Digital twin system construction method and system based on flexible job shop scheduling

Abstract

The invention discloses a digital twinning system construction method and system based on flexible job shop scheduling, and relates to the technical field of digital twinning, and the method specifically comprises the steps: constructing a twinning model of a physical workshop according to the production data of the physical workshop; building a workshop scheduling discrete event simulation model based on the twinborn model of the physical workshop; based on an internal and external virtual-real interaction mechanism, virtual-real mapping interaction between the twin model of the physical workshop and the physical workshop and virtual-real mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model are realized, and a virtual workshop is generated; data generated in the interaction process is stored in a twin data pool in real time; a workshop service system is constructed based on a three-dimensional visual integration technology, so that a physical workshop and a virtual workshop carry out interactive feedback, and construction of a scheduling digital twin system is completed. According to the method, the self-adaptive capability of the system in the face of complex production tasks is enhanced, and flexible scheduling according to actual conditions is supported.

Description

Technical Field

[0001] The present invention relates to the field of digital twin technology, and more specifically to a method and system for constructing a digital twin system based on flexible job shop scheduling. Background Art

[0002] At present, digital twin technology, as an innovative solution, creates a digital representation of physical entities in virtual space, which can reflect the status of physical entities in real time and predict and optimize the behavior of entities through data analysis. For flexible job shop scheduling, digital twin modeling combines the latest achievements in modern information technology, simulation technology and artificial intelligence, aiming to optimize actual production activities by building a highly realistic virtual environment, thereby helping enterprises improve their competitiveness.

[0003] However, there are many shortcomings in the traditional digital twin modeling system for flexible job shop scheduling. First, the decision-making ability and adaptability of the twin model are low, which limits its ability to make optimal decisions in a complex and changing production environment. Secondly, the dynamic iterative optimization of scheduling decision performance is poor, and it cannot respond quickly to changes in production demand. Furthermore, the communication load of the twin model in a stable state is high, resulting in resource waste and inefficiency. In addition, the virtual-real interaction and evolutionary update of the twin model are difficult to implement, affecting the flexibility and maintainability of the system. In practical applications, these limitations make the traditional system lack a more practical twin modeling system, making it difficult to support the rapid construction, model reuse and stable operation of twin scenarios.

[0004] Therefore, improving the dynamic optimization performance of the digital twin system for flexible job shop scheduling and reducing the communication load are issues that technical personnel in this field urgently need to solve. Summary of the invention

[0005] In view of this, the present invention provides a method and system for constructing a digital twin system based on flexible job shop scheduling, which overcomes the above-mentioned defects.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A method for constructing a digital twin system based on flexible job shop scheduling, the specific steps are:

[0008] Building a twin model of the physical workshop according to the production data of the physical workshop;

[0009] Building a workshop scheduling discrete event simulation model based on the twin model of the physical workshop;

[0010] Based on the internal and external virtual-real interaction mechanism, the virtual-real mapping interaction between the twin model of the physical workshop and the physical workshop, and the virtual-real mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model are realized to generate a virtual workshop; and the data generated during the interaction process are stored in the twin data pool in real time;

[0011] A workshop service system is constructed based on three-dimensional visualization integration technology, so that the physical workshop and the virtual workshop can interact and feedback with each other, thus completing the construction of the scheduling digital twin system.

[0012] Furthermore, the twin model of the physical workshop includes a production factor model, a production behavior logic model and a scheduling knowledge rule model, and its construction steps are:

[0013] Generate the production factor model based on the production data of the physical workshop using an ontology modeling method;

[0014] Classifying manufacturing resources in the production factor model, assigning time attributes to the workshop state transitions of each duration in each category, generating multiple agents, and constructing the production behavior logic model;

[0015] The scheduling knowledge rule model is constructed based on the interaction rules and scheduling rule knowledge of each Agent in the production behavior logic model.

[0016] Furthermore, the production behavior logic model is constructed by using a timed colored Petri net.

[0017] Furthermore, the ontology modeling method is expressed as:

[0018] O = {C, A, R, I, M};

[0019] In the formula, C represents the basic class; A represents the attribute set of production factors; R represents the relationship between attributes; I represents the instance set; and M represents the mapping relationship between production factor entities and concepts.

[0020] Furthermore, the internal and external virtual-reality interaction mechanism includes an external virtual-reality interaction mechanism and an internal virtual-reality interaction mechanism. The external virtual-reality interaction mechanism is the interaction between the physical workshop, the virtual workshop, the twin data pool and the workshop service system; the internal virtual-reality interaction mechanism includes: the production factor model interacts with the production behavior logic model and the scheduling knowledge rule model iteratively interacts with the workshop scheduling discrete event simulation model.

[0021] Furthermore, the iterative interaction between the scheduling knowledge rule model and the workshop scheduling discrete event simulation model is realized based on the scheduling rule knowledge mining model of NGEP-FS and the scheduling decision Agent adaptive scheduling model.

