Battery pack production line multi-end monitoring system based on digital twinning

Abstract

The application provides a battery pack production line multi-end monitoring system based on digital twinning, which comprises a twinning data processing hub and a visual interactive terminal in communication with each other; the twinning data processing hub is used for collecting multi-source heterogeneous equipment data from the battery pack production line, and establishing a mapping relationship between physical equipment and virtual nodes in a digital twinning model; the visual interactive terminal is used for driving visual presentation of the digital twinning model, and receiving operation instructions of a user for the physical production line; the twinning data processing hub further comprises a multi-user dynamic management unit, which is used for providing data isolated access services for the visual interactive terminals, constructing a three-dimensional permission matrix with roles, scenes and tasks as dimensions, and realizing multi-user dynamic fine-grained permission control; the visual interactive terminal further comprises a virtual-real interactive control unit and a multi-mode simulation unit, and realizes multi-mode and multi-rate simulation based on production line data or user input parameters.

Description

Technical Field

[0001] This invention belongs to the field of industrial intelligent manufacturing and digital twin technology, specifically relating to a multi-terminal monitoring system for battery pack production lines based on digital twins. Background Technology

[0002] Battery pack production lines are a crucial link connecting battery cell manufacturing and end-use applications, and their production quality and efficiency directly affect the reliability of energy supply. Currently, battery pack production lines are showing a trend towards multi-category and flexible production; however, traditional production line monitoring systems generally suffer from weak data integration, rigid simulation debugging, and poor system scalability. These inherent defects have become a major obstacle to industrial upgrading, severely restricting the process of transparent management and intelligent upgrading of production lines.

[0003] Digital twin technology, by constructing a high-precision virtual mapping of physical entities and establishing a real-time data link between physical and virtual spaces, provides core technical support for solving traditional monitoring challenges. Based on this technology, multi-source heterogeneous data can be integrated into a unified virtual platform, enabling visualized presentation of production status; through a virtual-physical collaboration mechanism, interference-free debugging can be carried out in virtual space, significantly reducing the trial-and-error costs of physical production lines; its flexibility allows for more precise matching of the production needs of multiple battery pack types, providing a feasible path for the intelligent upgrading of production lines.

[0004] Chinese patent CN121455576A discloses a visualization and multi-terminal collaborative control method and system for automotive welding workshops. It achieves multi-terminal collaborative control by constructing a 3D scene database with varying geometric precision, solving technical problems such as the contradiction between rendering performance and realism, inconsistent multi-terminal operations, and the separation of real-time control and visualization. However, it lacks a refined permission division and data isolation mechanism for different roles, failing to meet the differentiated data needs and security isolation requirements of different departments. Chinese patent CN121392160A discloses a production line debugging method and system, which designs the operating logic and process parameters of the physical entities of the production line in advance in a virtual model. Controlling the physical entity of the production line through a virtual model significantly reduces debugging time, but it only supports debugging mode and cannot flexibly switch between multiple working modes such as production line monitoring, offline debugging, and online control. Chinese patent CN121523274A discloses a control system and method for a tungsten wire copper plating production line based on digital twins. It realizes real-time display of virtual and real mapping screens, quality prediction curves, alarm information, and system recommended operation instructions through multi-source sensing and real-time digital twin engine. However, its simulation function only revolves around real-time quality prediction and lacks support for multi-scenario simulation modes that are adapted to the entire life cycle and flexible production of battery pack production lines, making it difficult to adapt to the needs of flexible production.

[0005] Specifically, existing technologies suffer from the following three main shortcomings: First, the lack of a multi-user access mechanism means that different departments, such as production management, equipment maintenance, and process optimization, need to share production line data. However, the system lacks permission division and isolation design, which not only fails to meet the differentiated data needs of each department but also poses security risks such as data leakage and unauthorized operation, making it difficult to balance the core requirements of data sharing and security isolation. Second, insufficient virtual-real interaction and scenario adaptability mean that while some systems have achieved visualization of virtual models, they lack stable two-way interaction capabilities. Control commands issued in the virtual space cannot be accurately and in real time sent to the physical production line, nor can they achieve two-way real-time synchronous feedback of the physical production line's operating status. Third, limited simulation functionality means that it cannot flexibly switch between scenarios such as real-time monitoring, process parameter optimization, fault simulation, and production line debugging. Furthermore, the adjustment of simulation parameters is cumbersome and difficult to quickly adapt to the production needs of different battery pack specifications, resulting in the failure to fully realize the technological value of digital twins. Summary of the Invention

