Complex production scenarios includes large equipment scale, complex processes and sub-model constraints are highly coupling at each level. It is difficult in inheritance of similar equipment information, causing the low reusability of models in the reconfiguration process of twin platform. To address this problem, a component based rapid modeling method for digital twin is proposed, which can enable efficient reuse of model information. Mainly includes: (1) for the decoupling problem of complex models, a hierarchical division scheme and a component based twin model are designed. (2) a K-means clustering algorithm integrating subjective and objective weights to filter templates is proposed to address the efficiency and executability of information reuse. (3) for the inheritance problem of subjective information, an information inheritance process is designed to process and quantify the subjective information inheritance. Finally, the circuit breaker digital twin workshop system is used as an case to exame. The efficiency of dynamic, static and control twin model construction of the researched sub-model are improved by 44.74%, 48.96% and 72.55%. The actual development cycle of multiple projects can be reduced by about 33% compared to the expected cycle.
Research Article
Research on digital twin modeling method for complex production scenarios
https://doi.org/10.21203/rs.3.rs-2220974/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
digital twin
rapid modeling
information reuse
componentization
The construction of digital twin model for complex production scenarios is a complex process, which has complex production equipment operation logic and many constraint relationships between geometry and behavior among sub-models[1–4]. Meanwhile, with the digital transformation of manufacturing enterprises, the agility requirements for twin product development are increasing. These characteristics and the current situation bring great challenges to the rapid modeling of digital twin systems. To tackle the issue, we carry out research on digital twin modeling method for complex production scenarios.
The digital twin model in complex production scenarios consists of models of multiple scales and levels, with complex and diverse twin model information and highly coupled sub-models[5–10]. It brings great challenges to the repetitive model building and maintenance and makes it difficult to meet the market agility requirements. The complexity of the twin model is mainly reflected in: (1) a large number of types and quantities of physical entity object parts, and complex production equipment process logic. (2) The association between many parts and components of physical entities, which includes geometric-spatial configuration relationships and behavioral logic relationships, results in a high coupling between sub-models. When a large number of twin bodies are developed for customization, iterative design is tedious because of lack of a unified model to separate and integrate the attributes of different entities. (3) The lack of information reuse design for similar equipment models causes a large amount of information not to be reused, which further prolongs the construction cycle of the twin platform.
For the problem of rapid modeling of digital twin in complex production scenarios, existing research focuses on two aspects: on the one hand, it provides architecture-level reference for similar twin systems by designing a common modeling method and system framework. On the other hand, by building a flexible model attribute configuration mechanism, the information model is efficiently reconstructed to reduce the system development workload. Ref[11] constructs a model-based definition (MBD) including public information in different aspects of the model life cycle and establishes a generic information model in different phases of different models. Ref[12] proposes a generic system architecture for digital twin(DT) model construction and reveals the core technology for developing DT to provide a reference model for similar product-level DT implementations in the future. Ref[13] proposes a generic advanced information-physical system architecture for digital factories, which lays the foundation for smart factory management by virtualizing physical assets into cyberspace. Ref[14] proposes a generic architecture for digital twin visualization of flexible manufacturing systems, described Cyber-Physical(C-P) modeling of digital twins with multi-source heterogeneous information, and explores 3D visualization of human-computer interaction of digital twin scene information in the proposed architecture. The above researchs can effectively describe the functional structure of the production system, nevertheless, a mechanism to map the coupling relationship of sub-models is not constructed, and the hierarchical description of information data and the attribute description of application and twin objects are lack, which is difficult to apply to the actual deployment of twin systems.