[0022] Furthermore, the update mechanism of the digital twin system is:

[0023] Adopting a state-based update strategy, the current state data of the workshop is compared with the historical data, the problematic data attribute entries are identified, and corresponding corrections are made.

[0024] A digital twin system construction system based on flexible job shop scheduling, comprising:

[0025] A twin model building module, used to build a twin model of the physical workshop according to the production data of the physical workshop;

[0026] A simulation model building module, used to build a workshop scheduling discrete event simulation model based on the twin model of the physical workshop;

[0027] A virtual interaction module, based on the internal and external virtual-real interaction mechanism, realizes the virtual-real mapping interaction between the twin model of the physical workshop and the physical workshop, and the virtual-real mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model, to generate a virtual workshop; and stores the data generated during the interaction process in real time to the twin data pool;

[0028] The system generation module builds a workshop service system based on three-dimensional visualization integration technology, enables the physical workshop and the virtual workshop to interact and feedback, and completes the construction of the scheduling digital twin system.

[0029] It can be seen from the above technical solutions that the present invention discloses a method and system for constructing a digital twin system based on flexible job shop scheduling, which has the following beneficial effects compared with the prior art:

[0030] Improve scheduling efficiency and accuracy: By building a scheduling digital twin model for the flexible job shop, a formal expression of the production factors and their relationships in the physical shop is achieved, making scheduling decisions more accurate and able to respond to changes in the production environment in real time, thereby improving the overall production scheduling efficiency.

[0031] Enhance system adaptability: The hybrid modeling method of multi-agent system and Petri net is introduced to construct the MATCPN behavior logic model, which not only takes into account the behavior logic of manufacturing resources, but also integrates the knowledge rules of multi-agents, enhances the system's adaptive ability in the face of complex production tasks, and supports flexible scheduling according to actual conditions.

[0032] Optimize dynamic performance iteration: Based on the internal and external virtual-real interaction mechanism, by integrating the interaction between production factors, behaviors and rules, virtual-real mapping is achieved, supporting the agent's adaptive scheduling decision-making and dynamic iterative optimization of performance, which helps to continuously improve problems in the production process.

[0033] Reduce network communication burden: In order to ensure that the twin model does not increase unnecessary network communication burden due to frequent data updates in a stable operating state, the present invention proposes a state-based twin model update strategy, which effectively reduces the frequency of data transmission and reduces bandwidth occupancy and energy consumption while ensuring information accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0035] Figure 1 A schematic diagram of the process of constructing a digital twin model for flexible job shop scheduling provided by the present invention;

[0036] Figure 2 A schematic diagram of the modeling and interaction system of the digital twin model for flexible job shop scheduling provided by the present invention;

[0037] Figure 3 A schematic diagram of the classification of physical workshop data provided by the present invention;

[0038] Figure 4 A schematic diagram of a manufacturing resource behavior model based on MACTPN provided by the present invention;

[0039] Figure 5 A schematic diagram of the MACTPN model of the multi-agent scheduling decision knowledge rules provided by the present invention;

[0040] Figure 6 A schematic diagram of a MACTPN workshop production operation logic model including scheduling rules provided by the present invention;

[0041] Figure 7 A schematic diagram of the interactive operation mechanism of the digital twin provided by the present invention;

[0042] Figure 8 A schematic diagram of the external interactive operation mode of the twin system provided by the present invention;

[0043] Fig. 9 A schematic diagram of the interaction between the simulation environment and the scheduling decision model provided by the present invention;

[0044] Fig.10 A schematic diagram of the update evolution mechanism of the workshop scheduling twin model provided by the present invention;

[0045] Fig.11This is a flow chart of the update rules of the workshop scheduling twin model provided by the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] On the one hand, the embodiment of the present invention discloses a method for constructing a digital twin system based on flexible job shop scheduling, such as Figure 1 and Figure 2 As shown, the specific steps are:

[0048] Step 1: Build a twin model of the physical workshop based on the production data of the physical workshop;

[0049] Step 2: Build a discrete event simulation model for workshop scheduling based on the twin model of the physical workshop;

[0050] Step 3: Based on the internal and external virtual-real interaction mechanism, the virtual-real mapping interaction between the twin model of the physical workshop and the physical workshop, as well as the virtual-real mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model are realized to generate a virtual workshop; and the data generated during the interaction process is stored in the twin data pool in real time;

[0051] Step 4: Build a workshop service system based on 3D visualization integration technology to enable interactive feedback between the physical workshop and the virtual workshop to complete the construction of the digital twin system.

[0052] Furthermore, for the virtual model layer in the adaptive scheduling framework of the flexible job shop, an ontology-based production factor model is constructed based on various data generated in the production process of the physical workshop; the production operation process of the workshop is described by an object-oriented timed Petri net, and a multi-agent timed Petri net production behavior logic rule model is proposed to improve the decision-making ability and adaptability of the model. An internal and external virtual-real interaction mechanism of the digital twin system is proposed to support the interaction between the agent and the simulation model, and realize the dynamic iterative optimization of the scheduling decision performance. In order to effectively reduce the communication load of the twin model in the stable state, a state-based twin model update strategy is proposed to realize the virtual-real interaction and evolutionary update of the twin model.

