coordinates.

[0045] SEK j ={ETK id SKst EK type FK, SK

[0046] S st ={-1,0,1}

[0047] In the formula, SEK j Let SK represent the j-th independent space. st This indicates the current operating status of the device, where SK st =-1 indicates an abnormal state in the space, SK st =0 indicates that the space is free, SK st =1 indicates the usage status of the space; FK represents the function set of the specific space, and SK represents the location information.

[0048] Furthermore, step 1.2 is characterized as follows:

[0049] OT j ={PN j ,N,ST};

[0050] PN j = {(x,y),x∈N,y∈R};

[0051] N = {1, 2, ..., 6};

[0052] ST = {1, 2, 3, 4}.

[0053] In the formula OT j PN represents the j-th process, derived after uniformly numbering the processes on the production line; j It is a binary tuple representing the type and number of elements contained in the j-th element, N represents the type of element, 1-6 represent collection equipment, logistics equipment, assembly equipment, operator, storage space, and production space, respectively; ST represents the classification of the process to be implemented, 1 to 4 represent preparation, inspection, operation, and cleaning, respectively.

[0054] Furthermore, since a process is composed of multiple elements, the realization of its different functions must be achieved by combining different elements according to the process logic; the element combination logic can be formally expressed as follows:

[0055]

[0056] R P (p,q)∈[-1,0,1,2].

[0057] In the formula: R P This is a matrix representing the logical relationships between elements within a process, where m represents operators, n represents equipment, and R... P(p,q) = -1 indicates a competitive relationship between the operator and the corresponding device. This represents the relationship when two or more operators simultaneously select the same device. A competitive relationship means that a column of the matrix will necessarily contain two or more -1 values. In this case, a priority matrix needs to be created to classify priorities. R P (p,q) = 0 indicates that the operator has no connection with the equipment, meaning they will not use the equipment at this workstation. R P (p,q) = 1 indicates the operator's dependence on the equipment, meaning the operator has complete control over and uses of the equipment at their workstation. P (p,q) = 2 indicates that the operators cooperate in using the equipment.

[0058] Furthermore, the process flow modeling in step 1.3 is as follows:

[0059]

[0060]

[0061]

[0062] Where: MOF i This represents the i-th process flow in the workshop, where each process flow contains K types of processes, OT. i,k Let R be the k-th process in the process flow. OT This is a matrix representing the logical relationships between processes based on their functions. The matrix element × indicates that there is no relationship between workstations. → indicates a parallel relationship, and → indicates a serial relationship.

[0063] This invention combines previous research to construct the vertical hierarchy of the workshop entity from multiple scales, enriches the horizontal attributes and relationships of the entity from multiple dimensions, deconstructs the workshop production mode from the data dimension, and combines the workshop model with data evolution and prediction to establish a complete assembly workshop model, which facilitates the digital management of the workshop. Attached Figure Description

[0064] Figure 1 A hierarchical structure diagram of the assembly workshop digital twin model provided by this invention;

[0065] Figure 2 The multi-level data fusion framework diagram provided by this invention. Detailed implementation method:

[0066] A multi-scale model is a logical model that is constructed at four scale levels: elements, processes, production lines, and workshops. It reflects the attributes, structure, and interrelationships of various production elements in the assembly workshop from a micro perspective, enabling it to map to physical entities.

[0067] A multi-dimensional model builds upon a multi-scale model by modeling from four dimensions: geometry, physics, behavior, and rules. It employs a data-driven approach, aggregating the characteristics of each dimension of the model using data and fusing it according to multi-source information fusion theory to achieve the fusion of the multi-scale and multi-dimensional models.

[0068] The specific steps are as follows:

[0069] Step 1: Construction of Multi-Scale Digital Twin Model

[0070] The hierarchical construction of the virtual model requires a clear definition of its hierarchy. A multi-scale modeling approach is employed to establish a model set based on four levels: elements, processes, production lines, and workshops. Each level focuses on information and objects with a clear inheritance relationship, and its specific structure is as follows: Figure 1 The element scale is a secondary division of production factors such as people, machines, materials, methods, and environment within the workshop. Its characteristic is that each element is an independent object that performs a specific function, and there is no coupling or overlap between them. These elements form the basis of the three scales: process, production line, and workshop scale, participating in the construction of these three scales. Formal mathematical models are constructed to map physical entities.