Ref[15] combines CPS, DT modeling and other techniques to propose a system framework that provides guidance for rapid system configuration of DT-based CPPS. In the Ref[16], a digital twin model for reconfigurable manufacturing systems is built to address the need to create increasingly complex customizable products. A metamodel concept is proposed to set a set of parameters that determine the functional configuration of the system, and reconfiguration of the model is achieved by configuring the system parameters. In the Ref[17], for open architecture machine tools, the manufacturing system is reconfigured by flexibly configuring the individual modules in the machine tool to meet the process planning. The Ref[18] integrates four models, such as behavioral and collaborative information, and defines different twin models by configuring the parameters and mapping relationships between the four models. The above references reconfigure the model parameters or structure in order to reconfigure the original information model, however, the reconfiguration flexibility in this way is limited, and the final result of the reconfiguration depends on the configuration logic developed in advance, so the change of the structure and function of the model may lead to the failure of the established configuration scheme and the inability to construct a new twin through the configuration, abandoning the model information that could have been reused.
Component based design is one of the important theories of software science, widely used in the field of product design, and its effectiveness has been fully verified. Its basic idea is to decompose a complete system into multiple mutually independent components, which rely on interfaces to assemble each other to fulfill functional requirements, and each component is a reusable object, which greatly improves the efficiency of product development [19, 20]. For the problem of rapid modeling of digital twin in complex production scenarios, there are two main challenges: (1) how to realize the separation and integration of model attributes to achieve effective decoupling of model in the face of complex production scenario environments. (2) How to improve the flexibility of information reconfiguration and reuse under process customization conditions and improve the utilization of information.
Therefore, based on the idea of componentization, this paper proposes a modeling design scheme adapted to complex production scenarios to support information inheritance and reconfiguration, in further, an information reuse strategy to improve the efficiency of digital twin construction. Finally, using a small circuit breaker digital twin manufacturing workshop as a case, it is verified that the proposed method can significantly improve the reuse rate of model information and accelerate the twin model construction.
For the complex motion logic and geometric configuration, this paper designs a component based rapid modeling method for digital twin, and the scheme is designed as shown in Fig. 1, divided into three stages: hierarchical decomposition, model design and encapsulation, and reuse strategy design. By improving the modularity of the overall device model and the ability of information inheritance, the construction and reconfiguration of the twin model is accelerated.
(1) Hierarchical decomposition design: rely on the set hierarchical division rules to decompose the complex device model into several relatively independent sub-models to improve the modularity of the model.
(2) Model design and encapsulation: Separate and integrate the static, dynamic and control information of twin objects, and modularize the twin models. The information of each sub-model and its interrelationships are described by formal semantics such as association matrix and finite state machine, then the data structure of model information is instantiated by relies on model encapsulation design.
(3) Reuse strategy design: statistical device model, identify the sub-model used in high frequency, obtain samples of this sub-model in the scenario, quantify the sample characteristics, and perform the calculation of subjective and objective weights, and filter the samples by K-means clustering algorithm integrating mixed weights to build the template library. Design the information inheritance process to improve the efficiency of template information inheritance.
As shown in Fig. 2, production equipment of complex electromechanical product assembly line through a specific motion logic, sequentially install multiple parts on the workpiece of the vehicle, and finally complete the processing of the product. The assembly unit is the basic component unit of the production line, and is also the basic unit of the production line structure design. This paper takes the equipment unit as the research object for modeling method research.
3.1 Hierarchical decomposition design
Considering the actual process behavior characteristics of the assembly unit, the assembly unit is divided into the following modules by function: (1) clamping function (2) part conveying function (3) workpiece conveying function. Among them, the clamping function is divided into equipment trunk adjustment and end execution functions, the part conveying function is divided into part storage, ejection and flow functions, and the workpiece conveying function is divided into workpiece flow and positioning functions. The part conveying conveys the material from the storage point to the loading point by the single free motion of the mechanism, and the workpiece flow carries out the operation of flowing, positioning and turning of the processed parts by the single free motion of the mechanism. The clamping function mainly includes the following work: (1) Installing the parts from the loading table to the specified position of the workpiece. (2) Multi-degree-of-freedom spatial movement of the part during part convey. (3) Spatial multi-freedom spatial movement of the part during the workpiece flow. In the flow of Fig. 3-a, firstly, the single-degree-of-freedom moving parts are centered, and the geometric entities are merged into groups to form motion objects by relying on parent-child and contact relationships, and then the interrelated motion objects are merged into groups according to their corresponding functions, and each functional object is divided into groups in turn until the whole assembly unit. Figure 3-b shows the result of the hierarchical division of the assembly unit.