[0053] In one embodiment, the twin model of the physical workshop includes a production factor model, a production behavior logic model, and a scheduling knowledge rule model, and the construction steps are:

[0054] Step 11: Generate a production factor model based on the production data of the physical workshop using the ontology modeling method;

[0055] Step 12: Classify manufacturing resources in the production factor model, assign time attributes to the workshop state transitions of each category, generate multiple agents, and build a production behavior logic model;

[0056] Step 13: Construct a scheduling knowledge rule model based on the interaction rules and scheduling rule knowledge of each Agent in the production behavior logic model.

[0057] Furthermore, in step 11, the data generated in the production process of the physical workshop is collected to construct a production factor model based on the ontology, specifically:

[0058] Based on the data generated by the production process of the physical workshop, the scheduling knowledge is mined and applied, and the updating, optimization and development of the physical / virtual model are promoted through integration and fusion. At the same time, the attributes and relationships of the production factors are uniformly formalized for use by the twin system. The model structure of the production factors is described by ontology modeling, and the mapping process from the ontology model structure to the data is completed, so as to realize the instantiation of the production factors and complete the construction of the production factor model.

[0059] Furthermore, in the process of modeling the production factor model, the data collected from the physical entity object and its environment and the data generated after the simulation of various models are applied. As a semantic technology, ontology models the production factors in a structured way. The ontology of the production factor is represented by a five-tuple (O = {C, A, R, I, M}). The production factor is instantiated through ontology modeling to complete the construction of the production factor twin model. It provides power for the update and optimization of the digital twin model.

[0060] Among them, C represents the basic class, that is, the definition of the production factor entity; A represents the attribute set of the production factor; R represents the relationship, which describes the relationship between attributes; I represents the instance set, the concrete object; M represents the mapping relationship between the production factor entity and the concept.

[0061] In one embodiment, a production behavior logic model is constructed by using a timed colored Petri net.

[0062] In step 12, the data in the physical workshop production process is processed to form a logical model of the workshop production behavior, specifically:

[0063] Step 121. After completing step 1, the operation behavior of the workshop is modeled, and a production behavior logic model based on the Multi-Agent Timed Color Petri Net (MATCPN) is constructed. For the transition between states with a certain time interval, the state transition with a duration should be given a time attribute. At the same time, the Timed Color Petri Net (TCP-net) is applied, and the coloring function is introduced into the simplified Petri net model to give different color values ​​to different manufacturing resources in the workshop. The operation mode adopts the cooperation of multiple agents to expand the knowledge rules describing the library and transition based on the object model in the Petri net.

[0064] Step 122, using the composition and coordinated interaction relationship of the multi-agent scheduling decision system of the flexible job shop, the operation process of each agent is mapped to the library and transition in the Petri net model, and the interactive cooperation of the equipment resource agent, the production task agent, the state monitoring agent and the scheduling decision agent is constructed to integrate the MATCPN model of multi-agent scheduling knowledge rules, that is, the scheduling knowledge rule model. In order to effectively solve the token conflict and transition conflict problems during the production state transfer, the cooperative operation of multiple agents is adopted to expand the knowledge rules describing the library and transition based on the object model in the Petri net, so as to effectively respond to the evolution of the operation state of the flexible job shop. At the same time, in order to improve the modeling efficiency of the workshop production behavior logic model and enhance the accuracy and flexibility of the model, this embodiment adopts a hierarchical form to construct the MATCPN model, and the specific definition is as follows:

[0065] MATCPN={APS, TC, AC, KB, M0};

[0066] APS represents the finite set of Agents in the workshop production behavior logic model. n}, AP i Represents the equipment resource agent, production task agent, status monitoring agent, and scheduling decision agent in the workshop; TC represents the communication connection used for information transmission between agents; A represents the connection arc function between agent and transition; C represents the color set associated with each transition and token in the Petri net; KB represents the set of scheduling knowledge rules associated with the library; M0 represents the initial mark of the MATCPN model.

