Intelligent manufacturing real-time decision-making method and system based on digital twinning

Abstract

The invention discloses an intelligent manufacturing real-time decision-making method and system based on digital twinning, and relates to the technical field of intelligent manufacturing. Comprising the following steps: S1, collecting multi-source operation data in a physical production system in real time; s2, dynamically constructing and updating a digital twinborn model of the physical production system based on the multi-source operation data acquired in real time; and S3, based on the digital twin model, monitoring the operation state of the physical production system in real time. According to the method, the multi-source operation data of the physical production system are collected in real time, and the digital twin model is dynamically updated, so that no lag deviation between the model and the physical system is ensured; when a dynamic disturbance event is identified, event-driven real-time simulation and predictive analysis are started immediately, delay caused by traditional batch processing is avoided, and disturbance influence can be accurately evaluated in combination with historical data and expert experience; and then an optimal decision scheme is generated and screened in real time through a multi-objective optimization algorithm.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, specifically to a real-time decision-making method and system for intelligent manufacturing based on digital twins. Background Technology

[0002] As industrial manufacturing evolves towards greater intelligence, digital twin technology, as a key bridge connecting the physical world and cyberspace, has been widely applied to construct virtual production system models. This technology drives the synchronous operation and simulation of the virtual model by collecting real-time data on equipment status, material flow, and process parameters from the physical production line, thereby achieving dynamic mapping and visual monitoring of the production process. Based on this, the manufacturing system can utilize historical data and real-time information to analyze, predict, and optimize production activities, providing decision support for managers and improving overall operational efficiency.

[0003] However, existing decision support methods based on digital twins generally suffer from decision delays due to the mapping between the virtual and physical systems when facing dynamic disturbances. Specifically, when unexpected situations occur on the production floor, such as sudden equipment malfunctions, material supply interruptions, or instantaneous drift in process parameters, although the changes in the physical system's state can be collected in real time and mapped to the virtual twin model, there is a significant lag between the system sensing the change and generating a response strategy and feeding it back to the physical side for execution. This lag mainly stems from the fact that current decision models typically rely on periodic or batch-based data analysis and simulation, making it difficult to quickly assess the impact of disturbances and dynamically optimize multiple solutions within a very short time. The consequence is that by the time a decision instruction is issued, the actual state of the physical system may have further evolved, rendering the formulated decision inapplicable or even exacerbating production fluctuations, severely restricting the efficient and stable operation of intelligent manufacturing systems. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a real-time decision-making method and system for intelligent manufacturing based on digital twins, which solves the problem of decision delay due to virtual-real mapping when facing dynamic disturbance events in existing technologies.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a real-time decision-making method for intelligent manufacturing based on digital twins, comprising: S1. Real-time acquisition of multi-source operational data from the physical production system; S2. Based on the real-time collected multi-source operational data, dynamically construct and update the digital twin model of the physical production system; S3. Based on the digital twin model, monitor the operating status of the physical production system in real time to identify dynamic disturbance events in the physical production system; S4. Upon identifying a dynamic disturbance event, immediately perform real-time simulation and predictive analysis based on the digital twin model to quickly assess the potential impact of the dynamic disturbance event on the physical production system. S5. Based on the real-time simulation and prediction analysis results, dynamically generate and optimize multiple candidate decision-making strategies to deal with the dynamic disturbance events, and obtain the optimal real-time decision-making scheme. S6. Send the optimal real-time decision-making scheme instruction to the physical production system in real time to drive the physical production system to respond quickly and execute the optimal real-time decision-making scheme.

[0006] Furthermore, the real-time acquisition of multi-source operational data from the physical production system includes: By deploying production equipment and sensor networks in the physical production system, real-time operating status data of the production equipment is acquired, including equipment load, energy consumption and fault diagnosis information, as well as process parameter data, including temperature, pressure, flow rate and product quality indicators. By establishing a data interface with the material management system of the physical production system, the flow information data of the materials can be obtained in real time, including material inventory levels, batch information, and flow location.

[0007] Furthermore, the dynamic construction and updating of the digital twin model of the physical production system includes: Construct a three-dimensional geometric model, a topological model, and a behavioral model of the physical production system to form the physical layer and model layer of the digital twin model; Establish data communication links and data interaction protocols between various devices and systems within the physical production system to form the network layer of the digital twin model; The real-time monitoring, fault diagnosis, performance prediction, and decision optimization service interfaces provided by the digital twin model are encapsulated to form the service layer of the digital twin model.