[0071] The process scale is defined by combining various elements defined by the element scale with the specific production processes in physical space. Different processes are classified according to their technological and physical characteristics, and then a unified model is created for each type of process.

[0072] The production line scale is defined by combining various processes in the process scale with basic elements in the element scale as supplements. It is a full-element mapping of a single production line in physical space and a process flow that combines processes.

[0073] The workshop scale is mainly constructed by combining production lines, independent processes, and independent elements, and it is a mapping of the overall production process.

[0074] Step 1.1 Based on the characteristics of the assembly workshop, the element-scale model includes two main subclasses: entity classes and spatial classes. The entity class includes four elements: data acquisition equipment, logistics equipment, assembly equipment, and operators. The spatial class includes three elements: storage space, production space, and transportation space. Therefore, the formal representation is as follows:

[0075] P = O∪S

[0076] O={E c E l E a ,H}

[0077] S={S s ,S a ,S t}

[0078] Where P represents the set of all elements at the element scale, O represents the set of entity classes, S represents the set of spatial classes, and E... c E represents the set of data acquisition devices. l E represents a collection of logistics equipment. a H represents the set of assembly equipment, S represents the set of assembly personnel, and S represents the set of assembly equipment. s S represents the set of storage spaces. a S represents the assembly space set. t It represents a collection of transport spaces.

[0079] Because of the different types of elements, the key information affecting production efficiency also varies, and they need to be characterized separately according to their types. Based on the different types of elements mentioned above, the characterization is as follows:

[0080] The data acquisition equipment mainly includes video surveillance equipment, RFID data acquisition equipment, and heterogeneous sensors at the assembly site. The primary function of these devices is to monitor the real-time status of personnel and materials at the assembly site and collect their status information. For each device itself, its own status information needs to be considered, using three state variables—on, off, and abnormal—to reflect its own state. This is formally represented as follows:

[0081] SEC j ={ETC id EC type CEC st}

[0082] EC type ={1,2,3,4}

[0083] CEC st ={-1,0,1}

[0084] In the formula SEC j Let j be the j-th data collection device, ETC id This indicates the unique serial number of the data acquisition device within the production workshop, EC. type This indicates the specific equipment category of the data acquisition device, where 1 represents video surveillance equipment, 2 represents RFID data acquisition equipment, 3 represents heterogeneous sensors, and 4 represents other data acquisition devices; CEC st This indicates the current operating status of the data acquisition device, where -1 indicates an abnormal state, 0 indicates a shut-down state, and 1 indicates a startup state.

[0085] Logistics equipment mainly includes AGVs and conveyor belts on the production floor, and its primary function is to transfer and transport materials. Logistics equipment needs to transfer materials, semi-finished products, and finished products according to a certain rhythm, which is characterized by the following standards:

[0086] Cl ={C AGV C con C el}

[0087] Where C AGV Indicates AGV,C con Indicates a conveyor belt, C el This refers to other logistics equipment.

[0088] AGVs encompass translational and rotational movements on a plane, and also possess various transfer functions on their upper structure, such as using small conveyor belts to control the movement of goods. Conveyor belts facilitate the flow of assembled products, and specialized conveyor belts can be coupled with sensors to control the position of goods. Key data includes the AGV's spatial position, transfer action signals, AGV status signals, and conveyor belt start / stop signals. Due to the close interrelationship and high coupling between the operation of logistics equipment and other elements, many characteristics need to be determined at the process level in conjunction with other elements. Formal characterization at the element level is as follows:

[0089] SEW j ={ETW id EW type ,CEW st ,SW}

[0090] EW type ={1,2,3}

[0091] CEW st ={-1,0,1,2}

[0092] Wherein, SEW j Let ETW represent the j-th logistics device. id This indicates the unique serial number of the logistics equipment within the production workshop, EW type This indicates the specific equipment category of the logistics equipment, where 1 represents AGV, 2 represents conveyor belt, and 3 represents other logistics equipment; CEW st This indicates the current operating status of the logistics equipment, where -1 indicates an abnormal state, 0 indicates a closed state, 1 indicates a working state, and 2 indicates a waiting state; SW represents the location information set of the AGV and the conveyor belt.