3.2 Model design and escapsulation
As shown in Fig. 4, the static, dynamic and control information of the model is extracted and encapsulated into components and fused into a unified model. The model can represent any twin body in the scene, and the model contains three types of components: entity component, dynamic component and control component. The entity component records the geometric and physical information of the model, the dynamic component records the motion logic information of the model, and the control component records the signal communication and parsing mechanism of the model. The entity component need to be assembled according to a certain spatial configuration, and based on the actual model information of the twin, it is determined whether the configuration of dynamic and control components is required. Each twin model has the potential to become a child object of another twin model, so component pointers are designed to map this composition relationship. The next step is to design the specific modeling details for each component of this model.
3.2.1 Static information model based on association matrix
Formula 1–1 is the static information model of all the objects at a certain level of the twin unit at a certain moment. X11, X22, X33, etc. represent the geometric and physical information of the object's sub-component entities, including geometric information such as size and shape, and physical properties such as material and mechanical performance parameters, etc. X12, X23, etc. represent the relationships between entities, including assembly, force and position relationships.
\(X=\left[ {\begin{array}{*{20}{c}} {{X_{11}}}&{{X_{12}}}&{...}&{{X_{1j}}}&{...}&{{X_{1{\text{n}}}}} \\ {{X_{21}}}&{{X_{22}}}&{...}&{{X_{2{\text{j}}}}}&{....}&{{X_{2{\text{n}}}}} \\ {...}&{...}&{...}&{...}&{...}&{...} \\ {{X_{{\text{i1}}}}}&{{X_{{\text{i2}}}}}&{...}&{{X_{{\text{ij}}}}}&{...}&{{X_{{\text{in}}}}} \\ {...}&{...}&{...}&{...}&{...}&{...} \\ {{X_{{\text{n2}}}}}&{{X_{{\text{n2}}}}}&{...}&{{X_{{\text{nj}}}}}&{...}&{{X_{{\text{nn}}}}} \end{array}} \right]\) (1–1)
Where:
X ——static information of a hierarchical object
X ij (i = j) ——static information of subcomponent entities
X ij (i ≠ j) ——interrelationship between subcomponent entities
Figure 5 shows the static model encapsulation of the clamping mechanism, whose geometric structure is shown in Fig. 5-a. According to the principle of hierarchical division, the division process is shown in Fig. 5-b, and it is merged into groups according to the relationship of parent-child and contact. Figure 5-c shows the encapsulated hierarchical structure, and Fig. 5-d shows the data encapsulation format of the model.
3.2.2 Behavioral information modeling based on finite state machine
Dynamic information describes the twin model response to external drives and disturbance effects, and Fig. 6 shows the state model of unit equipment behavior based on Moore-type finite state machine. Different assembly units have different specific processes and different sequences of parallel and serial execution. Considering that all assembly processes will eventually affect the machined workpiece, a unit process state tracking method based on the information of machined part properties is designed, and the main process is as follows.
(1) The information of the machined workpiece is abstracted into a ternary tuple(trans,rot,type), where trans and rot record the overall position posture information of the workpiece and type records the type of the mounted part.
(2) Design the ordered set of states of the process. As shown in Fig. 6, S0,S1,S2, etc. represent the individual process states of the unit, which form the finite state set of the unit; σ0,σ1, etc. represent the finite set of inputs, which are the trigger signals generated when the machined workpiece reaches the specified position; δ0,δ1, etc. are the state transfer functions; λ0,λ1, etc. are the state output sets of the machined workpiece, which represent the state of the workpiece after a process; ω0,ω1, etc. represent the output action of each state, which is specifically expressed as the specific mechanical movement of each process.