[0008] Furthermore, identifying dynamic disturbance events in the physical production system includes: Define the key performance indicators of the physical production system and set dynamic thresholds for the key performance indicators; The real-time values ​​of the key performance indicators are compared with the dynamic thresholds; When the real-time value of any of the key performance indicators exceeds the dynamic threshold, the dynamic disturbance event is determined to have occurred. The dynamic disturbance event includes sudden equipment malfunction, material supply interruption, or instantaneous drift of process parameters.

[0009] Furthermore, the rapid assessment of the potential impact of the dynamic disturbance event on the physical production system includes: Based on the pre-defined disturbance propagation path and impact mechanism in the digital twin model, the diffusion simulation of the disturbance event in the spatiotemporal dimension is performed to simulate the evolution of the disturbance event in the physical production system. By combining historical disturbance datasets accumulated in the physical production system with expert experience knowledge bases, the evolution trend and final impact of the disturbance events are predicted and analyzed. Quantify the direct and indirect potential losses of disturbance events to production efficiency, product quality, equipment utilization, material consumption, and energy efficiency.

[0010] Furthermore, the dynamic generation and optimization of multiple candidate decision strategies to respond to the dynamic disturbance events includes: Based on the type of the dynamic disturbance event and the assessed impact, candidate decision-making strategies for adjusting production plans, equipment scheduling, or process parameters are selected from a predefined strategy template library or intelligently generated. For each candidate decision strategy, a rapid simulation is performed in the digital twin model to evaluate its performance in restoring production, reducing losses, and optimizing resource utilization. By using multi-objective optimization algorithms, heuristic algorithms, or reinforcement learning-based decision models, the multiple candidate decision strategies are dynamically optimized in real time to select the real-time decision scheme with the best performance.

[0011] Furthermore, the step of sending the instruction for the optimal real-time decision-making scheme to the physical production system in real time includes: Through the Industrial Internet of Things (IIoT) gateway, the instructions for the optimal real-time decision scheme are converted into control signals that can be recognized by the programmable logic controller, robot control system, or actuator in the physical production system. The control signals are transmitted via low-latency industrial Ethernet, wireless LAN, 5G network, or dedicated industrial bus. The system monitors in real time the execution of the optimal real-time decision-making scheme instructions by the physical production system and sends the execution result data back to the digital twin model for verification and evaluation of the decision-making effect.

[0012] Furthermore, the method also includes: Based on the execution result data and effect evaluation of the optimal real-time decision-making scheme, update the historical disturbance dataset and expert experience knowledge base; By utilizing the updated historical disturbance dataset and expert experience knowledge base, the identification threshold, simulation parameters, and decision-making strategy optimization model of the dynamic disturbance event are adaptively adjusted.

[0013] Furthermore, the real-time simulation and predictive analysis includes: An event-driven discrete event simulator or continuous system simulator for constructing the digital twin model is used to immediately start high-fidelity simulation when dynamic disturbance events occur, avoiding periodic batch processing. In the simulator, refined simulation boundary conditions and dynamic evolution rules are set according to the type, initial state and evolution law of the dynamic disturbance event, and high-performance parallel computing technology is used to quickly complete the deduction of the impact of different disturbance scenarios and the prediction of future states.