[0093] Assembly equipment and operators are the most important elements in completing the entire workshop assembly process.

[0094] Assembly equipment mainly includes ion guns, dust removal guns, crack spotlights, etc., used on the production site. Different equipment has different functions. First, it is necessary to characterize the assembly equipment from several main directions, such as equipment coding, equipment status sets, and equipment function sets.

[0095] SEZ j ={ETZ id ,CEZ st ,FZ}

[0096] CEZ st ={-1,0,1}

[0097] Wherein, SEZ j Let ETZ represent the j-th assembly equipment. id This indicates the unique serial number of the equipment within the production workshop, CEZ. st This indicates the current operating status of the equipment, where -1 indicates an abnormal state, 0 indicates a closed state, and 1 indicates an occupied state; FZ represents the functional information set of the AGV and the conveyor belt.

[0098] Because the three types of equipment may have overlapping functions, a relationship matrix needs to be established to represent their interrelationships. The relationships between equipment are defined as follows: competitive relationship, no relationship, and merger relationship.

[0099] R SE (m,n)∈{-1,0,1}

[0100] R SE (m,n) represents the relationship between the m-th and n-th devices, where R SE (m,n) = -1 indicates that m and n are in a competitive relationship, R SE (m,n) = 0 indicates that there is no relationship between m and n, R SE (m,n)=1 indicates that m and n are in a merge relationship.

[0101] Warehousing space, assembly space, and logistics space are all subsets of the space category set. They share similar basic characteristics, differing only in their functional features. The basic characteristics of a space include its state, function, and location information. The state includes idle state, occupied state, and abnormal state. The function set management subdivides functions, which are defined in this invention as warehousing, production, and transportation. The location information is determined by three-dimensional coordinates. In conjunction with actual production, the location information in this invention is a fixed constant.

[0102] SEK j ={ETK id SK st EK type FK, SK

[0103] S st ={-1,0,1}

[0104] In the formula, SEK jLet SK represent the j-th independent space. st This indicates the current operating status of the device, where SK st =-1 indicates an abnormal state in the space, SK st =0 indicates that the space is free, SK st =1 indicates the usage status of the space; FK represents the function set of the specific space, and SK represents the location information.

[0105] Step 1.2 After completing the modeling of different types of equipment at the element scale, it is necessary to combine different types of elements at the process level. First, the characteristics of the process level itself need to be analyzed. Based on the production characteristics of the assembly workshop, the execution order of processes is distinguished into parallel processes and sequential processes. According to the actual situation of the factory, combined with several basic elements, the basic information at the process level should include the types and quantities of elements involved, the number of process steps implemented by each element type, and the classification of processes. Based on actual production, processes are divided into four categories: preparation, inspection, operation, and cleaning, represented as follows:

[0106] OT j ={PN j ,N,ST};

[0107] PN j = {(x,y),x∈N,y∈R};

[0108] N = {1, 2, ..., 6};

[0109] ST = {1, 2, 3, 4}.

[0110] In the formula OT j PN represents the j-th process, derived after uniformly numbering the processes on the production line; j It is a binary tuple representing the type and number of elements contained in the j-th element, N represents the type of element, 1-6 represent the collection equipment, logistics equipment, assembly equipment, operator, storage space, and production space, respectively; ST represents the classification of the process to be implemented, 1 to 4 represent the preparation class, detection class, operation class, and cleaning class, respectively.

[0111] Furthermore, since a process is composed of multiple elements, the realization of different functions must be achieved by combining different elements according to the process logic. Considering the actual production situation, the element combination logic is formally expressed as follows:

[0112]

[0113] R P (p,q)∈[-1,0,1,2].