[0006] To address the shortcomings and deficiencies of existing technologies, this invention provides a multi-terminal monitoring system for battery pack production lines based on digital twins. This system employs a dual-core architecture combining a digital twin data processing hub and a visual interactive terminal, overcoming the core pain points of traditional battery pack production line monitoring systems, such as insufficient multi-user collaboration capabilities, poor security in virtual-physical interaction, and rigid and limited simulation functions. The digital twin data processing hub is compatible with multiple industrial communication protocols, enabling real-time acquisition and stable transmission of data from heterogeneous equipment throughout the production line. It establishes a precise mapping between physical devices and virtual nodes, and uses a time-series database to achieve persistent storage and value-based hierarchical management of production line data. Simultaneously, it constructs a three-dimensional permission matrix based on roles, scenarios, and tasks, combined with a triple isolation mechanism for data, permissions, and access, providing dedicated access services with data isolation and refined dynamic permission control for multiple terminals. The visual interactive terminal enables multi-precision dynamic rendering of digital twin models and physical field visualization of core battery pack processes. It incorporates a four-level verification process (syntax validation, semantic validation, process interlock validation, and security review) and a confidence level transmission risk assessment mechanism, achieving high-security two-way closed-loop interaction between the virtual space and the physical production line. It supports seamless switching between three working modes: production line monitoring, offline debugging, and online control. Furthermore, based on a three-layer architecture of simulation center-scenario engine-deduction sandbox, it enables multi-mode, multi-rate simulation adapted to the entire lifecycle of the production line. This invention achieves multi-user safe collaborative monitoring, virtual-physical fusion closed-loop control, and flexible multi-scenario simulation for battery pack production lines, significantly improving the collaborative security, interactive reliability, and digital decision-making efficiency of production line management, and adapting to the multi-category, flexible production needs of battery pack production lines.

[0007] The specific technical solution adopted by this invention to solve its technical problem is as follows:

[0008] A multi-terminal monitoring system for a battery pack production line based on digital twins includes a twin data processing hub and a visual interactive terminal that are interconnected. The twin data processing hub is used to collect data from multiple heterogeneous devices on the battery pack production line and establish a mapping relationship between physical devices and virtual nodes in the digital twin model. The visual interactive terminal is used to drive the visual presentation of the digital twin model and receive user operation commands for the physical production line.

[0009] The twin data processing hub also includes a multi-user dynamic management unit, which provides data isolation access services for multiple visual interactive terminals, constructs a three-dimensional permission matrix based on roles, scenarios, and tasks, and realizes dynamic and refined permission control for multiple users.

[0010] The visualization interactive terminal also includes a virtual-real interaction control unit, which is communicatively connected to the multi-user dynamic management unit and the virtual model presentation unit. It is used to sequentially execute a four-level verification process of syntax verification, semantic verification, process interlock verification and security review for user operation commands that have passed the permission verification. It calculates a comprehensive risk score through a confidence transfer mechanism and controls the issuance of commands to the physical production line based on the comprehensive risk score.

[0011] The visualization interactive terminal also includes a multi-mode simulation unit, which is communicatively connected to the virtual model presentation unit, the virtual-real interaction control unit, and the twin data processing center. Based on the three-layer architecture of simulation center-scene engine-deduction sandbox, it realizes multi-mode multi-rate simulation based on production line data or user input parameters.

[0012] Furthermore, the multi-user dynamic management unit is configured with a triple isolation mechanism of data isolation, permission isolation, and access isolation; wherein data isolation achieves physical separation of different user data through database table partitioning, and access isolation restricts access behavior through dedicated communication ports and access control lists; the three-dimensional permission matrix is ​​a binary three-dimensional matrix, expressed as follows: The three dimensions of the matrix correspond to the number of categories of roles, scenes, and tasks, respectively. Matrix elements take values ​​of 0 or 1, corresponding to the role's permission status (no permission or permission granted) when performing the corresponding task in the corresponding scene. The system permission determination function is... ,in For user u's role, This is a temporary authorization marker; the temporary permissions will be automatically revoked upon completion of the task.