[0014] The present invention also provides a real-time decision-making system for intelligent manufacturing based on digital twins, applied to any of the above-described real-time decision-making methods for intelligent manufacturing based on digital twins, comprising: The data acquisition module is used to collect multi-source operational data from the physical production system in real time. A digital twin modeling module is used to dynamically construct and update a digital twin model of the physical production system based on the real-time collected multi-source operational data. A disturbance identification module is used to monitor the operating status of the physical production system in real time based on the digital twin model, so as to identify dynamic disturbance events in the physical production system. The impact assessment module is used to immediately perform real-time simulation and predictive analysis based on the digital twin model after identifying dynamic disturbance events, so as to quickly assess the potential impact of the dynamic disturbance events on the physical production system. The decision optimization module is used to dynamically generate and optimize multiple candidate decision strategies to deal with the dynamic disturbance event based on the real-time simulation and predictive analysis results, so as to obtain the optimal real-time decision scheme. The instruction issuing module is used to issue the instruction of the optimal real-time decision scheme to the physical production system in real time, so as to drive the physical production system to respond to and execute the optimal real-time decision scheme and report back the execution status.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention ensures no lag or deviation between the model and the physical system by real-time acquisition of multi-source operational data from the physical production system and dynamic updates to the digital twin model. Upon identification of a dynamic disturbance event, it immediately initiates event-driven real-time simulation and predictive analysis, avoiding the delays associated with traditional batch processing. It also combines historical data and expert experience to accurately assess the impact of the disturbance. Subsequently, a multi-objective optimization algorithm generates and selects the optimal decision-making scheme in real time, and the instructions are transmitted to the physical system for execution with low latency, effectively solving the decision-making delay problem in existing technologies. Simultaneously, the system can update its knowledge base based on the decision execution results, adaptively adjusting the disturbance identification threshold, simulation parameters, and optimization model to improve decision accuracy and continuous adaptability, ensuring the stable and efficient operation of the intelligent manufacturing system under disturbances. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart illustrating the simulation and decision optimization process of the present invention. Figure 3 This is a system structure diagram of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1-2 This invention provides a real-time decision-making method for intelligent manufacturing based on digital twins, including: S1. Real-time acquisition of multi-source operational data from the physical production system; S2. Based on real-time collected multi-source operational data, dynamically construct and update a digital twin model of the physical production system; S3. Based on the digital twin model, monitor the operating status of the physical production system in real time to identify dynamic disturbance events in the physical production system. S4. After identifying a dynamic disturbance event, immediately perform real-time simulation and predictive analysis based on the digital twin model to quickly assess the potential impact of the dynamic disturbance event on the physical production system. S5. Based on the results of real-time simulation and predictive analysis, dynamically generate and optimize multiple candidate decision-making strategies to deal with dynamic disturbance events, and obtain the optimal real-time decision-making scheme. S6. Send the optimal real-time decision-making scheme instruction to the physical production system in real time to drive the physical production system to respond quickly and execute the optimal real-time decision-making scheme.

[0019] Specifically, this implementation achieves rapid response through a real-time design across the entire process. First, real-time acquisition of multi-source operational data from the physical production system is performed. A distributed sensor network is deployed at key parts of the production equipment, and a real-time data interface is established with the material management system to ensure that equipment operating status data and material flow information are collected at a rate of seconds, providing a real-time data foundation for subsequent steps. Based on this real-time acquired data, a digital twin model is dynamically constructed and updated. During model construction, physical system data is synchronized in real time, and equipment parameters and process logic in the model are continuously adjusted to avoid lag or deviation between the model and the physical system. Subsequently, the operating status of the physical production system is monitored in real time based on this digital twin model. The model continuously maps various indicators of the physical system, and once an abnormality is detected, dynamic disturbance events are immediately identified without waiting for periodic detection. Upon identification of a disturbance event, the digital twin model immediately initiates real-time simulation and predictive analysis, using an event-driven model instead of traditional batch processing. The simulation process begins immediately after the disturbance occurs, quickly simulating the impact of the disturbance on the physical system. Based on the simulation and prediction results, the system dynamically generates multiple candidate decision strategies, evaluates the effectiveness of each strategy through rapid deduction, and then uses optimization algorithms to find the optimal solution in real time. Finally, the optimal solution instruction is sent to the physical production system via low-latency transmission, ensuring that the instruction arrives quickly and drives the equipment to execute. The entire process significantly shortens the time from disturbance identification to decision execution, effectively solves the decision delay problem, and ensures that the physical production system can still operate stably and efficiently under disturbances.

[0020] In this embodiment, multi-source operational data from the physical production system are collected in real time, including: By deploying production equipment and sensor networks in the physical production system, real-time operating status data of production equipment is acquired, including equipment load, energy consumption and fault diagnosis information, as well as process parameter data, including temperature, pressure, flow rate and product quality indicators. By establishing a data interface with the material management system of the physical production system, real-time information on material flow can be obtained, including material inventory levels, batch information, and flow location.

[0021] Specifically, when collecting multi-source operational data from the physical production system in real time, the first step is to deploy suitable sensors on the production equipment. For operational status data such as equipment load and energy consumption, current sensors, power sensors, and fault diagnosis sensors are installed on key equipment such as motors and machine tools to acquire real-time load changes, energy consumption fluctuations, and potential fault signals during equipment operation. For process parameter data, temperature sensors, pressure sensors, and flow sensors are installed at locations such as reaction vessels and conveying pipelines on the production line. Simultaneously, product quality indicators such as product size and purity are collected through product quality inspection equipment. Furthermore, a standardized data interface is used to connect with the material management system of the physical production system to read inventory level data from the material inventory database in real time, track production batch information for each batch of materials, and obtain the flow location information of materials within the production workshop through a material positioning system.