(3)Generation of twin units. First, the pose state of each part of the unit is initialized based on the state data of the real physical unit, and the parent-child relationship and physical constraints are added to each part model to give the same physical motion properties as the real part. Then the pose of the machined workpiece and the information of the installed parts are read in real time. When the workpiece arrives at the specified position, the process state transfer instruction σ0 is obtained by judging the information of the ternary. The unit migrates from S0 to S1 state through the transfer function δ0; at the same time, the process-driven part motion ω0 is executed, and the output λ0 is output after the state is executed. The processed workpiece is output to enter the next state of excitation, which is executed sequentially until all process states of a unit are completed.
Figure 7 shows the unit behavior hierarchy model, where each process activation will invoke the state of sub-levels layer by layer up to the action library. The behavior information of an object at a certain level includes the basic constituent actions and action invocation logic.
By analyzing the finite state machine model with behavioral information it is clear that any twin model containing dynamics can be decomposed into simple motion attributes of translation and rotation and a motion logic that coordinates these attributes. Therefore, the dynamic model information is divided into two modules of motion attributes and motion logic.
Taking the clamping function module as an example, the motion attribute escapsulation is shown in Fig. 8-a. Each serial number corresponds to the geometry of Fig. 5-a, containing rotation and translation library, with the following meanings: (1) Serial number 1 represents arm2. (2) Serial number 2 represents arm3. (3) Serial number 3 represents wrist4. (4) Serial number 4 represents actuator. Attitude adjustment and end execution correspond to the motion attribute escapsulation of two sub-functions, clamping is the entire motion attribute escapsulation of the mechanism. Figure 8-b represents the motion logic topology of the mechanism, where a,b,c,d,e,f represent the transfer points of each object state, the trigger of the stored action and the index of the specified action in the action library, and serial numbers 1,2,3,4 correspond to the serial numbers in Fig. 8-a, that is, represent the action sequence library of each subcomponent. The logic diagram runs as follows: (1) When the workpiece reaches the assembly point and the material arrives at the point to be assembled, point a opens and the clamping mechanism starts to operate. (2) b,c point open, function I-1 attitude adjustment begins to execute, after the execution, d point open, function I-2 end clamping begins to execute, in turn, after the execution of serial number 3 wrist action open e point, the execution of hand grip clamping parts. (3) continue to activate point b,c, while changing the index, call the corresponding animation. Execute iteratively until all specified actions are completed, and turn on point f to signal the completion of assembly to the downstream twin. Figure 8-c shows the dynamic data encapsulation format.
3.2.3 Interaction-oriented control information modeling
As shown in Fig. 9, the twin model receives control signals from two sources, a entity control signal from the workshop sensors and a simulation control signal between the upstream and downstream twins of the virtual scene. The physical control signal requires the construction of network communication protocols, and the information must also be transformed into information acceptable to the twin model. To ensure the normal communication and parsing of control signals, three principles are defined for the model to accept single, multiple signals and the case of accepting entity and simulation signals at the same time.
(1) Serial processing principle, the sequential relationship of adjacent twin information transfer, where the downstream twin cannot parse the signal when the upstream twin does not transfer the signal, and the upstream twin component cannot perform the signal transfer when the downstream twin component does not parse the signal.
(2) Parallel processing rules, when the same twin receives multiple signals, it needs to consider its own production behavior and system simulation rules to determine the order of signal processing.
(3) Entity priority principle, the entity object takes the initiative to send control signals to the twin object, and the entity control signal resolution priority is higher than the twin simulation signal. When the twin accepts the entity signal, it needs to mask the simulation signal, parse and execute the operation and then enter the accept signal wait state, if only the simulation signal is executed, execute the serial and parallel rules, parse and execute and then enter the signal wait state.
As shown in Fig. 10, the control model escapsulation of the clamping function module, for example, consists of two modules, external signal operation and internal signal operation. Separate the internal and external operations and constructe the signal resolution model. At the same time, the dynamic model transfer point is bridged, and its output signal is used as the control signal of the dynamic model. As shown in Fig. 10-a, the model needs to complete the configuration of the protocol information first, and its workflow mainly contains the following two points: (1) The external signal directly controls the transfer point of the twin entity through the module transformation, while the internal signal operation is suspended. (2) When the external signal stops transmitting, the internal signal operation starts to execute and the control point is operated according to the predefined simulation logic. Figure 10-b shows the encapsulated format of the control information.