[0013] Furthermore, in the four-level verification process, syntax verification is used to verify the instruction protocol format, semantic verification is used to verify whether the instruction parameters are within the process threshold range, process interlock verification is used to verify the matching between the instruction and the process interlock rule base, and the security review uses a dynamic Bayesian network risk assessment method to comprehensively assess operational risks; the confidence transfer mechanism calculates a comprehensive risk score through a weighted geometric mean operator, and the comprehensive risk score calculation formula is as follows: ,in The confidence level for each level of verification. To correspond to the weighting coefficient, control commands can only be issued to the physical production line when the overall risk score is lower than the preset safety threshold.

[0014] Furthermore, the simulation hub of the multi-mode simulation unit is used to maintain the battery pack product model, equipment behavior model and process rule library. The scene engine is used to load and configure the corresponding simulation scene and output the scene initialization parameters. The simulation sandbox has a built-in time-series thruster, which is used to drive the model to run in the virtual space-time, adapting to the multi-rate simulation requirements of millisecond-level transient process and hour-level long-term equipment aging process.

[0015] Furthermore, the twin data processing hub also includes a multi-source device access unit. This unit is used to collect data in real time from various heterogeneous data sources, including PLCs, sensors, and visual recognition devices on the battery pack production line, and is compatible with multiple industrial communication protocols. The multi-source device access unit achieves breakpoint resumption of core process data through an improved sliding window method. This improved sliding window method dynamically adjusts the window length based on network transmission quality, and the dynamic adjustment formula for the window length is as follows: ,in Based on the window length, This is the jitter adjustment coefficient. To account for network transmission jitter, when resuming transmission after a breakpoint, three core process parameters—post welding current, cell pressing pressure, and welding temperature—are prioritized for retransmission.

[0016] Furthermore, the twin data processing hub also includes a data mapping and storage unit. This unit adopts a four-level storage structure: production line, workstation, equipment, and parameters, supporting multi-dimensional data queries based on production line workstation, equipment type, and time range. The data mapping and storage unit uses a time-series database for persistent storage of production line operation data, and implements value grading of production line data based on a time-degradation mechanism. The data value scoring function is as follows: ,in As the benchmark value score, For time-dependent decay function, For the time since the data was generated, For process attribute coefficients, To access feature coefficients, a differentiated storage strategy is automatically triggered based on the scoring results.

[0017] Furthermore, the visualization interactive terminal also includes a virtual model presentation unit, which is used to drive the visualization presentation of the digital twin model according to the mapping data, and supports switching between multiple perspective visualization modes such as global view, local workstation focus view and equipment detail view; the digital twin model includes a multi-precision model with three levels of detail: geometric level, process level and fault level. The virtual model presentation unit adopts a dual triggering mechanism of frustum culling and process hot zone to achieve dynamic rendering, and only renders high-precision detail levels of the model within the user's field of view and within the process hot zone.

[0018] Furthermore, the virtual model presentation unit, targeting the core processes of battery pack terminal welding and cell assembly, employs the finite element method to calculate the temperature, pressure, and stress fields, and visualizes the physical field through volumetric cloud rendering and particle system rendering. The finite element method utilizes an intrinsic orthogonal decomposition model to achieve real-time acceleration, and the reconstructed physical field formula after order reduction is as follows: Reconstruct the physical field at time t. For a reduced-order basis matrix, These are low-dimensional modal coefficients.

[0019] Furthermore, the virtual-physical interaction control unit supports seamless switching between three modes: production line monitoring, offline debugging, and online control. During the switching process, the work-in-process status is saved through a buffer. In the production line monitoring mode, the physical production line operating status is mapped to the digital twin model in real time. In the offline debugging mode, the control commands only drive the digital twin model to run. In the online control mode, compliant control commands are sent to the physical production line to drive the physical equipment to run and synchronously update the digital twin model status.

