Cloud manufacturing system digital twinning migration modeling method considering preference

Abstract

The invention discloses a cloud manufacturing system digital twin migration modeling method considering preference, and the method comprises the steps: constructing a knowledge graph of a cloud manufacturing system based on the production relation of the cloud manufacturing system; constructing a digital twin model set in a cloud model library of the cloud manufacturing system; the digital twin model comprises a geometric model, a behavior model, a logic model and a performance model; based on the characteristic that the cloud manufacturing digital twinning model can be migrated, constructing a demand set of digital twinning modeling of a cloud manufacturing service demand side; according to an attention mechanism, an interaction and preference relationship between a modeling demand and a to-be-selected model is captured, a recommendation algorithm is constructed, and an optimal model is screened out from a digital twin model library for migration. According to the migration modeling method considering the preference, an efficient and accurate modeling method is provided for accurate application of digital twinning of the cloud manufacturing system, and the migration modeling method has important value for improvement of intelligence of the cloud manufacturing system.

Description

Technical Field

[0001] The present invention relates to a digital twin migration modeling method, and in particular to a preference-based cloud manufacturing system digital twin migration modeling method. Background Art

[0002] With the rise of intelligent manufacturing, new technical means and tools such as the Internet of Things, big data and artificial intelligence continue to emerge. These new technologies and tools help the transformation and upgrading of traditional manufacturing, thereby promoting the high-quality development of the manufacturing industry. Cloud manufacturing makes full use of network technology and new generation information technology to achieve cooperation in product design, manufacturing, management, etc. within and across supply chains. Through the Internet, companies in the entire supply chain can share manufacturing resources and achieve large-scale rapid response collaborative manufacturing, thereby shortening the production cycle, quickly responding to customer needs, and improving production flexibility.

[0003] The introduction of digital twin technology has greatly changed the operation mode of cloud manufacturing. The key to the application of digital twin technology in cloud manufacturing lies in the construction of digital twin models. The construction of digital twin models is to build a multi-dimensional, multi-scale, full-factor, high-fidelity physical space twin model in the virtual space through the comprehensive perception and integration of physical space information.

[0004] In the rapid evolution of today's manufacturing industry, manufacturing companies are facing increasingly complex market demands and production challenges. The order-based mixed-line production model has become an important means to improve production efficiency and meet personalized customer needs. This model requires the manufacturing system to be able to flexibly respond to diverse product demands while maintaining high efficiency and low cost. However, traditional manufacturing systems often find it difficult to adapt to this rapidly changing production environment, and a new solution is needed to improve production flexibility and responsiveness. To this end, the industry has proposed a cloud manufacturing model, and the cloud manufacturing model based on digital twins can predict and analyze the future through the support of digital twin systems, and provide optimization solutions. However, it takes a lot of modeling time to build a complete digital twin model that maps the virtual and real entities of manufacturing resources from scratch.

[0005] In this context, traditional digital twin modeling methods often fail to meet the needs of these advanced manufacturing models. Traditional methods may lack sufficient flexibility to adapt to rapidly changing production needs, or fail to effectively handle the complexity and dynamics of production. Therefore, cloud manufacturing platforms are in urgent need of an efficient digital twin modeling method that can support cloud manufacturing. Summary of the invention

[0006] Purpose of the invention: In response to the problems existing in the above-mentioned technical background, the present invention provides a digital twin migration modeling method for a cloud manufacturing system that takes preferences into consideration, which can migrate knowledge between different production tasks and manufacturing resources, improve the adaptability and generalization ability of the model; and can reflect the status of manufacturing resource entities in the cloud manufacturing system in real time, predict future production conditions, and provide support for production decisions.

[0007] Invention content: A digital twin migration modeling method for a cloud manufacturing system considering preferences according to the present invention comprises the following steps:

[0008] (1) Based on the production relations of the cloud manufacturing system, a knowledge graph of the cloud manufacturing system is constructed;

[0009] (2) constructing a set of digital twin models in a cloud model library of a cloud manufacturing system; the digital twin model includes a geometric model, a behavioral model, a logical model, and a performance model;

[0010] (3) Based on the transferability of cloud manufacturing digital twin models, a set of requirements for digital twin modeling of cloud manufacturing service demanders is constructed;

[0011] (4) Based on the attention mechanism, the interaction and preference relationship between modeling requirements and the models to be selected is captured, and a recommendation algorithm is constructed to select the optimal model from the digital twin model library for migration.