[0114] In the formula: R PThis is a matrix representing the logical relationships between elements within a process, where m represents operators, n represents equipment, and R... P (p,q) = -1 indicates a competitive relationship between the operator and the corresponding device. This represents the relationship when two or more operators simultaneously select the same device. A competitive relationship means that a column of the matrix will necessarily contain two or more -1 values. In this case, a priority matrix needs to be created to classify priorities. R P (p,q) = 0 indicates that the operator has no connection with the equipment, meaning they will not use the equipment at this workstation. R P (p,q) = 1 indicates the operator's dependence on the equipment, meaning the operator has complete control over and uses of the equipment at their workstation. P (p,q)=2 indicates that operators cooperate in using the equipment. If this occurs in the matrix, then there will definitely be two or more such cases in that column.

[0115] Step 1.3 After completing the process-scale modeling, analyze the characteristics of the production line model. This requires focusing on the processing characteristics and technological flow of the production line, and then using the process model and its elements as the foundation for further modeling. Actual production lines are mainly composed of process flows, which are formed by various processes combined with elements. The process flow can be formally modeled as follows:

[0116]

[0117]

[0118]

[0119] Where: MOF i OT represents the i-th process flow (containing K types of processes) in the workshop. i,k Let R be the k-th process in the process flow. OT This is a matrix representing the logical relationships between processes based on their functions. The matrix element × indicates that there is no relationship between workstations. → indicates a parallel relationship, and → indicates a serial relationship.

[0120] Step 1.4 synthesizes the characteristics of elements, workstations, and production lines to propose the basic logical relationships at the workshop scale, forming a workshop logical structure centered on workshop structure and technological tasks. Workshop structure refers to the layout relationships of various production lines and workstations within the workshop; their spatial locations are also key data information. The operating time data of logistics equipment moving between production lines, workstations, and elements in different locations is the basis for corresponding scheduling. Technological tasks mainly refer to the overall assembly task performed in the workshop and its sub-tasks. These sub-tasks are aggregated through the dimensions of elements, workstations, and production lines to form the workshop model.

[0121] Step 2: At the model type level, a multi-dimensional modeling method is adopted. Modeling is performed from four levels: geometric, physical, behavioral, and rule-based. Model-based definition (MBD) technology is based on the geometric model of mechanical equipment, attaching process information, physical information, and numbering attribute information to the geometric model to achieve efficient model application. Drawing on the ideas of MBD technology, a digital twin modeling technology based on Bill of Process (BOP) is proposed. Using the geometric model as the foundation, information involving physical and rule models, such as process information and product physical information, is attached to the geometric model to achieve information encapsulation. For behavioral models reflecting system behavioral logic, a model-based twin mapping method is proposed. The formal description of virtual mapping is:

[0122] V = M + P + B + I;

[0123] P Δ B Δ I→M;

[0124] In the formula, V represents the virtual workshop model; M, P, B, and I represent the geometric model, physical model, behavioral model, and rule model, respectively; Δ represents the connection relationship; and → represents the attachment relationship.

[0125] In the actual modeling process, physical, behavioral, and rule models are all presented in the form of data or information in the virtual space (geometric models are essentially also data and information, but they have three-dimensional visual attributes and are classified separately here). Therefore, after the basic geometric model is constructed, a data model based on data node flow is constructed as the carrier of physical, behavioral, and rule models.

[0126] Step 2.1 Establishing data nodes based on the multi-scale model: Establish basic data nodes by using a multi-scale model based on elements and workstations, and determine data nodes that match different types of elements and workstations as well as data nodes with independent control functions.

[0127] Considering the simulation requirements for all elements, processes, and business aspects of workshop manufacturing, it is necessary to construct models at multiple scales, including elements, workstations, and production lines. Since elements themselves do not possess data acquisition capabilities, basic elements require the use of data acquisition devices at the element scale or their built-in acquisition mechanisms to collect on-site data. However, as elements, they require basic physical and behavioral information to support their operation within the model. The granularity of data nodes in the model needs to be determined based on the actual data requirements of the service functions, and this granularity corresponds to the granularity under multi-scale modeling.