[0067] The MATCPN model has a i The internal state change behavior is encapsulated and structured as: AP i = {P i , TC i , Ai , Ci, K i , M.P. i , R.S. i , M oi};

[0068] Where P i Represents the set of places in the agent, describing the state change inside the Agent, Pi = {P i j,， j=1,2,3……n p}.TC i Represents the transition set TC within the proxy position j =TC i j,， j=1,2,3……n TC}, according to the time characteristics, it can be divided into sequential transition, deterministic delay transition and random delay transition; A i Represents the place P i and Transition TC i The set of connecting arc functions, A i ={A i j,， j=1,2,3……n T}; C i Indicates the proxy position and the library P i 、Change TC i A set of associated colors. K i K represents the knowledge rule set inside the Agent. i ∈DT kno , DT kno is the scheduling knowledge rule model in the twin model, DT kno ={Kno obj , Kno eqp , Kno job}, Kno obj The knowledge rules and Kno eqp Corresponding machine allocation sub-agent knowledge rules, Kno job Corresponding to the knowledge rules of the workpiece sorting sub-agent. i Represents the message bit set of the library, MP i ={IM i , OM i}, with the input message bit set IM i and output message bit set OM i It is the communication interface between agents; RS i It represents the information base, which can share the internal real-time status information of the Agent with other Agents, providing information support for the operation and scheduling of the production process;oi Indicates AP i The initial identification of the Petri net model describes the initial distribution position of the tokens in the Petri net model. According to the above definition, based on the operation mode of coordinated interaction between multiple agents and the operation mode of internal state change of a single agent, a hierarchical and structured MATCPN production behavior logic rule model is constructed, which effectively describes and characterizes the evolution of the production operation state and operation logic of the entire manufacturing workshop.

[0069] Step 123, after completing step 121 and step 122, the MATCPN model constructed based on the hierarchical modeling idea provides a reconfigurable modeling system for the workshop production system. At the same time, by mixing the advanced Petri net with the multi-agent system, the production factor model, the production behavior logic model and the scheduling knowledge rule model are effectively integrated to realize the digital twin scheduling simulation model of the flexible workshop. According to the composition and coordinated interaction relationship of the multi-agent scheduling decision system of the flexible workshop, the operation process of each agent is mapped to the library and transition in the Petri net model, and the interactive and cooperative scheduling process of the equipment resource agent, the production task agent, the state monitoring agent and the scheduling decision agent is described to construct a MATCPN model integrating multi-agent scheduling knowledge rules. The interconnection between the input and output bits of the MATCPN model of different agents is completed by communication transitions. The occurrence of transitions will cause the transfer of the internal state of the agent, so that the agent enters the scheduling decision reasoning stage.

[0070] In one embodiment, the internal and external virtual-reality interaction mechanism includes an external virtual-reality interaction mechanism and an internal virtual-reality interaction mechanism. The external virtual-reality interaction mechanism is the interaction between the physical workshop, the virtual workshop, the twin data pool and the workshop service system; the internal virtual-reality interaction mechanism includes: the interaction between the production factor model and the production behavior logic model, and the iterative interaction between the scheduling knowledge rule model and the workshop scheduling discrete event simulation model.

[0071] In one embodiment, the iterative interaction between the scheduling knowledge rule model and the workshop scheduling discrete event simulation model is implemented based on the scheduling rule knowledge mining model of NGEP-FS and the scheduling decision agent adaptive scheduling model.

[0072] Furthermore, the production factor ontology model and the workshop production operation behavior model are integrated into the virtual simulation model to carry out virtual-real interaction of digital twins, specifically:

[0073] The physical workshop PS, virtual workshop VS, twin data pool DD, and workshop service system SSS interact with each other to realize external interaction, ensuring the adaptive scheduling optimization and control of the workshop at the outer level; the interaction between the components of the physical workshop PS, virtual workshop VS, twin data pool DD, and workshop service system SSS is internal interaction, which provides internal guarantee through the adaptive scheduling operation mode of workshop self-organization, self-learning and self-optimization.

[0074] In one embodiment, the update mechanism of the digital twin system is:

[0075] Adopting a state-based update strategy, the current state data of the workshop is compared with the historical data, the problematic data attribute entries are identified, and corresponding corrections are made.

[0076] Furthermore, in the update mechanism of the digital twin model, a state-based update strategy is adopted to reduce the network communication burden caused by frequent data updates in a stable state. Attribute parameters representing the operating state of the model are extracted from the database, and different attribute changes correspond to different update requirements. By comparing and analyzing the current state data of the workshop with the historical data in the database, the data attribute entries with problems are obtained and the causes of the problems are analyzed.

[0077] On the other hand, this embodiment discloses a digital twin system construction system based on flexible job shop scheduling, carrying the above method, including:

[0078] The twin model building module is used to build a twin model of the physical workshop based on the production data of the physical workshop;

[0079] A simulation model building module is used to build a discrete event simulation model for workshop scheduling based on the twin model of the physical workshop;

[0080] The virtual interaction module realizes the virtual-reality mapping interaction between the twin model of the physical workshop and the physical workshop, and the virtual-reality mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model based on the internal and external virtual-reality interaction mechanism, generates a virtual workshop, and stores the data generated during the interaction process in real time to the twin data pool;

[0081] The system generation module builds a workshop service system based on 3D visualization integration technology, enabling the physical workshop and the virtual workshop to interact and feedback, thus completing the construction of the digital twin system.