[0012] Furthermore, the implementation process of step (1) is as follows:

[0013] The production process of the cloud manufacturing system is analyzed, and the production activities of the cloud manufacturing system are described as the entire process in which the product is processed and assembled on different production units according to the production schedule until it is finally completed; the key elements of the cloud manufacturing system production process include orders, products, processes and machines; the production unit represents one or more manufacturing resource entities abstracted from each cloud manufacturing service provider;

[0014] The knowledge graph of the cloud manufacturing system describes the real-time manufacturing status of the cloud manufacturing system with nodes and connecting edges. Nodes represent entities in the cloud manufacturing system, and connecting edges represent the relationship between nodes. The formula of the knowledge graph G is as follows:

[0015] G=(V,E)

[0016] V=(M,O,Pd,Pc)

[0017] E=(MO,MPd,MPc,OPd,…,PdPc)

[0018] Among them, V represents the node set, E represents the connection edge set, M is the production unit, O is the order, Pd represents the product, and Pc represents the process; MO is the connection between the machine and the order, including two states: the machine is processing the order and the machine has not processed the order; MPd is the connection between the machine and the product, including two states: the machine is processing the product and the machine has not processed the product; it is determined by the position of the product in the actual production process; MPc is the connection between the machine and the process, indicating whether the machine executes a specific process; OPd indicates that the order consists of products, and its connection can be represented by the processing progress of the products in the order; PdPc is the connection between the product and the process, indicating whether the product is executing the process.

[0019] Furthermore, the geometric model in step (2) is:

[0020] DTM GM =(ME,GP)

[0021] Among them, ME represents the external model and GP represents the geographical location.

[0022] Furthermore, the behavior model in step (2) represents the real-time response and behavior of the manufacturing resource entity under the influence of the external environment and the internal operation mechanism, and is formed by coupling multiple manufacturing resource entity sub-behavior models, which are:

[0023]

[0024] Among them, M n Indicates the number of manufacturing resource entities, DTM EU The sub-behavior model of each manufacturing resource entity is represented by the following formula:

[0025]

[0026] Where X represents the input event set, Y represents the output event set, DI represents the input data set, DO represents the output data set, D represents the coupling member name set, {F i} represents a sub-function set, EIC represents an external input coupling function set, DIC represents an input data coupling function set, IEC represents an internal event coupling function set, IDC represents an internal data coupling function set, EOC represents an output event coupling function set, and DOC represents an output data coupling relationship set.

[0027] Furthermore, the logic model in step (2) represents the internal logic of the actual production process and operation of the mapping manufacturing resource entity, which is:

[0028] DTM LM =(DTM GM ,DTM BM ,DTMOR )

[0029] Among them, DTM OR Represents a set of operating rules, including behavioral information transmission rules between different digital twin logic models, information interaction rules between models, synchronous advancement rules between superior and subordinate models, and mutual exclusion priority rules between models of the same level.

[0030] Furthermore, the performance model in step (2) includes a prediction or decision model for performance indicators, which is expressed as:

[0031] DTM PM =(DTM MM ,DTM DDM )

[0032] Among them, DTM MM Representation mechanism model, DTM DDM Represents a data-driven model;

[0033] The jth digital twin model in the cloud model library is represented as:

[0034]

[0035] in, Represents the jth digital twin model in the cloud model library; They represent the geometric model, behavioral model, logical model and performance model of the j-th digital twin model respectively;

[0036] The digital twin model set is recorded as:

[0037]

[0038] Furthermore, the implementation process of step (3) is as follows:

[0039] The requirements for building the i-th digital twin model are expressed as:

[0040]

[0041] in, Indicates the construction requirements of the i-th digital twin model; They respectively represent the geometric model requirements, behavioral model requirements, logical model requirements, and performance model requirements in the i-th digital twin model construction requirements;

[0042] The demand set for building the digital twin model of cloud manufacturing service demanders is recorded as:

[0043]

[0044] Furthermore, in step (4), the root captures the interaction and preference relationship between the modeling requirements and the model to be selected according to the attention mechanism, and the implementation process is as follows:

[0045] The preference of the i-th build requirement for the j-th candidate model is expressed as:

[0046]

[0047] in, They represent preferences for geometry, behavior, logic, and performance, respectively. The symbols in parentheses represent the importance of the same group of models, that is, the preference value. The preference value is not less than 0, and x gm +x bm +x lm +x pm =1.