[0020] Furthermore, the simulation modes supported by the multi-mode simulation unit include real-time operation simulation, process optimization simulation, fault simulation simulation, and production line debugging simulation. In the process optimization simulation mode, the system completes multiple rounds of iterative optimization of process parameters based on intelligent optimization algorithms and outputs the optimal parameter configuration. In the fault simulation mode, the system injects faults into the virtual model through parameter coverage and deduces the fault propagation chain and impact range. In the production line debugging simulation mode, the system completes equipment layout collision detection and equipment operation cycle simulation verification.

[0021] Compared to existing technologies, this invention and its preferred solution address the core pain point of the lack of multi-user access mechanisms in existing technologies. This solution effectively balances the data sharing needs and security isolation requirements of multi-department collaboration on the production line through refined dynamic permission control and multi-dimensional isolation mechanisms. It ensures flexibility in cross-role collaboration while fundamentally avoiding security risks such as data leakage and unauthorized operations, significantly improving the management efficiency and security of concurrent access from multiple terminals. Addressing the insufficient virtual-physical interaction capabilities and weak security control in existing technologies, this solution achieves stable bidirectional closed-loop interaction between the virtual space and the physical production line, supports seamless switching between multiple working modes, and can complete production line debugging and process verification without interfering with actual production. Simultaneously, through multi-level verification and risk assessment mechanisms, it achieves full-process security control of control commands, ensuring the flexibility of production line management. Meanwhile, it minimizes safety risks in production operations. Addressing the inherent limitations of existing simulation technologies, such as limited functionality and poor scenario adaptability, this solution constructs a multi-mode, multi-rate simulation system covering the entire production line lifecycle. This system can flexibly adapt to the flexible production needs of different battery pack specifications, providing reliable digital support for scenarios such as production line process optimization, emergency fault handling, and production line layout debugging. It fully leverages the core value of digital twin technology in the intelligent upgrading of production lines. Furthermore, this solution streamlines the entire process of production line data acquisition, mapping, storage, visualization applications, interactive control, and simulation, effectively solving the problems of weak data integration and poor system scalability in traditional monitoring systems. This comprehensively improves the transparency and intelligence of battery pack production line management, providing a complete and reliable technical solution for the digital upgrading of power battery production lines. Attached Figure Description

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0023] Figure 1 A schematic diagram of the overall architecture of a multi-terminal monitoring system for a battery pack production line based on digital twins provided in an embodiment of the present invention;

[0024] Figure 2 A flowchart illustrating the three-dimensional permission matrix workflow of a multi-user dynamic management unit provided in this embodiment of the invention;

[0025] Figure 3 A flowchart of the four-level instruction security verification of the virtual-real interaction control unit provided in this embodiment of the invention;

[0026] Figure 4 The flowchart of the three-layer architecture of the multi-mode simulation unit, namely simulation hub, scene engine, and inference sandbox, provided in the embodiments of the present invention;

[0027] Figure 5 This is a schematic diagram of the deployment architecture of a multi-terminal monitoring system for battery pack production lines based on digital twins, provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the features and advantages of the present invention more apparent and understandable, specific embodiments are described below in detail:

[0029] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0030] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0031] To overcome the shortcomings of existing technologies, this invention proposes a multi-terminal monitoring system for battery pack production lines based on digital twins, addressing the deficiencies of existing digital twin technologies in areas such as multi-user access, virtual-real interaction, and simulation capabilities. By constructing a complete solution integrating data fusion, multi-user isolation, virtual-real interaction control, and multi-mode simulation, it achieves multi-user safe collaborative monitoring, virtual-real fusion closed-loop control, and flexible simulation for battery pack production lines.

[0032] Its systematic implementation includes a twin data processing hub and a visual interactive terminal.

[0033] The twin data processing hub includes:

[0034] The multi-source device access unit is used to collect battery pack production line equipment data in real time from multiple heterogeneous data sources. It supports multiple industrial communication protocols such as OPC UA and Modbus TCP, and realizes the core data breakpoint resume transmission through an improved sliding window method.

[0035] The data mapping and storage unit works in conjunction with the multi-source device access unit to map the collected physical device data to virtual nodes, and uses a time-series database to persistently store production line operation data. It also implements production line data value classification and multi-dimensional data query based on the time decay mechanism.