[0082] In one embodiment, it also includes a data acquisition module, which collects equipment status, workpiece status and environmental parameters in real time through sensors and information systems, builds an ontology-based production factor model, completes the description of production factors, and the data source of the physical workshop, such as Figure 3 shown.

[0083] In the twin model construction module, ontology technology is used to uniformly model production factors and achieve standardized descriptions of equipment, materials, and production tasks. As a semantic technology, ontology abstracts and digitally describes the workshop manufacturing process and can clearly explain the characteristics and logical relationships of production factors. The model structure of production factors is described by ontology modeling, and the mapping process from the ontology model structure to data is completed to achieve the instantiation of production factors, thereby completing the construction of the production factor model.

[0084] The timed Petri net based on the multi-agent system describes the dynamic process of workshop production and supports the scheduling logic of concurrent and asynchronous operations. There are many types of production equipment in the actual manufacturing workshop, among which the operation logic of processing equipment is the most complex and is also the core of workshop production scheduling. Therefore, this embodiment uses the operation logic of processing equipment to cover the operation logic of workshop transportation equipment, tooling fixtures and other production auxiliary equipment, and constructs a MATCPN-based equipment resource agent behavior logic model to describe the state transition behaviors of workshop equipment resources such as processing, loading, and failure. The reasoning process of the model behavior is as follows Figure 4 As shown in the figure: When the equipment resource agent receives the message requesting the workpiece to be processed, it enters the message bit IM I A token is generated. If there is a token in the processing equipment idle bit MReady_p, the sequential transition Enter_ti is triggered, and the equipment processing preparation bit Setup_r is entered. After loading, the workpiece processing status bit Processing_p is entered. During the equipment processing, Processing_p can trigger three delayed transitions, Failure1_ts, Failure2_te and Normal_te, according to the knowledge and probability set by the dynamic event trigger mechanism, which respectively indicate equipment failure, reprocessing and normal processing. When the workpiece is processed normally, the delayed transition Normal_te and the workpiece unloading delayed transition Unload1_te are triggered, and the output message bit OM1 is entered, and the state of the processing equipment is changed to the idle state. When the workpiece requires reprocessing, the delayed transition Failure2_te is triggered. After a certain processing time, the workpiece enters the waiting unloading status bit Unload2_w, and the unloading delayed transition Unload2_te is triggered, and then the output message bit OM2 is entered, and the state of the processing equipment is changed to the idle state. When a processing device fails, the random delay transition Failure1_ts is triggered and the device enters the maintenance state MRepair_w. After the random delay transition MRepair_ts, the device enters the maintenance completed state MRepair_c and triggers the detection delay transition MCheck_te. The device is in an idle state. At the same time, the workpiece in the processing device enters the unload state Unload2_w after detection, and triggers the unload delay transition Unload2_te, entering the output message bit OM.i2 At the same time, the device turns to idle state. The operation logic of this model is composed of Figure 5 As shown in the figure: There is a token in the initial state of the manufacturing workshop, P1, which indicates that the production system has been initialized. When the order production plan is released in the workshop, the production task agent and the resource agent accept the order at the same time, triggering transitions T1 and T2 to complete the order acceptance. After that, the token is transferred to places P2 and P3, triggering the corresponding communication connections T3 and T4, entering the production task agent agent position and the resource agent agent position, and recording the status information of the workpiece and equipment resources. After the status information of the manufacturing element is sent by the production task agent and the resource agent, the communication transition T7 is triggered. The real-time monitoring agent judges whether the current workshop status meets the scheduling conditions based on the received current workshop status information and its own capabilities. Place P 12 and P 13 Must have tokens to achieve communication transition T 13 Triggering conditions. This means that the scheduling decision agent can only make scheduling decisions after receiving real-time status information sent by the production task agent and resource agent.

[0085] In the actual manufacturing environment, the production process of workpieces often requires multiple devices to operate together, which will inevitably lead to token conflicts. Therefore, from the perspective of the overall operation of the workshop, a workshop production behavior logic model integrating scheduling knowledge rules is constructed to form a token-driven workshop production operation system. Based on a specific control strategy, this model successfully connects the independent agents in the workshop and builds a token-driven production operation system. Figure 6The workshop MATCPN model shown in the figure has the following operation logic description: the token in the input information bit IM indicates that the request for workpiece processing is received. By triggering the communication connection represented by the instantaneous transition T_enter, the agent bit AP_mt of the production task agent and the agent bit AP_eqp of the resource agent are entered. These two agent agent bits respectively obtain the real-time status information of the current workpiece and equipment in the workshop, and communicate with AP_rm. Based on the established interaction rule MATCPN model, the corresponding communication transition T_scheduling is triggered, and the agent bit AP_sc of the scheduling decision agent is entered to make a decision. Then, the decision agent obtains the corresponding scheduling rules from the scheduling knowledge rule base, and converts the rules into a specific scheduling plan according to the workshop information, determines the workpiece to be processed and the processing equipment, triggers the corresponding communication transition T_mi, and enters the agent bit AP_i of the processing equipment i. When the workpiece needs to continue to complete the processing of the next process on the equipment, the communication transition T_mi_r is triggered; otherwise, the communication transition T_i is triggered, and the output message bit OM is entered to complete the processing of the workpiece. When the workpiece needs to be transported to other equipment for the next process, the input message bit IM will receive the token of the workpiece and continue the workshop scheduling. Through this continuous cycle, the production task of the workshop is completed. Figure 6 Where AP_sc, AP_eqp, AP_mt and AP_rm represent the status bits of the scheduling decision agent, equipment resource agent, production task agent and status monitoring agent respectively.