4.1 Template library design strategy
The encapsulation of twin entities needs to be based on sorting out the specific static and dynamic and control information, and using componentization reasonably. At the beginning of twin production line construction, typical objects are selected to be encapsulated into templates for the sub-models with high frequency of production line construction.The corresponding template library is formed. later when new twins of the same type are constructed, the information of these templates can be inherited to speed up the twin construction process. At the same time, while new twins are generated continuously, new twins are screened to enrich the original template library, which is described in this paper for the special sub-model objects,that is the clamping function module model.
The difficulty in screening suitable templates lies in two aspects: (1) the quantitative description of twin features. (2) The design of a suitable clustering algorithm. To solve the above problems, a feature extraction method based on dynamic logic topology map is proposed. It is known from the behavioral logic and geometric information of the assembly line that the more similar the physical information of the objects with similar motion logic. Extracting the number of serial and parallel serial numbers of each functional layer of the topology can refine the key information of the topology diagram, and supplement the key static attributes on this basis, as shown in Fig. 11-a. Figure 11-b shows the K-means clustering algorithm integrating subjective and objective weights, which mathematizes the thinking process of decision, making through the organic combination of qualitative and quantitative. it makes the subjective judgment more reliable. The main process is as follows.
(1) For a certain type mechinism, the corresponding feature attributes are extracted to form a (m,n) two-dimensional sample matrix, with m being the number of samples and n being the feature dimensions.
(2) Based on the relationship between the extracted attributes, the judgment matrix is constructed by comparing the importance of two-two attributes with expert experience and assigning values in the range of 1 ~ 9. After passing the consistency test, the subjective weights are calculated based on the feature vector corresponding to the maximum eigenvalue.
(3) By standardizing the sample matrix, the objective weights are calculated using the information quantity, and the final mixed weights combining the subjective and objective weights are calculated
(4) Combining the mixed weights, the nearest neighbor calculation formula in the K-means clustering process is derived, and relying on the K-means iterative process, the templates are screened in order to form a template library. Retrieving templates according to Eqs. 1–4, when the number of new twins reaches a certain threshold, the clustering center of K-means should be increased to increase the number of templates in the template library of that type.
where：
d ——Euclidean distance with mixed weight
x ——Template feature properties
y ——Characteristic attribute of twin model to be built
n ——Twin characteristic dimension
ω ——Mixed weight
4.2 Information inheritance process design
According to the dynamic logic topology diagram, the inheritance process of the twin model is designed, as shown in Fig. 12. Firstly, the inheritance of dynamic model information is carried out according to the difference between the dynamic logic topology diagram of the template and the twin model to be built to complete the reconfiguration and reuse of the dynamic component, and then the static model is adjusted according to the change of node structure in the topology diagram and also combined with the geometric information of the entity model to complete the reconfiguration of the entity component. Finally, the reconstruction of control information is carried out according to the structure of transfer points adjusted by dynamic components, and debugging is carried out after the information inheritance is completed, and iterative model adjustment is carried out again if it fails, and the new twin is finally output.
The circuit breaker digital twin workshop system is used as an example for verification. Each production line mainly includes the assembly line for assembling parts and the detection line for testing product functions, and contains a total of 24 manufacturing processes such as single-stage riveting, automatic assembly and sweeping pad printing, etc. Figure 13-a shows the process flow diagram of the circuit breaker production line, in which all six production lines have a total of 47415 parts models, 4860 assembly actions and 4548 detection actions, with a large variety and number of production equipment and a complex action The logic is complex.
The project team developed the corresponding twin system for digital manufacturing of small circuit breakers in the early stage. The experimental software environment is Microsoft Visual Studio 2015 and Unity3D 5.5.2f1 development platform, and the hardware environment is Intel(R) Core(TM) i7-11800H @ 2.3GHz processor, NVIDIA GeForce RTX3060 display adapter.