[0048] Furthermore, the implementation process of constructing the recommendation algorithm in step (4) is as follows:

[0049] The interaction matrix between the model construction requirements and the digital twin model is recorded as Given an interaction matrix Under the conditions of and knowledge graph G, the goal of migration modeling is to predict whether the model in the digital twin model library will be selected according to the needs of model construction. The prediction formula is:

[0050]

[0051] in, Indicates the requirements built according to the model cm de Select vm pr The probability of performing migration modeling, ω is the parameter to be trained for the prediction model.

[0052] Furthermore, the process of selecting the optimal model from the digital twin model library for migration in step (4) is as follows:

[0053] When new modeling requirements arise, the recommendation algorithm f predict (.) Push out the optimal digital twin model from the cloud model library for migration modeling; Denoted as the dot product of the model building requirement representation and the candidate model representation Represents the dot product of the vector, constructing the interaction between the label of the candidate model and the actual migration, that is, when in cm,vm =1, indicating that according to cm de Select VM pr Migrate in cm,vm =0, no migration.

[0054] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention utilizes transfer learning technology to transfer knowledge between different production tasks and manufacturing resources, thereby improving the adaptability and generalization ability of the model; the present invention recommends a suitable model in the model library through a preference algorithm to match it, and after fine-tuning, quickly constructs a digital twin model, which can reflect the status of manufacturing resource entities in the cloud manufacturing system in real time, predict future production conditions, and provide support for production decisions; it provides a powerful tool for the modern manufacturing industry to cope with the challenges brought by advanced manufacturing models such as network collaborative manufacturing and cloud manufacturing, and help enterprises improve production efficiency, reduce costs, and enhance market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the cloud manufacturing digital twin migration modeling method proposed in the present invention;

[0056] Figure 2 It is a flow chart of geometry, behavior, logic and performance modeling in the cloud manufacturing model library proposed by the present invention;

[0057] Figure 3 It is a flow chart of the training method of the recommendation algorithm proposed in the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described below in conjunction with the accompanying drawings.

[0059] like Figure 1 As shown, the present invention provides a method for modeling digital twin migration of a network system manufacturing system considering preferences, which specifically includes the following steps:

[0060] Step S1: Based on the production relationship of the cloud manufacturing system, a knowledge graph of the cloud manufacturing system is constructed, which includes two processes: analyzing the production process of the cloud manufacturing system and constructing the knowledge graph of the cloud manufacturing system.

[0061] The cloud manufacturing system is a system based on the concept of "manufacturing as a service". It integrates information technology, manufacturing technology, Internet of Things technology, etc., and provides high value-added, low-cost manufacturing services through cloud computing platforms. Its core concepts include using the network and cloud manufacturing platform to organize manufacturing resources according to user needs and provide various on-demand manufacturing services. The cloud manufacturing system consists of three parties: the cloud manufacturing platform, the cloud manufacturing service provider, and the cloud manufacturing service demander. The cloud manufacturing service provider virtualizes and encapsulates all kinds of manufacturing resources it owns into manufacturing capabilities in the form of cloud services and stores them in the resource pool of the cloud service platform. The cloud manufacturing service demander selects manufacturing capabilities that meet the requirements in terms of function, performance, price, time and service quality from the resource pool of the cloud manufacturing platform according to actual needs, combines them according to business processes, meets the manufacturing task requirements and organizes production.

[0062] The production relationship of the cloud manufacturing system is reflected in the following aspects: using network technology to coordinate the planning tasks and manufacturing links of different enterprises in different geographical locations, and to achieve the optimal allocation of dynamic resources and efficient coordination of the manufacturing process. The knowledge graph of the cloud manufacturing system represents the conversion relationship between the graph model data and the production data established by the system under the production state at a certain time.