[0036] The multi-user dynamic management unit, connected to the data mapping and storage unit, provides data-isolated access channels and subscription services for multiple visual interactive terminals, and constructs a three-dimensional permission matrix of roles, scenarios and tasks to achieve fine-grained permission management.

[0037] The visual interactive terminal includes:

[0038] The virtual model presentation unit is used to drive the visualization of the digital twin model based on the mapping data. It can achieve dynamic rendering of multi-precision models through process feature triggering methods, and realize the visualization of battery pack process features by combining finite element, volume cloud and particle system rendering methods.

[0039] The virtual-physical interaction control unit is connected to the multi-user dynamic management unit and the virtual model presentation unit to enable users to interact and control the physical production line through the virtual interface. It achieves safe interaction through a four-level verification of syntax, semantics, process interlock, and security.

[0040] The multi-mode simulation unit, connected to the virtual model presentation unit and the virtual-real interaction control unit, is based on the simulation hub-scene engine-deduction sandbox architecture to realize different simulation modes based on production line data or user input parameters.

[0041] This invention realizes multi-dimensional dynamic access control, high-confidence virtual-real closed-loop control and multi-mode flexible simulation for battery pack production lines, significantly improving the collaborative security, interactive reliability and digital decision-making efficiency of the production line.

[0042] As a preferred implementation, the multi-source device access unit supports the access of various heterogeneous data sources, including PLCs, sensors, and vision recognition devices, to collect production line equipment operation data, process parameter data, environmental data, and product quality inspection data. The built-in data preprocessing module removes redundant and invalid data through data filtering, deduplication, format standardization, and noise reduction. The improved sliding window method dynamically adjusts the window length based on network transmission quality; the dynamic adjustment formula is as follows: ,in Based on the window length, This is the jitter adjustment coefficient. To account for network transmission jitter, when resuming transmission after a breakpoint, three core process parameters—post welding current, cell pressing pressure, and welding temperature—are prioritized for retransmission.

[0043] As a preferred implementation, the data mapping and storage unit adopts a four-level storage structure of production line, workstation, equipment, and parameters, enabling multi-dimensional data queries based on production line workstation, equipment type, and time range. Data value grading employs a dynamic evaluation mechanism based on data timeliness, process attributes, and access characteristics, with the data value scoring function being: ,in D is the benchmark value score. t ( ) is the time-dependent decay function. For the time since the data was generated, For process attribute coefficients, To access feature coefficients and automatically trigger differentiated storage strategies based on evaluation results, core data is stored using both a time-series database and local disks.

[0044] As a preferred implementation method, the multi-user dynamic management unit supports the dynamic creation, cancellation, and resource reclamation of users, and automatically completes the allocation and release of user-specific channels without manual intervention.

[0045] More specifically, the isolation mechanism of the multi-user dynamic management unit includes data isolation, permission isolation, and access isolation. Data isolation achieves physical separation of different user data through database table partitioning; permission isolation is implemented based on a three-dimensional permission matrix; and access isolation implements access restrictions through dedicated communication ports and access control lists, ensuring that data between different users is not leaked. The role-scenario-task three-dimensional permission matrix P... rst ∈{0,1} NR×NS×NT Dimensional cardinality N R N S N T These represent the number of characters, scenes, and task categories, respectively, and matrix element P. rst This represents the permission status of role r in scenario s when executing task t; the system determines valid permission items by querying and matching in the three-dimensional permission matrix based on the user's login scenario s and the current task t. The permission determination function is... ,in For users The role to which they belong. This is a temporary authorization marker; the temporary permissions will be automatically revoked upon completion of the task.

[0046] As a preferred implementation method, the virtual model presentation unit supports multi-view visualization presentation modes, including a global view, a local workstation focus view, and a device detail view, to meet the needs of different scenarios such as overall monitoring, key observation, and fault diagnosis, and improve the flexibility of monitoring.

[0047] More specifically, the virtual model rendering unit has three levels of detail: geometric, process, and fault. The process-triggered dynamic rendering adopts a dual triggering mechanism of view frustum culling and process hot zone, rendering high-precision detail only for models within the user's field of view and within the process hot zone.