The general scenario of the system and some of the functional modules and fault modules are shown in Fig. 13-b and 13-c; the general scenario of the system includes the functions of production simulation, equipment status monitoring and production data display to realize the model-driven and optimized simulation of virtual equipment on the one hand; on the other hand, it realizes the operation and maintenance services such as health management, fault prediction and result visualization through data extraction and mining.
Based on the proposed decomposition flowchart, the twin objects are encapsulated in a static model, and the entities are merged and grouped by parts contact and mechanism functions to divide the hierarchy as shown in Fig. 14.
The motion sequences of each motion object of the clamping mechanism are grouped and the corresponding motion sequence libraries of the motion objects are constructed. Meanwhile, the mapping of the motion logic topology diagram of the clamping module is completed by the state machine, as shown in Fig. 18-a,18-b and 18-c.
`As shown in Fig. 16, the corresponding C# code is written to instantiate the control information model data package of the clamping mechanism, including network configuration, signal acceptance and parsing, control point output, basic sensor information and entity and twin connection configuration.
Figure 17 shows the information inheritance process of the clamping module, and the specific details of the inheritance process of each information are: (1) static information: region 1 of Fig. 17-a is the object hierarchy, region 2 is the 3D entity rendering, and region 3 is responsible for adjusting the grid and spatial configuration of the parts, for example, replacing the end execution parts of the clamping module, replacing the corresponding grid in region 3, and adjusting the control position to complete the information inheritance operation. (2) Dynamic information: By replacing the action sequence in the state machine of the original object and adjusting the structure of the state machine at the same time, the inheritance of dynamic information is completed. (3) Control information: add or delete the original transfer points, and adjust the corresponding parsing statements, modify the rest of the object property information, and complete the inheritance of control information.
By analyzing the original twin production line, a total of 40 clamped twins were screened, and characteristic attributes of some of their are shown in Table 1. Based on the calculation method of subjective and objective weights, the final mixture weights of each attribute were derived as shown in Table 2. On this basis, K-means clustering is performed, as shown in Fig. 18 when the number of clustering centroids is 4, the average distance of each sample point to the center of the clustering starts to decrease at a decreasing rate when the number of clustering points is increased again, i.e. The centroid is 4 as the number of cost-performance optimal centroids, so the total number of template components of the initial clamping mechanism is 4.
|
Serial number |
Function I-1 serial number |
Function I-1 parallel number |
Function I-2 serial number |
Function I-2 parallel number |
Rotating pair |
Moving pair |
|---|---|---|---|---|---|---|
|
1 |
2 |
0 |
1 |
2 |
1 |
5 |
|
2 |
1 |
0 |
2 |
0 |
0 |
4 |
|
3 |
1 |
0 |
1 |
2 |
0 |
6 |
|
4 |
1 |
0 |
1 |
4 |
0 |
10 |
|
Serial number |
Function I-1 serial number |
Function I-1 parallel number |
Function I-2 serial number |
Function I-2 parallel number |
Rotating pair |
Moving pair |
|---|---|---|---|---|---|---|
|
Objective weight |
0.111131 |
0.435235 |
0.048439 |
0.136733 |
0.196405 |
0.072058 |
|
Subjective weight |
0.091642 |
0.222239 |
0.169122 |
0.404098 |
0.046356 |
0.066543 |
|
Blend weights |
0.055272 |
0.524957 |
0.044461 |
0.299874 |
0.049412 |
0.026023 |
Based on the proposed information inheritance process, the number of operations of dynamic model, static model and control model are compared respectively. As shown in Table 3, in the comparative tests of multiple sets of clamped module twin builds with different number of units, it is found that the proposed componentized rapid method has a significant improvement in the build efficiency compared with the traditional build engineering. The dynamic, static and control construction efficiencies are improved by 44.74%, 48.96% and 72,55%, respectively, in the 4-unit tests such as single-stage nail-piercing. Clamping module as the logical core of the twin unit, the rapid construction of this module is significant for the overall unit construction, which can effectively improve the efficiency of the entire unit model build. As Fig. 19 shows the actual R&D case project, the expected project development cycle is compared with the actual cycle, project one is shortened from 48 days to 33 days, project two from 28 days to 18 days, and project three from 33 days to 22 days.