[0063] Through the analysis of the production process of cloud manufacturing system, the production activities of cloud manufacturing system are described as the whole process of processing and assembling products in different production units according to the production schedule until they are finally completed. The key elements of the production process of cloud manufacturing system include orders, products, processes and machines; production units represent one or more manufacturing resource entities abstracted from each cloud manufacturing service provider.

[0064] The knowledge graph of the cloud manufacturing system describes the real-time manufacturing status of the cloud manufacturing system with nodes and connecting edges. Nodes represent entities in the cloud manufacturing system, such as orders, products, processes, and machines; connecting edges represent the relationship between nodes; the knowledge graph G formula is as follows:

[0065] G=(V,E)

[0066] V represents the node set, and the formula is:

[0067] V=(M,O,Pd,Pc)

[0068] Among them, M is the machine (in the cloud manufacturing system, it refers to the production unit of the cloud manufacturing service provider consisting of multiple production equipment, transfer equipment and warehousing equipment, etc.), O is the order, Pd represents the product, and Pc represents the process.

[0069] E represents the connected edge set, and the formula is:

[0070] E=(MO,MPd,MPc,OPd,…,PdPc)

[0071] Among them, MO is the connection between the machine and the order, indicating whether the machine is executing the order, including two states: the machine is processing the order and the machine is not processing the order; MPd is the connection between the machine and the product, indicating whether the machine is processing the product, including two states: the machine is processing the product and the machine is not processing the product, which is determined by the position of the product in the actual production process; MPc is the connection between the machine and the process, indicating whether the machine executes a specific process; OPd means that the order consists of products, and its connection can be represented by the processing progress of the products in the order; PdPc is the connection between the product and the process, indicating whether the product is executing the process.

[0072] Step S2: constructing a digital twin model set in the cloud model library of the cloud manufacturing system, including two processes: representation of the digital twin model in the cloud model library and digital twin model set (domain).

[0073] The digital twin of the cloud manufacturing system includes digital twin models, manufacturing resource entities, simulation models, relational models and data models; the digital twin model is connected to the manufacturing resource entities through simulation models, relational models and data models; the simulation model is connected to the relational model, and the relational model is connected to the data model; the digital twin model is the digital mapping of the manufacturing resource entity in cyberspace; the simulation model is an intelligent algorithm that supports the digital twin model to run in an actual production state; the relational model is the operational logical relationship of the manufacturing resource entity in actual production.

[0074] The cloud model library is a container for storing digital twin models in the cloud manufacturing system. It supports the creation, update, query, deletion and reuse of digital twin models. Digital twin models are diverse, covering four aspects: geometric models, behavioral models, logical models and performance models.

[0075] like Figure 2 As shown in the figure, the representation of the digital twin model in the cloud model library covers four aspects: geometry, behavior, logic and performance models.

[0076] The geometric model is a 3D model corresponding to the manufacturing resource entity, which can be constructed by 3D modeling tools such as UG, CATIA, Pro / E, SolidWorks, and ZWCAD. The geometric model is represented as:

[0077] DTM GM =(ME,GP)

[0078] Among them, ME (Model External) represents the external model, and GP (Geographical Position) represents the geographical position.

[0079] The behavior model represents the real-time response and behavior of manufacturing resource entities under the influence of the external environment and internal operating mechanism, which can cover the functions of manufacturing resource entities such as production equipment, transfer equipment, and storage equipment. The production equipment state machine model, transfer equipment state machine model, and storage equipment state machine model constructed by the finite state machine are used as sub-models of the cloud manufacturing provider's manufacturing resource behavior model to realize their own state transfer, and then the sub-models are assembled into the behavior model of the entire cloud manufacturing system according to the production relationship of cloud manufacturing; the finite state machine model represents a mathematical model of a finite number of states and the transitions and actions between these states. The state reflects the input changes from the beginning of the system to the present moment, the transition represents the switch from one state to another, and the action describes the activities that need to be performed at a given moment; the behavior model is expressed as:

[0080]

[0081] The behavior model is composed of multiple manufacturing resource entity sub-behavior models coupled together. n Indicates the number of manufacturing resource entities, DTM EU The sub-behavior model of each manufacturing resource entity is represented by the following formula:

[0082]

[0083] Where X represents the input event set, Y represents the output event set, DI represents the input data set, DO represents the output data set, D represents the coupling member name set, {F i} represents a sub-function set, EIC represents an external input coupling function set, DIC represents an input data coupling function set, IEC represents an internal event coupling function set, IDC represents an internal data coupling function set, EOC represents an output event coupling function set, and DOC represents an output data coupling relationship set.