[0048] As a further preferred implementation, the process hotspot refers to the equipment model area corresponding to core processes such as terminal welding and cell pressing in the battery pack production line. It is a key area for monitoring the production line's operating status and displaying process details. The process-triggered dynamic rendering adopts a dual triggering mechanism of frustum culling and process hotspot, rendering high-precision detail levels only on the model within the user's field of view and within the process hotspot.

[0049] The virtual model presentation unit targets the core processes of battery pack terminal welding and cell assembly. It uses the finite element method (FEM) to calculate the temperature, pressure, and stress fields, and visualizes the physical field through volumetric cloud rendering and particle system rendering. The FEM employs an eigenorthogonal decomposition model for real-time acceleration. By extracting principal modes from the snapshot matrix through singular value decomposition, a reduced-order basis matrix is ​​constructed, projecting the original high-dimensional finite element space to a low-dimensional modal space. The reconstructed physical field formula after the reduction is as follows: ,in Reconstruct the physical field at time t. For a reduced-order basis matrix, These are low-dimensional modal coefficients.

[0050] The virtual-physical interactive control unit supports switching between three modes: production line monitoring, offline debugging, and online control. During the switching process, the work-in-process status is saved through a buffer to avoid data loss. In production line monitoring mode, the physical production line operating status is automatically and synchronously mapped to the digital twin model to maintain real-time consistency between the virtual and physical states. In offline debugging mode, control commands only drive the digital twin model to run and provide feedback on simulation results without affecting physical production line production. In online control mode, compliant control commands can be issued to the physical production line to drive the physical equipment to run and synchronously update the virtual model status.

[0051] The virtual-physical interaction control unit incorporates a built-in command validity verification mechanism, including a four-level check process: syntax parsing, semantic verification, process interlock verification, and safety review. Syntax parsing uses regular expressions to verify the command protocol format; semantic verification verifies whether parameters are within process threshold ranges; process interlock verification verifies the matching of process interlock rule bases through rule matching; and safety review employs a dynamic Bayesian network risk assessment method, comprehensively considering operational history, equipment operating conditions, environmental factors, and personnel status to conduct risk assessment. The four levels of verification work collaboratively through a confidence propagation mechanism. The confidence level and weight of each verification level participate in the calculation of the comprehensive risk score. The verification confidence calculation uses the confidence level of adjacent preceding verification levels as input weights, ultimately outputting a comprehensive risk score. The formula for calculating the comprehensive risk score is as follows: ,in For each level of verification confidence, This is a weighting coefficient; instructions can only be issued when the score is below the safety threshold.

[0052] As a preferred implementation method, the simulation modes of the multi-mode simulation unit include real-time operation simulation, process optimization simulation, fault simulation simulation and production line debugging simulation, which are suitable for different application scenarios such as multi-category production, process iteration, and fault emergency response.

[0053] More specifically, the multi-mode simulation unit includes a simulation hub, a scenario engine, and a simulation sandbox. The simulation hub is used to acquire real-time or historical production line data and maintain the battery pack product model, equipment behavior model, and process rule library required for simulation. The scenario engine is used to load and configure real-time running, process optimization, fault simulation, and production line debugging simulation scenarios according to user selection or system triggering, and output scenario initialization parameters to the simulation sandbox. The simulation sandbox has a built-in timing actuator to drive the model in the simulation hub to run in virtual spacetime based on the scenario initialization parameters, supporting millisecond-level process transients and hour-level equipment aging multi-rate simulations, and outputting the simulation process and result data to the virtual model presentation unit.

[0054] Compared with existing technologies, this invention addresses core pain points in battery pack production line monitoring scenarios, such as the lack of multi-user access mechanisms, insufficient security of virtual-real interaction, and limited simulation functionality with poor adaptability to production scenarios. It proposes a multi-terminal monitoring system for battery pack production lines based on digital twins. This system constructs a twin data processing hub to collect, map, store, and dynamically manage data from multiple heterogeneous devices in real time. High-fidelity virtual presentation, two-way interactive control, and multi-scenario simulation are achieved on a visual interactive terminal. Furthermore, through the synergistic effect of role-scenario-task three-dimensional permission management, syntax-semantics-process interlocking-security four-level verification, and a three-layer architecture of simulation hub-scenario engine-deduction sandbox, it solves the problems of limited concurrent multi-terminal access, fragmented virtual-real interaction and high security risks, and fixed simulation scenarios in traditional monitoring systems. This significantly improves the collaboration, flexibility, security, and adaptability of production line monitoring. Compared to traditional monitoring systems with weak multi-terminal service capabilities, rigid virtual-real interaction, and single simulation modes, this invention can realize multi-user secure collaborative access, virtual-real fusion closed-loop control, and flexible simulation in multiple scenarios for battery pack production lines, providing reliable technical support for the digital upgrade of intelligent battery pack manufacturing.