Table 3 comparison between traditional building and component building in this paper (clamping module)
|
Unit |
Dynamic model operation num times |
Static model operation |
Control model operation |
||||||||
|
name |
Traditi-on |
Compo-nent |
Optimiza-tion rate |
Traditi-on |
Compo-nent |
Optimiza-tion rate |
Traditi-on |
Compo-nent |
Optimiza-tion rate |
||
|
Single stage nailing |
28 |
16 |
42.86% |
31 |
13 |
58.06% |
100 |
30 |
70.00% |
||
|
Single stage nailing Automatic magnetization |
54 |
32 |
40.74% |
62 |
30 |
51.61% |
234 |
62 |
73.50% |
||
|
Single stage nailing Automatic magnetization Yoke assembly |
67 |
38 |
43.28% |
84 |
44 |
47.62% |
312 |
98 |
68.59% |
||
|
Single stage nailing Automatic magnetization Yoke assembly Multistage riveting |
76 |
42 |
44.74% |
96 |
49 |
48.96% |
408 |
112 |
72.55% |
||
Digital twin is a prominent smart manufacturing technology that shows high-value application potential in complex production scenarios. In order to shorten the modeling cycle of digital twins, a rapid component based digital twin modeling method is proposed to provide an accurate and efficient construction method for twin modeling. The method improves the overall model modularity through hierarchical decomposition design, and designs a component based twin model to separate and integrate the attributes of the twin model to achieve an effective decoupling of the model. It also designs an efficient strategy for information reuse by considering the frequency of use of production equipment sub-models and the characteristics of twin modeling information inheritance. Using the circuit breaker digital twin workshop system for multi-unit and multi-scene comparison experiments, in the multi-unit comparison experiments, compared with the traditional one-time construction method, the optimization results obtained by this method have better results in the loading and clamping model building efficiency, the dynamic model building efficiency improved by 40.74%-44.74%, the static model improved by 47.62%-58.06%, the control model improved The rapid modeling method proposed in this paper can effectively shorten the R&D cycle by 31.25%-35.71% and improve the market responsiveness in the process of multi-scene actual project development.
Acknowledgements The authors would like to express sincere gratitude to the anonymous reviewers for the invaluable comments and suggestions that have improved the quality of the paper.
Funding This research is supported by Zhejiang Province Key Research and Development Program Project, 2021C01046.
Conflict of interest The authors declare no competing interests.
Availability of data and materia The datasets used or analysed during the current study are available from the corresponding author on reasonable request.
Code availability Not applicable.
Ethics approval This chapter does not contain any studies with human participants or animals performed by any of the authors. Authors acknowledge the work is original and all the relevant text from the literature has been properly cited.
Consent to participate Authors agree to the authorship order.
Consent to publication All authors have read and agreed to the published version of the manuscript.
Author Contribution All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by GaoJujian, YangYanfang and ShuLiang. The first draft of the manuscript was written by GaoJujian and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Corresponding author Correspondence to GaoJujian.
- Tao F, Zhang M, Cheng J et al (2017) Digital twin workshop: a new paradigm for future workshop. Computer Integrated Manufacturing Systems 23:1–9. https://doi.org/10.13196/j.cims.2017.01.001.
- Tao F, Qi QL, Wang LH et al (2019) Digital twins and cyber–physical systems toward smart manufacturing and industry 4.0: correlation and comparison. Engineering 5(4):653–661. https://doi.org/10.1016/j.eng.2019.01.014.
- Guo H, Zhu Y, Zhang Y, et al (2021) A digital twin-based layout optimization method for discrete manufacturing workshop. The International Journal of Advanced Manufacturing Technology 112(5):1307-1318. https://doi.org/10.1007/s00170-020-06568-0.