[0084] The logical model represents the internal logic of the actual production process and operation of the manufacturing resource entity. It is based on the manufacturing resource entity, follows the information such as geographical layout and production mode, and is constructed according to the behavior and response (operation) rules of the production unit in the actual production process, so that it has behavioral characteristics, response mechanisms and state conversion capabilities; the logical model can be expressed as:

[0085] DTM LM =(DTM GM ,DTM BM ,DTM OR )

[0086] Among them, DTM OR Represents a set of operating rules, including behavioral information transmission rules between different digital twin logic models, information interaction rules between models, synchronous advancement rules between superior and subordinate models, and mutual exclusion priority rules between models of the same level.

[0087] The performance model is an intelligent analysis model built on the basis of geometric model, behavior model and logic model according to the operation rules of cloud manufacturing system. It includes prediction or decision-making models for performance indicators such as production progress, yield rate, production cost and energy consumption. This is the core of digital twin model. The performance model can be expressed as

[0088] DTM PM =(DTM MM ,DTM DDM )

[0089] DTM MMRepresentation mechanism model, DTM DDM Represents a data-driven model.

[0090] Combining the above geometric model, behavior model, logic model and performance model, the jth digital twin model in the cloud model library is expressed as:

[0091]

[0092] in, Represents the jth digital twin model in the cloud model library; They respectively represent the geometric model, behavioral model, logical model and performance model of the j-th digital twin model.

[0093] The digital twin model set (domain) is recorded as:

[0094]

[0095] Step S3: Based on the transferable characteristics of the cloud manufacturing digital twin model, a set of requirements for digital twin modeling (including geometric models, behavioral models, logical models and performance models) of cloud manufacturing service demanders is constructed.

[0096] The parts and parameters of the model that can be migrated are: geometric model (size, assembly relationship and material, etc.), behavioral model (production equipment state machine model, transfer equipment state machine model and storage equipment state machine model, etc.), logical model (spatial position, process flow and constraint rules, etc., which are constructed according to the behaviors and operating rules that occur in the actual production process of cloud manufacturing, so that it has behavioral characteristics, response mechanisms, and state conversion capabilities), performance model (the total model of the cloud manufacturing system assembled from manufacturing resource entity sub-models, which contains prediction or decision-making models for indicators such as production progress, yield, production cost, and energy consumption).

[0097] The cloud manufacturing service demand side digital twin model construction requirements are expressed, which covers four aspects: geometry, behavior, logic and performance models. Let the i-th digital twin model construction requirement be expressed as:

[0098]

[0099] in, Indicates the construction requirements of the i-th digital twin model; They respectively represent the geometric model requirements, behavioral model requirements, logical model requirements, and performance model requirements in the i-th digital twin model construction requirements.

[0100] The demand set (domain) for the digital twin model construction of cloud manufacturing service demanders is recorded as:

[0101]

[0102] Step S4: select the most suitable model from the digital twin model library for migration based on the preference recommendation algorithm.

[0103] The preference algorithm captures the interaction / preference relationship between modeling requirements and the model to be selected based on the attention mechanism; the recommendation algorithm is a prediction algorithm that is converted by predicting whether the model in the digital twin model library will be selected. The specific implementation process is as follows Figure 3 shown.

[0104] The attention mechanism is used to capture the interaction / preference relationship between modeling requirements and the model to be selected. The preference of the i-th construction requirement for the j-th candidate model is expressed as:

[0105]

[0106] in, They represent preferences for geometry, behavior, logic, and performance, respectively. The symbols in parentheses represent the importance of the same group of models, that is, the preference value. The preference value is not less than 0, and x gm +x bm +x lm +x pm =1.

[0107] For example, select model B as the base model, and then fine-tune model B to establish model A. The importance of model migration is That is, when model migration selection is required, logic and performance are given high priority (preference value) in the same set of models.