[0128] Data nodes can be formally represented as:

[0129]

[0130] D i,j,k ={D p D b D r ,…}

[0131]

[0132] In the formula, OD i,j,k Represents a data node, where i, j, k represent its corresponding production line number, workstation number, and element type number, respectively; D i,j,k This represents the physical object data within a data node that does not change over time, where D p D represents physical property data. b D represents the behavioral information data reflected by this node. r This represents the rule information data for that node; Indicates at t N The node data information at different times, where E represents the energy dataset, S represents the velocity dataset, V represents the vibration dataset, D represents the displacement dataset, I represents the image dataset, and Q represents the voice dataset.

[0133] It should be noted that the representation of physical object data in the data node in the above formula is divided into three directions and includes physical information data, behavioral information data, and rule information data. Among them, physical information data mainly represents the dynamics and kinematic characteristics of the node's mapping in the physical world; behavioral data information is the information on the production behaviors that the node's physical world mapping can participate in; and rule data information is the necessary constraint information for connecting the physical world mappings of data nodes.

[0134] It should also be pointed out that data that changes over time is a type of situational data, which is composed of historical information in the database and real-time information received, and serves as the data foundation for the model to perform simulations and inferences.

[0135] Step 2.2 Construct a data model based on the data evolution process. Consider using the self-growth theory of trees to analogize the evolution process of data nodes and their information to the self-growth process of trees. Formalize the data evolution logic into a data model. Combine the nodes of the data model with the multi-scale model to form the following table in the analogy of the self-growth process of trees.

[0136] Table 1. An analogy between the evolution of workshop assembly data and the self-growth process of a tree.

[0137]

[0138] As can be seen from Table 1, the evolution of the overall data can be described as follows:

[0139] The production line can be likened to a "tree trunk," upon which all the "branches," "twigs," and "leaves" are built. Adjustments to the production line's function can be reflected in changes in the number and location of the branches; changes in data nodes are also reflected in changes in elements under different service conditions; and the growth of different "leaves" is reflected in the continuous increase of the data values ​​of each data node over time.

[0140] Considering the characteristics of multi-source heterogeneity and multiple spatiotemporal scales of manufacturing data in the physical world, and based on the evolution process described above, we introduce a matrix theory-based model for data association in the assembly workshop.

[0141] Constructing the assembly workshop coordinate matrix (CM) involves using the production line as the main line of association, then using production workstations as nodes on that line, and finally using the corresponding data to construct the coordinates of the remaining dimensions, forming a multi-dimensional matrix.

[0142] Step 2.3: Consider adopting the multi-source information fusion theory, and based on the general physical information fusion framework and multi-scale multi-dimensional digital twin model, establish a three-level data fusion mode, which is a three-layer fusion architecture from metadata fusion, feature-level fusion to decision-level fusion.

[0143] Metadata fusion first requires redundancy removal of dynamic data from each data node to eliminate noise in the original data and reduce congestion in the transmission channel. The data in the data node includes static attribute data and real-time status data of the process. Static attribute data serves as a prerequisite for real-time status data processing. An event matching mechanism is constructed in conjunction with the process-level status model. The event range that falls into is determined by the threshold range of the fused data, thus completing the fusion of metadata.

[0144] Feature-level fusion is formed on the basis of metadata fusion. It takes the model state set at the production line scale as the base point and treats it as a spatial object. Within the production line area, metadata fusion forms a fused dataset containing spatial information.

[0145] Decision-level fusion builds upon this by integrating digital twin data with the process characteristics of the workshop, enabling the data model to predict and control the behavior of the physical workshop based on workshop plans and dynamic task effects.

[0146] Three-level data fusion can correspond to the element-scale model, process-scale model and workshop-scale model of multi-scale models, that is, data fusion is carried out at three scale levels to realize the fusion of multi-dimensional models and multi-scale models.