[0022] In this embodiment, dynamically constructing and updating the digital twin model of the physical production system includes: Construct a three-dimensional geometric model, topological model, and behavioral model of the physical production system to form the physical layer and model layer of the digital twin model; Establish data communication links and data interaction protocols between various devices and systems within the physical production system to form the network layer of the digital twin model; The service layer of the digital twin model is formed by encapsulating the real-time monitoring, fault diagnosis, performance prediction, and decision optimization service interfaces provided by the digital twin model.

[0023] Specifically, when dynamically constructing and updating the digital twin model of a physical production system, the physical layer and model layer are built first. 3D modeling software is used to recreate the equipment layout and structural dimensions of the physical production system, forming a 3D geometric model. Then, the connections between equipment and the production process are analyzed to establish a topology model. Simultaneously, a behavioral model that simulates equipment operation is constructed by combining equipment operating patterns and process parameter changes. Next, a network layer is established. Based on the communication requirements of the equipment in the physical production system, suitable communication protocols such as industrial Ethernet and wireless LAN are selected to build data communication links between equipment and the system, and between the system and the model, ensuring efficient data transmission at each stage. Finally, a service layer is encapsulated. The real-time monitoring, fault diagnosis, performance prediction, and decision optimization functions of the digital twin model are packaged into standardized service interfaces. These interfaces can be directly called by subsequent modules such as disturbance identification and impact assessment, enabling the digital twin model not only to achieve real-time mapping of the physical system but also to provide direct service support for subsequent decision-making processes, improving the efficiency of the entire decision-making process.

[0024] In this embodiment, identifying dynamic disturbance events in the physical production system includes: Define the key performance indicators of the physical production system and set dynamic thresholds for the key performance indicators. Compare the real-time values ​​of key performance indicators with dynamic thresholds; When the real-time value of any key performance indicator exceeds the dynamic threshold, a dynamic disturbance event is determined to have occurred. Dynamic disturbance events include sudden equipment malfunctions, material supply interruptions, or instantaneous drift of process parameters.

[0025] Specifically, when identifying dynamic disturbances in a physical production system, key performance indicators (KPIs) are first defined based on production goals and system characteristics. These KPIs cover dimensions such as equipment operation, material supply, and process parameters, including equipment failure rate, material turnover rate, and stable process temperature range. Then, based on historical operating data of the physical production system, production process requirements, and industry standards, dynamic thresholds are set for each KPI. These thresholds are automatically adjusted according to changes in production conditions; for example, the thresholds for some equipment load indicators are appropriately relaxed during peak production periods. Subsequently, real-time collected KPI data is continuously compared with the dynamic thresholds. When the real-time value of equipment load suddenly exceeds the currently set dynamic threshold, a sudden equipment anomaly is identified; when the real-time value of material inventory level falls below the dynamic threshold, a material supply interruption is identified; when the real-time values ​​of parameters such as process temperature and pressure momentarily deviate from the dynamic threshold range, a momentary drift of process parameters is identified. This method enables timely and accurate identification of various dynamic disturbances, preventing disturbances from causing greater impact on the production system.

[0026] In this embodiment, the rapid assessment of the potential impact of dynamic disturbance events on the physical production system includes: Based on the pre-defined disturbance propagation path and impact mechanism in the digital twin model, the diffusion simulation of disturbance events in the spatiotemporal dimension is carried out to simulate the evolution of disturbance events in the physical production system. By combining historical disturbance datasets accumulated in the physical production system with expert experience knowledge bases, we can predict and analyze the evolution trend and final impact of disturbance events. Quantify the direct and indirect potential losses of disturbance events to production efficiency, product quality, equipment utilization, material consumption, and energy efficiency.

[0027] Specifically, when quickly assessing the potential impact of dynamic disturbance events on the physical production system, the propagation paths and impact mechanisms of different types of disturbances are first preset in the digital twin model. For example, sudden equipment failures can affect the operation of upstream and downstream equipment along the production line, and material supply interruptions can cause the processes that depend on the material to stagnate. After the disturbance event occurs, the model will perform a spatiotemporal diffusion simulation based on these preset information to simulate the evolution of the disturbance at different time points and different production links. At the same time, the historical disturbance dataset accumulated by the physical production system is called to find cases similar to the current disturbance. Combined with the judgment rules and coping experience in the expert experience knowledge base, the evolution trend and final impact of the disturbance event are predicted and analyzed. Finally, the direct and indirect potential losses that the disturbance event may cause are quantified from five dimensions: production efficiency, product quality, equipment utilization rate, material consumption and energy efficiency. Here, the multi-dimensional loss quantification formula is used to calculate the total potential loss. Formula (1) is as follows: (1) In the formula, The total potential loss caused by the disturbance event; The number of dimensions for loss assessment is set to 5 here; For the first The weight coefficients of each evaluation dimension are determined using the entropy weight method, an existing technology that assigns weights by analyzing the dispersion of the loss data for each dimension; the higher the dispersion, the greater the weight. For the first The normalized loss values ​​for each evaluation dimension need to be normalized using min-max to convert the original loss values ​​to the [0,1] interval, since the original loss dimensions are different. This ensures that the loss of each dimension can be directly used in the calculation. This formula can comprehensively reflect the overall impact of disturbances on the production system and provide an accurate basis for subsequent decision-making strategies.