[0086] The virtual-reality interaction module connects the virtual workshop with the real workshop through a data interface to ensure information synchronization and feedback between the two. Figure 7 As shown in the figure, the internal and external interaction modes of the interactive objects of the digital twin system are as follows: the two-to-two interactions between PS, VS, DD, and SSS are external interactions; the interactions between the components of PS, VS, DD, and SSS are internal interactions. The interactive objects of the digital twin system provide an external guarantee for the adaptive scheduling optimization and control of the workshop through two-to-two interactions. The internal interaction provides an internal guarantee for the adaptive scheduling operation mode of the workshop's self-organization, self-learning, and self-optimization.

[0087] External interaction mechanisms such as Figure 8 As shown, the interaction between PS and SSS, Figure 8Middle stage① The interaction between PS and SSS is the core of the digital twin workshop management process, which ensures that the manufacturing elements are optimally configured and meet the production task requirements. When the digital twin workshop receives a production task, SSS formulates an initial resource configuration plan for the manufacturing elements based on the production factor management data and other related data in DD. Subsequently, according to the principle of closed-loop iterative control system, the SSS system obtains the status data of the manufacturing elements in PS in real time, corrects and optimizes the initial plan based on these data, and guides PS to adjust the status of the manufacturing elements through control instructions. PS continuously feeds back real-time data to SSS so that SSS can monitor and adjust in real time. When the real-time data does not match the plan, SSS will adjust the plan again. This data-based real-time feedback and iterative optimization make the management of manufacturing elements more accurate and efficient. Through the interaction between PS and SSS, the digital twin workshop can gradually optimize resource allocation and production plan until the optimal state is reached. At the same time, the data generated in this process is stored in DD to provide data support and driving force for subsequent stages. Therefore, the interaction between PS and SSS is an important means for the digital twin system to achieve intelligent and efficient production. It ensures the smooth completion of production tasks and improves the overall operation efficiency of the workshop through dynamic real-time management and optimization of manufacturing elements. In the interaction between SSS and VS, stage ② plays a core role in the iterative optimization of production plans in the digital twin workshop. In this stage, SSS and VS work closely together to promote the optimization of production plans. Based on the rich data provided by DD, VS uses multiple models to simulate and analyze the production plan and feeds back the results to SSS. SSS makes necessary adjustments and optimizations to the production plan based on these feedback data. This optimization process is an iterative cycle process. SSS and VS continuously exchange information and jointly improve the production plan until the optimal state is reached. Through this interactive method, stage ② ensures the accuracy and efficiency of the production plan, providing a strong guarantee for the smooth operation of the digital twin workshop. At the same time, all the data generated in this stage are stored in DD, providing valuable data support for subsequent stages and promoting the continuous optimization and development of the digital twin workshop. In the interaction between PS and VS, stage ③ is the key link for the digital twin workshop to achieve real-time iterative optimization of the production process, which reflects the dynamic real-time interaction between PS and VS. PS organizes production according to preset instructions and transmits real-time data to VS. VS updates its own status accordingly and compares it with the scheduled production plan. Once data differences are found, VS will quickly identify disturbance factors and correct the model. At the same time, VS conducts a comprehensive evaluation and optimization of the production process based on a variety of real-time and historical data. VS forms optimization measures through comprehensive analysis of all factors, all processes, and all businesses, and feeds back to PS in the form of real-time control instructions, thereby continuously optimizing the production process. Through the continuous iteration of this process, it is ensured that the production process can always maintain the optimal state. Through the iterative optimization of stages ①②③, while driving the model operation, the continuous updating and supplementation of DD is realized.

[0088] The production factor model interacts with the production behavior logic model, and the scheduling knowledge rule model interacts iteratively with the workshop scheduling discrete event simulation model.