Claims

1. A method for constructing a digital twin model for an assembly workshop, including the construction of a multi-scale model, the construction of a multi-dimensional model, and corresponding model fusion and model lightweighting methods; Step 1: Construction of a multi-scale digital twin model: Multi-scale refers to the four basic industrial scales of elements, processes, production lines, and workshops. The hierarchical modeling method is used to model from the smallest element scale to the largest scale. Step 1.1: The elements are divided into two main categories: entity class and space class. The entity class includes four elements: collection equipment, logistics equipment, assembly equipment, and operator. The space class includes three elements: storage space, production space, and transportation space. Analyze and establish a formal expression that includes the types, attributes, and interrelationships of elements. Step 1.2: Construct an element-based process mapping model, and divide the processes into four categories based on actual production: preparation, inspection, operation and cleaning. Different combinations of elements reflect different process functions and attributes. Step 1.3: Based on processes and elements, construct a formal representation of the production line; organize the processes using the process flow model, and then supplement them with different independent elements. Use different process organization forms to determine the organizational form of the production line, and determine the attributes and interrelationships of the production line. Step 1.4: Combine the three scale models of elements, processes, and production lines to construct an overall framework at the workshop scale, ensuring the consistency of internal logic and the independence of functions within the workshop; Step 2: Construction of a multi-dimensional digital twin model; Modeling is performed at four levels: geometric, physical, behavioral, and rule-based. Using the geometric model as a foundation, information from the physical, behavioral, and rule-based models, including process information and product physical information, is attached to the geometric model. The formal description of this virtual mapping is as follows: ； ； in, V represents the virtual workshop model; M, P, B, and I represent the geometric model, physical model, behavioral model, and rule model, respectively. Indicates a connection relationship; This indicates the attachment relationship; based on the different granularity models established in the multi-scale modeling stage, data nodes based on data of different granularity can be formed; the establishment of data nodes first requires determining the data and information to be covered by nodes at different levels, and then completing the establishment of node flows between different nodes, clarifying the composition logic and composition relationship between nodes; Step 2.1: Establish data nodes based on the multi-scale model. This mainly involves establishing basic data nodes based on element and process models, identifying data nodes that match different types of elements and processes, and data nodes with independent control functions. Considering the simulation requirements for all elements, processes, and businesses in workshop manufacturing, the model is constructed by combining elements, workstations, and production lines. In the model, the required granularity of data nodes is determined according to the data requirements of actual service functions, and this granularity corresponds to the granularity under multi-scale modeling. The data node is formally represented as: ； ； ； In the formula, Represents a data node, where These represent the corresponding production line number, workstation number, and element type number, respectively. This represents the physical object data within a data node that does not change over time. Representing physical property data, This indicates the behavioral information data reflected by this node. This represents the rule information data for that node; Indicates in The node data information at time, where E represents the energy dataset, S represents the velocity dataset, V represents the vibration dataset, D represents the displacement dataset, I represents the image dataset, and Q represents the voice dataset; The above formula represents the physical object data in the data node in three directions and includes physical information data, behavioral information data, and rule information data. Among them, physical information data mainly represents the dynamics and kinematic characteristics of the node's mapping in the physical world; behavioral data information is the information on the production behaviors that the node's physical world mapping can participate in; and rule data information is the necessary constraint information for connecting the physical world mappings of data nodes. Step 2.2: Based on the characteristics of the constructed data nodes, the evolution process of the data nodes is analogized to the growth process of a tree to determine the evolution logic of the data nodes; Step 2.3: Based on the theory of multi-source information fusion, a three-level data fusion model is established according to the general physical information fusion framework. Combined with the multi-scale model, a three-level data fusion model is formed, which realizes metadata fusion at the process level, feature-level data fusion at the production line level, and decision-level data fusion at the workshop level, thus completing the fusion of multi-scale and multi-dimensional models.