[0108] The interaction matrix between the model construction requirements and the digital twin model is recorded as Given an interaction matrix Under the conditions of and knowledge graph G, the goal of migration modeling is to predict whether the model in the digital twin model library will be selected according to the needs of model construction. The prediction formula is:

[0109]

[0110] in, Indicates the requirements built according to the model cm de Select vm pr The probability of performing migration modeling, ω is the parameter to be trained for the prediction model.

[0111] Due to the complex coupling between geometric, behavioral, logical and performance models, a digital twin model is usually a combination of geometric, behavioral, logical and performance models, and no single model can be selected separately. For example, in the cloud manufacturing service process, a certain aspect (such as the behavioral model) will be emphasized. When this behavioral model is selected, the geometric, logical and performance models associated with it will be selected together, that is, the overall twin model targeted by these models will be selected. When performing model selection in (0.2 indicates the importance of the same group of models), then the associated To be selected together, that is, to choose When the focus is on choice , then the associated Also selected, that is, selected

[0112] When new modeling requirements arise, the recommendation algorithm f predict (.) Push the most suitable digital twin model from the cloud model library for migration modeling. It can be expressed as the dot product of the model building requirement representation and the candidate model representation Represents the dot product of the vector, constructing the interaction between the label of the candidate model and the actual migration, that is, when in cm,vm =1, indicating that according to cm de Select VM pr Migrate in cm,vm =0, no migration.

[0113] Finally, the migration modeling method considering preferences is used to complete the efficient migration of the digital twin model of the cloud manufacturing system. During the operation of the digital twin model, the virtual data acquisition terminal collects data of the virtual manufacturing resource entity in real time and transmits the data to the cloud manufacturing platform. The data acquisition terminal reads data from the cloud manufacturing platform and acts on the manufacturing resource entity. The manufacturing resource entity performs production, transportation or warehousing actions to obtain the data of the manufacturing resource entity, and transmits the data of the manufacturing resource entity to the cloud manufacturing platform for comparison with the data of the digital twin model. If the data of the manufacturing resource entity is consistent with that of the digital twin model, it means that the model migration is successful.

[0114] The digital twin migration modeling method for cloud manufacturing systems taking preference into consideration provided by the present invention makes full use of the data and knowledge in cloud manufacturing to effectively migrate the existing models (fine-tune parameters). The theoretical method is simple and easy to implement, which greatly improves the modeling efficiency of the digital twin model and meets the requirements of accurate and efficient model migration. It is the precise application of digital twins in cloud manufacturing systems and has important value for improving the intelligent level of cloud manufacturing services. The present invention can be used to guide the accurate and rapid construction process of digital twin models in cloud manufacturing systems, solve the problems of long modeling time and high cost of twin models, and can effectively improve the efficiency of twin model construction.

[0115] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A digital twin migration modeling method for cloud manufacturing system considering preference, characterized in that: The following steps are involved: (1) Based on the production relations of the cloud manufacturing system, a knowledge graph of the cloud manufacturing system is constructed; (2) constructing a set of digital twin models in a cloud model library of a cloud manufacturing system; the digital twin model includes a geometric model, a behavioral model, a logical model, and a performance model; (3) Based on the transferability of cloud manufacturing digital twin models, a set of requirements for digital twin modeling of cloud manufacturing service demanders is constructed; (4) Based on the attention mechanism, the interaction and preference relationship between modeling requirements and the models to be selected is captured, and a recommendation algorithm is constructed to select the optimal model from the digital twin model library for migration.

2. According to a preference-based cloud manufacturing system digital twin migration modeling method according to claim 1, it is characterized in that: The implementation process of step (1) is as follows: The production process of the cloud manufacturing system is analyzed, and the production activities of the cloud manufacturing system are described as the entire process in which the product is processed and assembled on different production units according to the production schedule until it is finally completed; the key elements of the cloud manufacturing system production process include orders, products, processes and machines; the production unit represents one or more manufacturing resource entities abstracted from each cloud manufacturing service provider; The knowledge graph of the cloud manufacturing system describes the real-time manufacturing status of the cloud manufacturing system with nodes and connecting edges. Nodes represent entities in the cloud manufacturing system, and connecting edges represent the relationship between nodes. The formula of the knowledge graph G is as follows: G=(V,E) V=(M,O,Pd,Pc) E=(MO,MPd,MPc,OPd,…,PdPc) Among them, V represents the node set, E represents the connection edge set, M is the production unit, O is the order, Pd represents the product, and Pc represents the process; MO is the connection between the machine and the order, including two states: the machine is processing the order and the machine has not processed the order; MPd is the connection between the machine and the product, including two states: the machine is processing the product and the machine has not processed the product; it is determined by the position of the product in the actual production process; MPc is the connection between the machine and the process, indicating whether the machine executes a specific process; OPd indicates that the order consists of products, and its connection can be represented by the processing progress of the products in the order; PdPc is the connection between the product and the process, indicating whether the product is executing the process.

3. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The geometric model in step (2) is: DTM GM =(ME,GP) Among them, ME represents the external model and GP represents the geographical location.

4. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The behavior model in step (2) represents the real-time response and behavior of the manufacturing resource entity under the influence of the external environment and the internal operation mechanism, and is formed by coupling multiple manufacturing resource entity sub-behavior models, which are: Among them, M n Indicates the number of manufacturing resource entities, DTM EU The sub-behavior model of each manufacturing resource entity is represented by the following formula: Where X represents the input event set, Y represents the output event set, DI represents the input data set, DO represents the output data set, D represents the coupling member name set, {F i } represents a sub-function set, EIC represents an external input coupling function set, DIC represents an input data coupling function set, IEC represents an internal event coupling function set, IDC represents an internal data coupling function set, EOC represents an output event coupling function set, and DOC represents an output data coupling relationship set.

5. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The logical model in step (2) represents the internal logic of the actual production process and operation of the mapping manufacturing resource entity, which is: DTM LM =(DTM GM ,DTM BM ,DTM OR ) Among them, DTM OR Represents a set of operating rules, including behavioral information transmission rules between different digital twin logic models, information interaction rules between models, synchronous advancement rules between superior and subordinate models, and mutual exclusion priority rules between models of the same level.

6. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The performance model in step (2) includes a prediction or decision model for performance indicators, which is expressed as: DTM PM =(DTM MM ,DTM DDM ) Among them, DTM MM Representation mechanism model, DTM DDM Represents a data-driven model; The jth digital twin model in the cloud model library is represented as: in, Represents the jth digital twin model in the cloud model library; They represent the geometric model, behavioral model, logical model and performance model of the j-th digital twin model respectively; The digital twin model set is recorded as:

7. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The implementation process of step (3) is as follows: The requirements for building the i-th digital twin model are expressed as: in, Indicates the construction requirements of the i-th digital twin model; They respectively represent the geometric model requirements, behavioral model requirements, logical model requirements, and performance model requirements in the i-th digital twin model construction requirements; The demand set for building the digital twin model of cloud manufacturing service demanders is recorded as:

8. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The root of step (4) captures the interaction and preference relationship between the modeling requirements and the model to be selected according to the attention mechanism, and the implementation process is as follows: The preference of the i-th build requirement for the j-th candidate model is expressed as: in, They represent preferences for geometry, behavior, logic, and performance, respectively. The symbols in parentheses represent the importance of the same group of models, that is, the preference value. The preference value is not less than 0, and x gm +x bm +x lm +x pm =1.

9. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The implementation process of constructing the recommendation algorithm described in step (4) is as follows: The interaction matrix between the model construction requirements and the digital twin model is recorded as Given an interaction matrix Under the conditions of and knowledge graph G, the goal of migration modeling is to predict whether the model in the digital twin model library will be selected according to the needs of model construction. The prediction formula is: in, Indicates the requirements built according to the model cm de Select vm pr The probability of performing migration modeling, ω is the parameter to be trained for the prediction model.

10. The method for digital twin migration modeling of a cloud manufacturing system considering preference according to claim 1, characterized in that: The process of selecting the optimal model from the digital twin model library for migration in step (4) is as follows: When new modeling requirements arise, the recommendation algorithm f predict (.) Push out the optimal digital twin model from the cloud model library for migration modeling; Denoted as the dot product of the model building requirement representation and the candidate model representation Represents the dot product of the vector, constructing the interaction between the label of the candidate model and the actual migration, that is, when in cm,vm =1, indicating that according to cm de Select VM pr Migrate in cm,vm =0, no migration.

Images