[0089] Furthermore, the internal interactive operation mechanism is the key to efficient production scheduling. It consists of the interaction between the production factor model and the production behavior logic model, and the iterative interaction between the scheduling knowledge rule model and the workshop scheduling discrete event simulation model. The former interaction can not only reflect the production operation in real time, but also provide a basis for the mining of production scheduling knowledge and the learning and training of the agent, and drive the operation of the digital twin simulation environment. The latter interaction realizes the integration between the agent and the virtual simulation environment. The agent perceives the environment and acts on the environment, which improves the scheduling decision-making ability of the twin system. In the interaction between the production factor model and the production process operation logic model, the production factor model includes a geometric model and a physical model. Among them, the high-fidelity geometric model that reflects the characteristics of the physical entity size, material, etc. is the basic element for building a digital twin virtual scene. It can maintain good spatiotemporal consistency with the physical workshop. When building a digital twin virtual simulation environment, the geometric model is first imported into the virtual simulation environment, and the connection between the geometric model and the twin database is realized through the virtual-real interaction communication interface (Eqp_interface and M_interface), and the real-time dynamic physical attribute data is given to the geometric model based on the defined data mapping rules. Secondly, the real-time information of the geometric physical model is mapped to the message bit and information bit of the production behavior logic model of MATCPN, and the behavior change of the production operation logic model is realized through the defined state transition logic rules. Finally, the state node is converted into action code using programming language to drive the operation of the geometric model. The data generated during the interaction process is stored in the twin data pool in real time for sharing and use by other models and service systems, thereby ensuring the virtual-real interaction between the twin model and the physical entity. The interactive scheduling knowledge rule model between the simulation model and the multi-agent scheduling decision model consists of a scheduling rule knowledge mining model based on NGEP-FS and a scheduling decision agent adaptive scheduling model. The workshop scheduling simulation environment provides a workshop environment for training, learning and interaction for the scheduling rule knowledge mining model and the agent's adaptive scheduling. Based on the reinforcement learning mechanism and the scheduling rule knowledge base, through interaction with the simulation environment, the agent can continuously learn and optimize its own scheduling decision-making ability to improve the accuracy and efficiency of scheduling. In the twin service layer, the digital twin workshop scheduling simulation environment and the reinforcement learning training mechanism defined by Python encapsulation are defined as clients. Based on the communication protocol and interface address, the simulation software and the scheduling knowledge rule model are connected to realize the interaction between the simulation environment and the model. Fig. 9As shown in the figure. Through the external interaction of the digital twin, the model obtains real-time data to describe the workshop status and stores it in the database. The workshop status monitoring agent determines that the current workshop status meets the workshop scheduling conditions based on the workshop status data in the data pool. The scheduling decision agent will formulate a scheduling plan based on the knowledge rule model. The decision plan will be transmitted to the simulation model and then converted into specific scheduling events to drive the operation of the simulation environment. Through this cyclical interaction process, the continuity and efficiency of workshop production are effectively guaranteed.

[0090] Among them, the twin database is mainly composed of twin data, including two parts. The first part is data perception. The production-related equipment on the physical workshop site, such as CNC machine tools, AGV, RFID and other sensors, are connected to the OPC UA server through Ethernet or workshop field bus, and the status data of various production factors are collected in real time. In the twin model, based on the current operation data of the workshop, the multi-agent scheduling decision system will formulate a suitable scheduling plan and send these scheduling decision information to the OPC UA server. After that, the digital twin system converts the scheduling instructions into equipment operation commands, thereby promoting the production process of the physical workshop. The real-time data collection of various production factors in the physical workshop is achieved by deploying various sensor devices such as RFID and sensors on the workshop site, and connecting to various processing equipment and control devices such as CNC machine tools, PLC, robots, etc. in the production line, and connecting to the OPC UA server to realize the collection of multi-source heterogeneous data. The second part is data storage, using HBase for distributed data storage, the underlying HDFS storage architecture, and storing it in the MYSQL database according to the node configuration. The data stored here is not only used to monitor and update the digital twin model, but also provides key data support for identifying and detecting production anomalies.

[0091] It covers the necessary information data of the virtual workshop operation process, including static data, dynamic data, simulation operation data and historical production data of the physical workshop. Specifically, static data consists of information such as equipment processing capacity and workshop layout, which will not change during the long period of workshop operation; dynamic data contains real-time status information that characterizes the manufacturing elements of the workshop; simulation data is the data generated by the various hierarchical models in the virtual workshop during the simulation operation process, which can be used for learning and training of the multi-agent scheduling decision system; historical production data includes the historical status information of various manufacturing elements in the workshop.

[0092] In one embodiment, a dynamic optimization module is also included. In the dynamic optimization module, the scheduling strategy is dynamically adjusted based on real-time data analysis to optimize the production plan and resource allocation. When the digital twin system is running, the higher the accuracy of the system parameters, the more reliable the scheduling decision results. Before guiding the actual manufacturing workshop to make scheduling decisions, the scheduling plan needs to be simulated and verified in the twin model. The twin model that provides reliable decision results and the status information of the physical workshop iterate with each other to jointly promote the virtual and real evolution and update of the digital twin workshop. The update evolution mechanism of the workshop scheduling twin model is as follows: Fig.10 As shown in the figure, the process starts with real-time data, and after digital twin modeling and correction, simulation and interaction, scheduling information and product completion status are generated. Then a comparative analysis is performed to determine the consistency between PS and VS. If they are inconsistent, the disturbance factors are determined and the system data is corrected, and the twin model is updated; if they are consistent, the scheduling plan is planned. Then the plan is evaluated based on the VS simulation. If it does not meet the requirements, the scheduling plan is returned; if it meets the requirements, the scheduling plan is obtained.