2. The method for constructing a digital twin model for an assembly workshop as described in claim 1, characterized in that, The specific method for step 1.1 is as follows: The data acquisition equipment includes video surveillance equipment at the assembly site, RFID data acquisition equipment, and heterogeneous sensors. The data acquisition equipment uses three state variables—on, off, and abnormal—to reflect its own state; the characteristics are as follows: ； ； ； In the formula This represents the j-th data acquisition device. This indicates the unique identifier of the data acquisition device within the production workshop. This indicates the specific equipment category of the data acquisition device, where 1 represents video surveillance equipment, 2 represents RFID data acquisition equipment, 3 represents heterogeneous sensors, and 4 represents other data acquisition devices. This indicates the current operating status of the data acquisition device, where -1 indicates an abnormal state, 0 indicates a shut-down state, and 1 indicates a startup state. Logistics equipment mainly includes AGVs and conveyor belts on the production site. Their primary function is to transfer and transport materials. Logistics equipment needs to transfer materials, semi-finished products, and finished products according to a certain rhythm, as shown below. ； in Indicates AGV, Indicates a conveyor belt. Indicates other logistics equipment; An AGV (Automated Guided Vehicle) encompasses translational and rotational movements on a plane, and also possesses various transfer functions on its upper surface. Its data includes the AGV's spatial position data, transfer action signals, AGV status signals, and conveyor belt start / stop signals. Formal representation at the element level is as follows: ； ； ； Where, in the formula This represents the j-th logistics device. This indicates the unique serial number of the logistics equipment within the production workshop. This indicates the specific equipment category of the logistics equipment, where 1 represents AGV, 2 represents conveyor belt, and 3 represents other logistics equipment; This indicates the current operating status of the logistics equipment, where -1 indicates an abnormal state, 0 indicates a shut-down state, 1 indicates a working state, and 2 indicates a waiting state. This represents the set of position information between the AGV and the conveyor belt; Assembly equipment includes ion guns, dust removal guns, and crack spotlights used on the production floor, characterized as follows: ； ； Where, in the formula This represents the j-th assembly equipment. This indicates the unique serial number of the equipment within the production workshop. This indicates the current operating status of the equipment, where -1 indicates an abnormal state, 0 indicates a closed state, and 1 indicates a occupied state. This represents the set of functional information for the AGV and the conveyor belt; The relationships between devices are defined in three ways: competitive relationship, no relationship, and merger relationship. ； Represents the relationship between the m-th and n-th types of devices, where This indicates that m and n are in a competitive relationship. This indicates that there is no relationship between m and n. This indicates a merging relationship between m and n; Warehousing space, assembly space, and logistics space are all subsets of the space category set. They have similar basic characteristics, but their functional characteristics differ. The basic characteristics of a space include its status, function, and location information. The status includes idle status, occupied status, and abnormal status. The function is defined as three functions: warehousing, production, and transportation. The location information is determined by three-dimensional coordinates. ； ； In the formula, where, Let j represent the j-th independent space. This indicates the current operating status of the device, where Indicates an abnormal state of space. Indicates the free state of space. Indicates the usage status of the space; Represents a set of functions in a specific space. Indicates location information.

3. The method for constructing a digital twin model for an assembly workshop as described in claim 1, characterized in that, The characterization of step 1.2 is as follows: ； ； ； ； In the formula Indicates the first Each process is derived from a standardized process numbering system on the production line. It is a tuple representing the first The types and quantities of elements contained therein. The elements are categorized as follows: 1-6 represent data collection equipment, logistics equipment, assembly equipment, operator, storage space, and production space, respectively. This indicates the classification of the processes performed, with 1 to 4 representing preparation, inspection, operation, and cleaning, respectively. Furthermore, since a process is composed of multiple elements, the realization of different functions must be achieved by combining different elements according to the process logic; the element combination logic can be formally expressed as follows: ； In the formula: This is a matrix representing the logical relationships between elements within a process. Indicates the operator. Indicates equipment, This indicates a competitive relationship between the operator and the corresponding device. It represents the relationship when two or more operators simultaneously select the same device. In a competitive relationship, there will inevitably be two or more -1 values ​​in a certain column of the matrix. In this case, a priority matrix needs to be established to divide the priorities. This indicates that the operator is not associated with the equipment and will not use it at this workstation. This indicates the operator's dependent relationship with the equipment, meaning the operator has complete ownership and use of the equipment at their workstation. This indicates that operators cooperate in using the equipment.

4. The method for constructing a digital twin model for an assembly workshop as described in claim 1, characterized in that, The process flow modeling in step 1.3 is as follows: ； ； In the formula: Indicates the first in the workshop There are K process flows, each containing K types of processes. For the k-th process in the process flow, This is a matrix representing the logical relationships between processes based on their functions, with matrix elements... This indicates that there is no relationship between the workstations. Indicates parallel relationship, This indicates a sequential relationship.

Images