[0028] In this embodiment, multiple candidate decision strategies for responding to dynamic disturbance events are dynamically generated and optimized, including: Based on the type of dynamic disturbance event and the assessed impact, candidate decision-making strategies for adjusting production plans, equipment scheduling, or process parameters are selected from a predefined strategy template library or intelligently generated. For each candidate decision strategy, perform rapid simulations in the digital twin model to evaluate its performance in restoring production, reducing losses, and optimizing resource utilization; By utilizing multi-objective optimization algorithms, heuristic algorithms, or reinforcement learning-based decision models, multiple candidate decision strategies are dynamically optimized in real time to select the most efficient real-time decision scheme.

[0029] Specifically, when dynamically generating and optimizing multiple candidate decision strategies to deal with dynamic disturbance events, firstly, based on the type of dynamic disturbance event and the assessed scope and degree of impact, suitable candidate decision strategies are selected from a predefined strategy template library. For example, when equipment suddenly malfunctions, strategies such as equipment scheduling and activation of backup equipment are selected; when material supply is interrupted, strategies such as production plan adjustment and use of alternative materials are selected. If there is no perfectly matching strategy in the template library, new candidate strategies are intelligently generated based on the disturbance characteristics and the current status of the production system. For each candidate decision strategy, the execution process is simulated in a digital twin model to deduce the recovery status of the production system, the loss reduction effect, and the resource utilization efficiency after the strategy is implemented. Then, a multi-objective optimization algorithm is used to dynamically optimize multiple candidate decision strategies in real time. Here, a weighted summation method is used to construct the multi-objective optimization function, and the specific formula is shown in formula (2): (2) In the formula, This represents the overall effectiveness value of the candidate decision-making strategies; To optimize the target quantity, three objectives are included here: production recovery speed, loss control effectiveness, and resource utilization efficiency. For the first The weight coefficients of each optimization objective are determined using the analytic hierarchy process (AHP), an existing technology that obtains the weights by constructing a judgment matrix and calculating the weight vector. For the first The normalized performance values ​​of each optimization objective need to be normalized to the [0,1] interval using min-max normalization, since the original performance values ​​of each objective have different dimensions, to ensure consistency of computational dimensions. This is achieved by calculating the performance of each candidate strategy. Value, selection The strategy with the highest value is selected as the optimal real-time decision-making scheme, ensuring that the decision-making scheme can minimize the impact of disturbances.

[0030] In this embodiment, the instruction for the optimal real-time decision-making scheme is sent to the physical production system in real time, including: Through industrial IoT gateways, the instructions for optimal real-time decision-making schemes are converted into control signals that can be recognized by programmable logic controllers, robot control systems, or actuators in the physical production system. Control signals are transmitted via low-latency industrial Ethernet, wireless LAN, 5G network, or dedicated industrial bus. The system monitors the execution of the optimal real-time decision-making scheme instructions by the physical production system in real time, and transmits the execution results back to the digital twin model for verification and evaluation of the decision-making effect.

[0031] Specifically, when the optimal real-time decision-making scheme is sent to the physical production system, the instruction is first formatted via an industrial IoT gateway. This converts the parameter requirements and operational instructions in the decision-making scheme into digital control signals recognizable by the programmable logic controllers, robot control systems, or actuators in the physical production system. The converted control signals are then transmitted using a low-latency transmission method, selecting industrial Ethernet, wireless LAN, 5G network, or dedicated industrial bus based on the production workshop's network environment to ensure low latency and high stability during transmission. Simultaneously, the execution of instructions by the physical production system is monitored in real-time. Sensors and data acquisition devices acquire data such as the equipment's execution status and process parameter adjustment results. This execution result data is then fed back to the digital twin model and compared with the model's simulation results to verify the effectiveness of the decision-making scheme. If any deviations are detected, adjustments are made promptly to ensure the accurate implementation of the decision-making scheme.