[0093] Furthermore, in the update mechanism of digital twins, a state-based twin model update mechanism is adopted, which is mainly reflected in the adaptive update of state parameters. The specific update process is as follows: Fig.11 As shown in the figure, 17 attribute parameters representing the running status of the model are extracted from the database, and different attribute changes correspond to different update requirements. By comparing and analyzing the current status data of the workshop with the historical data in the database, the data attribute entries with problems are obtained and the causes of the problems are analyzed. First, the location of the comparison data is determined by the workpiece name and the completed process number. Then, the comparison data is corrected based on production attribute parameters such as the number of reported work, order batch, batch number and scrap number. Next, by analyzing the attribute parameters that represent the current execution of the system, such as the process table, equipment table, preparation time and equipment status, the changes in the workshop processing tasks are determined. Finally, the workpiece processing time is determined by analyzing different combinations of the remaining data.

[0094] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0095] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a digital twin system based on flexible job shop scheduling, characterized in that: The specific steps are: Building a twin model of the physical workshop according to the production data of the physical workshop; Building a workshop scheduling discrete event simulation model based on the twin model of the physical workshop; Based on the internal and external virtual-real interaction mechanism, the virtual-real mapping interaction between the twin model of the physical workshop and the physical workshop and the virtual-real mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model are realized to generate a virtual workshop; The data generated during the interaction is stored in the twin data pool in real time; A workshop service system is constructed based on three-dimensional visualization integration technology, so that the physical workshop and the virtual workshop can interact and feedback with each other, thus completing the construction of the scheduling digital twin system.

2. According to the method for constructing a digital twin system based on flexible job shop scheduling according to claim 1, it is characterized in that: The twin model of the physical workshop includes a production factor model, a production behavior logic model and a scheduling knowledge rule model, and its construction steps are as follows: Generate the production factor model based on the production data of the physical workshop using an ontology modeling method; Classifying manufacturing resources in the production factor model, assigning time attributes to the workshop state transitions of each duration in each category, generating multiple agents, and constructing the production behavior logic model; The scheduling knowledge rule model is constructed based on the interaction rules and scheduling rule knowledge of each Agent in the production behavior logic model.

3. The method for constructing a digital twin system based on flexible job shop scheduling according to claim 2 is characterized in that: The production behavior logic model is constructed by using timed colored Petri nets.

4. The method for constructing a digital twin system based on flexible job shop scheduling according to claim 2 is characterized in that: The ontology modeling method is expressed as: O = {C, A, R, I, M}; In the formula, C represents the basic class; A represents the attribute set of production factors; R represents the relationship between attributes; I represents the instance set; M represents the mapping relationship between production factor entities and concepts.

5. The method for constructing a digital twin system based on flexible job shop scheduling according to claim 2 is characterized in that: The internal and external virtual-real interaction mechanism includes an external virtual-real interaction mechanism and an internal virtual-real interaction mechanism. The external virtual-real interaction mechanism is a pairwise interaction between a physical workshop, a virtual workshop, a twin data pool, and a workshop service system. The internal virtual-real interaction mechanism includes: the production factor model interacts with the production behavior logic model, and the scheduling knowledge rule model interacts iteratively with the workshop scheduling discrete event simulation model.

6. The method for constructing a digital twin system based on flexible job shop scheduling according to claim 5 is characterized in that: The iterative interaction between the scheduling knowledge rule model and the workshop scheduling discrete event simulation model is realized based on the scheduling rule knowledge mining model of NGEP-FS and the scheduling decision Agent adaptive scheduling model.

7. The method for constructing a digital twin system based on flexible job shop scheduling according to claim 1 is characterized in that: The update mechanism of the digital twin system is: Adopting a state-based update strategy, the current state data of the workshop is compared with the historical data, the problematic data attribute entries are identified, and corresponding corrections are made.

8. A digital twin system construction system based on flexible job shop scheduling, characterized in that: include: A twin model building module, used to build a twin model of the physical workshop according to the production data of the physical workshop; A simulation model building module, used to build a workshop scheduling discrete event simulation model based on the twin model of the physical workshop; A virtual interaction module, which realizes virtual-reality mapping interaction between the twin model of the physical workshop and the physical workshop, and virtual-reality mapping interaction between the twin model of the physical workshop and the workshop scheduling discrete event simulation model based on the internal and external virtual-reality interaction mechanism, to generate a virtual workshop; The data generated during the interaction is stored in the twin data pool in real time; The system generation module builds a workshop service system based on three-dimensional visualization integration technology, enables the physical workshop and the virtual workshop to interact and feedback, and completes the construction of the scheduling digital twin system.

Images