- Xia L, Lu J, Zhang H, et al (2022) Construction and application of smart factory digital twin system based on DTME. The International Journal of Advanced Manufacturing Technology 120(5):4159-4178. https://doi.org/10.1007/s00170-022-08971-1.
- Zhuang CB, Liu JH, Xiong H (2018) Digital twin-based smart production management and control framework for the complex product assembly shop-floor. The International Journal of Advanced Manufacturing Technology 96(1):1149-1163. https://doi.org/10.1007/s00170-018-1617-6.
- Elisa N, Luca F, Marco M(2017) A review of the roles of digital twin in CPS-based production systems. Procedia Manufacturing 11:939-948. https://doi.org/10.1016/j.promfg.2017.07.198.
- Bao QW, Zhao G, Yu Y, et al (2021) Product Information Units Modeling Oriented to Digital Twin. IOP Conference Series: Earth and Environmental Science 726(1): 012014. https://doi.org/10.1088/1755-1315/726/1/012014.
- Qian WW, Guo Y, Cui K, et al (2021) Multidimensional data modeling and model validation for digital twin workshop. Journal of Computing and Information Science in Engineering 21(3):031005. https://doi.org/10.1115/1.4049634.
- Leng JW, Zhou M, Xiao YX, et al (2021) Digital twins-based remote semi-physical commissioning of flow-type smart manufacturing systems. Journal of Cleaner Production 306: 127278. https://doi.org/10.1016/j.jclepro.2021.127278.
- Biesinger F, Weyrich M (2019) The facets of digital twins in production and the automotive industry. 2019 23rd International Conference on Mechatronics Technology (ICMT) 2019:1-6. https://doi.org/10.1109/ICMECT.2019.8932101.
- Ruemler SP, Zimmerman KE, Hartman NW, et al (2016) Promoting model-based definition to establish a complete product definition. J Manuf Sci Eng. 139(5):051008. https://doi.org/10.1115/1.4034625.
- Zheng P, Sivabalan AS (2020) A generic tri-model-based approach for product-level digital twin development in a smart manufacturing environment. Robotics Comput Integr Manuf. 64:101958. https://doi.org/10.1016/j.rcim.2020.101958.
- Lee J, Bagheri B, Kao HA (2015) A cyber-physical systems architecture for industry 4.0-based manufacturing systems. Manufacturing Letters. 3:18-23. https://doi.org/10.1016/j.mfglet.2014.12.001.
- Fan YP, Yang JZ, Chen JH, et al (2021) A digital-twin visualized architecture for Flexible Manufacturing System. Journal of Manufacturing Systems. 60:176-201. https://doi.org/10.1016/j.jmsy.2021.05.010.
- Liu C, Jiang PY, Jiang WL (2020) Web-based digital twin modeling and remote control of cyber-physical production systems. Robotics Comput Integr Manuf. 64:101956. https://doi.org/10.1016/j.rcim.2020.101956.
- Kombaya Touckia J, Hamani N, Kermad L (2022) Digital twin framework for reconfigurable manufacturing systems (RMSs): design and simulation. The International Journal of Advanced Manufacturing Technology. 120(7): 5431-5450. https://doi.org/10.1007/s00170-022-09118-y.
- Leng JW, Liu Q, Ye S, et al (2020) Digital twin-driven rapid reconfiguration of the automated manufacturing system via an open architecture model. Robotics and Computer-Integrated Manufacturing. 63: 101895. https://doi.org/10.1016/j. rcim.2019.101895.
- Bao JS, Guo DS, Li J, et al (2019) The modelling and operations for the digital twin in the context of manufacturing. Enterprise Information Systems. 13(4): 534-556. https://doi.org/10.1080/17517575.2018.1526324.
- Bock C (2019) Componentization in the systems modeling language. Systems Engineering. 17(4): 392-406. https://doi.org/10.1002/sys.21276.
- Zhang J (2012) A Role-Based Feature Model Componentization Method. Electrical, Information Engineering and Mechatronics 2011. 2012:255-267. https://doi.org/10.1007/978-1-4471-2467-2_30.