[0032] In this embodiment, the method further includes: Based on the execution results data and effect evaluation of the optimal real-time decision-making scheme, update the historical perturbation dataset and expert experience knowledge base; By utilizing the updated historical perturbation dataset and expert experience knowledge base, the identification threshold, simulation parameters, and decision-making strategy optimization model for dynamic perturbation events are adaptively adjusted.

[0033] Specifically, after the optimal real-time decision-making scheme is executed, various data during the execution process are collected, including execution result data such as production recovery time, actual loss amount, and resource consumption, as well as the decision effect evaluation conclusions derived from these data. This information is entered into the historical disturbance dataset, and the strategy optimization points and disturbance identification improvement directions discovered during the evaluation process are added to the expert experience knowledge base to achieve dynamic updates of the knowledge base. Using the updated historical disturbance dataset and expert experience knowledge base, the system will adaptively adjust the identification threshold of dynamic disturbance events. Here, the threshold adaptive adjustment formula is used, as shown in formula (3): (3) In the formula, The new threshold after adjustment; The old threshold before adjustment; This is the threshold adjustment coefficient, with a value range of [0,1]. It is determined through feedback calibration of historical threshold adjustment effects to ensure that the adjustment range is reasonable. This represents the actual error in identifying the current disturbance event, which is the difference between the actual time of the disturbance and the system's identification time. The allowable identification error threshold is set based on the timeliness requirements of the production system's response to disturbances. Simultaneously, the simulation parameters and the weighting coefficients of the decision-making strategy optimization model are adjusted to make the optimization process more closely match current actual production needs, enabling the entire decision-making system to have continuous optimization capabilities.

[0034] In this embodiment, real-time simulation and predictive analysis includes: Event-driven discrete event simulators or continuous system simulators that build digital twin models are used to immediately start high-fidelity simulations when dynamic disturbance events occur, avoiding periodic batch processing. In the simulator, based on the type, initial state, and evolution law of dynamic disturbance events, refined simulation boundary conditions and dynamic evolution rules are set, and high-performance parallel computing technology is used to quickly complete the deduction of the impact of different disturbance scenarios and the prediction of future states.

[0035] Specifically, in real-time simulation and predictive analysis, an event-driven discrete event simulator or continuous system simulator is first built for the digital twin model. This simulator does not rely on periodic trigger signals and starts immediately when a dynamic disturbance event is detected, avoiding the time delay caused by traditional batch processing and ensuring that the simulation process is synchronized with the disturbance event. After the simulator starts, refined simulation boundary conditions are set according to the specific type of dynamic disturbance event, such as the location of equipment failure, the batch of material interruption, and the initial state and evolution law of the disturbance. The production links and time range covered by the simulation are clearly defined, and dynamic evolution rules are defined, such as the propagation rate of the disturbance between equipment and the degree of influence on process parameters. Subsequently, high-performance parallel computing technology is used to simulate and extrapolate multiple possible disturbance development scenarios simultaneously, significantly shortening the simulation time. During the simulation, the quantification of the potential impact of the disturbance is calculated using Formula 1 to calculate the total potential loss, quickly completing the predictive analysis of the impact range, evolution trend, and future state of the disturbance event, providing timely simulation support for subsequent decision-making.

[0036] Please see Figure 3 The present invention also provides a real-time decision-making system for intelligent manufacturing based on digital twins, comprising: The data acquisition module is used to collect multi-source operational data from the physical production system in real time. The digital twin modeling module is used to dynamically build and update a digital twin model of the physical production system based on real-time collected multi-source operational data. The disturbance identification module is used to monitor the operating status of the physical production system in real time based on the digital twin model in order to identify dynamic disturbance events in the physical production system. The impact assessment module is used to perform real-time simulation and predictive analysis based on a digital twin model immediately after a dynamic disturbance event is identified, so as to quickly assess the potential impact of the dynamic disturbance event on the physical production system. The decision optimization module is used to dynamically generate and optimize multiple candidate decision strategies to deal with dynamic disturbance events based on real-time simulation and predictive analysis results, so as to obtain the optimal real-time decision scheme. The instruction issuing module is used to issue the instruction of the optimal real-time decision scheme to the physical production system in real time, so as to drive the physical production system to respond to and execute the optimal real-time decision scheme and report back the execution status.

[0037] Specifically, in practical applications, the data acquisition module collects real-time operating status data and process parameter data of production equipment through a distributed sensor network deployed in various stages of the physical production system. Simultaneously, it connects to the material management system via a standardized data interface to obtain material flow information, achieving comprehensive acquisition and real-time transmission of multi-source data. The digital twin modeling module receives data transmitted from the data acquisition module and constructs a three-dimensional geometric model, topological model, and behavioral model of the physical production system based on this data, forming the physical and model layers of the model. It also establishes communication links and interaction protocols between equipment and the system to construct the network layer, encapsulates service interfaces to form the service layer, and dynamically updates the parameters of each layer of the model based on real-time data. The disturbance identification module reads the model operation data output by the digital twin modeling module in real time, compares the real-time values ​​of key performance indicators with preset thresholds, identifies dynamic disturbance events, and sends them to the impact assessment module. After receiving the disturbance identification signal, the impact assessment module calls the simulation function of the digital twin model to conduct disturbance diffusion simulation and trend prediction, quantifies the potential impact using Formula 1, and feeds it back to the decision optimization module. Based on the impact assessment results, the decision optimization module selects or generates candidate strategies from the strategy template library. After model deduction, it calculates the comprehensive effectiveness value of each strategy using Formula 2, determines the optimal solution, and sends it to the instruction issuance module. The instruction issuance module converts the optimal solution into a control signal, transmits it to the physical system through a low-latency network, monitors the execution status, and sends data back to the digital twin model. All modules work together to achieve real-time decision-making throughout the entire process.

[0038] In summary, this invention ensures no lag or deviation between the model and the physical system by real-time acquisition of multi-source operational data from the physical production system and dynamic updates to the digital twin model. Upon identifying a dynamic disturbance event, it immediately initiates event-driven real-time simulation and predictive analysis, avoiding the delays associated with traditional batch processing. Furthermore, it combines historical data and expert experience to accurately assess the impact of the disturbance. Subsequently, a multi-objective optimization algorithm generates and selects the optimal decision-making scheme in real time, and the instructions are transmitted to the physical system for execution with low latency, effectively solving the problem of decision-making delays in existing technologies. Simultaneously, the system can update its knowledge base based on the decision execution results, adaptively adjusting the disturbance identification threshold, simulation parameters, and optimization model to improve decision-making accuracy and continuous adaptability, ensuring the stable and efficient operation of the intelligent manufacturing system under disturbances.

[0039] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0040] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A digital-twin-based intelligent manufacturing real-time decision-making method, characterized in that, The method comprises the following steps: S1, real-time collection of multi-source operation data in a physical production system; S2, dynamic construction and update of a digital twin model of the physical production system based on the real-time collected multi-source operation data; S3, real-time monitoring of the operation state of the physical production system based on the digital twin model to identify a dynamic disturbance event in the physical production system; S4, after the dynamic disturbance event is identified, real-time simulation and prediction analysis based on the digital twin model are performed to quickly evaluate the potential impact of the dynamic disturbance event on the physical production system; S5, dynamic generation and optimization of multiple candidate decision strategies for responding to the dynamic disturbance event according to the results of the real-time simulation and prediction analysis to obtain an optimal real-time decision scheme; S6, real-time delivery of the optimal real-time decision scheme instruction to the physical production system to drive the physical production system to quickly respond and execute the optimal real-time decision scheme. 2.The digital-twin-based intelligent manufacturing real-time decision method of claim 1, wherein, The real-time collection of multi-source operation data in the physical production system comprises: real-time acquisition of the operation state data of the production equipment, including equipment load, energy consumption and fault diagnosis information, and process parameter data, including temperature, pressure, flow and product quality indicators, through the deployment of the production equipment and the sensor network in the physical production system; real-time acquisition of the flow information data of the materials, including material inventory level, batch information and flow location, through data interface with the material management system of the physical production system. 3.The digital-twin-based intelligent manufacturing real-time decision method of claim 1, wherein, The dynamic construction and update of the digital twin model of the physical production system comprises: construction of a three-dimensional geometric model, a topological structure model and a behavior model of the physical production system to form a physical layer and a model layer of the digital twin model; establishment of data communication links and data interaction protocols between each device and system in the physical production system to form a network layer of the digital twin model; encapsulation of real-time monitoring, fault diagnosis, performance prediction and decision optimization service interfaces provided by the digital twin model to form a service layer of the digital twin model. 4.The digital-twin-based intelligent manufacturing real-time decision method of claim 1, wherein, The identification of the dynamic disturbance event in the physical production system comprises: definition of key performance indicators of the physical production system, including equipment operation dimension indicators, material flow dimension indicators, process parameter dimension indicators and production efficiency dimension indicators, and setting of dynamic thresholds for the key performance indicators; comparison of real-time values of the key performance indicators with the dynamic thresholds; when the real-time value of any key performance indicator exceeds the dynamic threshold, it is determined that the dynamic disturbance event occurs, which includes device sudden abnormality, material supply interruption or process parameter instantaneous drift.

5. The digital-twin-based intelligent manufacturing real-time decision method according to claim 1, characterized in that, The quick evaluation of the potential impact of the dynamic disturbance event on the physical production system comprises: diffusion simulation of the disturbance event in the time and space dimensions based on the preset disturbance propagation path and influence mechanism in the digital twin model to simulate the evolution of the disturbance event in the physical production system; prediction analysis of the evolution trend and final impact degree of the disturbance event in combination with the historical disturbance data set accumulated by the physical production system and the expert experience knowledge base; Quantify the direct and indirect potential losses of production efficiency, product quality, equipment uptime, material consumption, and energy efficiency caused by the dynamic disturbance event.

6. The digital-twin-based intelligent manufacturing real-time decision method according to claim 1, characterized in that, The dynamic generation and optimization of multiple candidate decision-making strategies to cope with the dynamic disturbance event include: According to the type and evaluation impact of the dynamic disturbance event, candidate decision-making strategies for adjusting production plans, equipment scheduling, or process parameters are screened or intelligently generated from a pre-defined strategy template library; For each candidate decision-making strategy, a quick deduction is performed in the digital twin model to evaluate its performance in restoring production, reducing losses, and optimizing resource utilization; Using multi-objective optimization algorithms, heuristic algorithms, or reinforcement learning-based decision-making models, the multiple candidate decision-making strategies are dynamically optimized in real time to select the optimal real-time decision-making scheme.

7. The digital-twin-based intelligent manufacturing real-time decision method according to claim 1, characterized in that, The optimal real-time decision-making scheme instructions are real-time issued to the physical production system, including: Through an industrial Internet gateway, the optimal real-time decision-making scheme instructions are converted into control signals recognizable by programmable logic controllers, robot control systems, or actuators in the physical production system; The control signals are transmitted through low-latency industrial Ethernet, wireless local area network, 5G network, or dedicated industrial bus; Real-time monitoring of the physical production system's execution of the optimal real-time decision-making scheme instructions, and the execution result data is returned to the digital twin model for verification and decision-making effect evaluation. 8.The digital-twin-based intelligent manufacturing real-time decision method of claim 1, wherein, The method further includes: Based on the execution result data and effect evaluation of the optimal real-time decision-making scheme, the historical disturbance data set and expert experience knowledge base are updated; Using the updated historical disturbance data set and expert experience knowledge base, the identification threshold, simulation deduction parameters, and decision-making strategy optimization model of the dynamic disturbance event are adaptively adjusted. 9.The digital-twin-based intelligent manufacturing real-time decision method of claim 1, wherein, The real-time simulation and prediction analysis includes: Building an event-driven discrete event simulator or continuous system simulator for the digital twin model to start high-fidelity simulation immediately when a dynamic disturbance event occurs, avoiding periodic batch processing; In the simulator, according to the type, initial state, and evolution law of the dynamic disturbance event, set fine simulation boundary conditions and dynamic evolution rules, and use high-performance parallel computing technology to quickly complete the impact deduction and future state prediction of different disturbance scenarios.

10. The intelligent manufacturing real-time decision system based on digital twinning, applied to the intelligent manufacturing real-time decision method based on digital twinning of any one of claims 1-9, characterized in that, Including: A data acquisition module for real-time acquisition of multi-source operation data in the physical production system; A digital twin modeling module for dynamically building and updating the digital twin model of the physical production system based on the real-time acquired multi-source operation data; A disturbance identification module for real-time monitoring of the running state of the physical production system based on the digital twin model to identify dynamic disturbance events in the physical production system; An impact assessment module for immediately performing real-time simulation and prediction analysis based on the digital twin model after identifying a dynamic disturbance event to quickly evaluate the potential impact of the dynamic disturbance event on the physical production system; a decision optimization module configured to dynamically generate and optimize a plurality of candidate decision strategies for coping with the dynamic disturbance event according to the real-time simulation and the prediction analysis result, to obtain an optimal real-time decision scheme; an instruction issuing module configured to issue an optimal real-time decision scheme instruction to the physical production system in real time, so as to drive the physical production system to respond to and execute the optimal real-time decision scheme and return an execution state.

Images