[0055] By constructing a multi-user dynamic management unit, fine-grained data isolation and access control based on a three-dimensional permission matrix of roles, scenarios, and tasks are achieved. This unit establishes a permission mapping relationship between roles, scenarios, and tasks, covering multiple types of roles, scenarios, and tasks. It can automatically match corresponding permission items based on the user's login scenario and the currently executed task, and automatically reclaim temporary permissions upon task completion. Combined with a triple isolation mechanism of data, permissions, and access, it achieves dynamic, on-demand authorization with the least privilege. This mechanism can dynamically create dedicated access channels for different departments such as production, operation and maintenance, and process, effectively resolving the contradiction between data sharing and secure isolation. Its benefits include ensuring that departments can collaborate based on a unified data source while completely eliminating the risks of unauthorized operations and data leaks. User channel allocation and release can be completed without manual intervention, significantly improving access management efficiency and providing a secure and reliable technical foundation for multi-role, cross-departmental collaborative management of the battery pack production line.

[0056] By constructing a virtual-physical interactive control unit, stable, accurate, and secure two-way interaction between the virtual space and the physical production line is achieved. This unit supports seamless switching between three modes: production line monitoring, offline debugging, and online control. During switching, a buffer saves the work-in-process status to prevent data loss. In production line monitoring mode, the physical production line's operating status is automatically and synchronously mapped to the digital twin model, maintaining real-time consistency between the virtual and physical states and providing a foundation for production line monitoring. In offline debugging mode, control commands only drive the digital twin model and provide simulation results, suitable for process verification and fault simulation. The online control mode incorporates a four-level verification mechanism: syntax, semantics, process interlocking, and security. It employs confidence-based collaborative decision-making, and outputs a comprehensive risk score through regular expression syntax parsing, process threshold semantic verification, rule-based process interlocking verification, and dynamic Bayesian network security review. Only compliant commands can be issued to the physical production line. The benefits include allowing users to perform production line debugging, process verification, and fault simulation risk-free within the virtual model, significantly reducing trial-and-error costs. Simultaneously, the four-level verification mechanism maximizes the avoidance of security risks associated with erroneous or unauthorized command issuance, significantly improving the security and flexibility of production line control.

[0057] By constructing a multi-mode simulation unit, relying on a three-layer architecture of simulation hub, scenario engine, and simulation sandbox, a flexible, multi-scenario, multi-rate simulation capability is provided. In this unit, the simulation hub serves as the core knowledge base, maintaining three core models: battery pack products, equipment behavior, and process rules. The scenario engine, as a scenario orchestrator, can quickly load and configure four types of simulation scenarios—real-time operation, process optimization, fault simulation, and production line debugging—based on user needs or system triggers. The simulation sandbox, as a secure, isolated virtual testbed, supports millisecond-level process transients and hourly-level equipment aging multi-rate simulations, completely isolated from the physical production line, and synchronously outputs simulation process and result data. The benefit is that users can quickly switch simulation scenarios according to the production categories or process requirements of different battery packs, enabling the system to quickly adapt to multi-category, flexible production needs. This not only enhances the adaptability of the digital twin system to complex production environments but also provides powerful data-driven decision support for production line optimization, fault contingency planning, and personnel training, achieving a closed-loop iteration of simulation-feedback-optimization and fully releasing the application value of digital twins.

[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention are described clearly and completely below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0059] This invention provides a multi-terminal monitoring system for battery pack production lines based on digital twins. By constructing a collaborative architecture between a twin data processing hub and a visual interactive terminal, the system achieves a complete closed loop for battery pack production lines, from data acquisition and storage management to multi-user visual monitoring and interactive control. Figure 1 This invention demonstrates the overall architecture of the digital twin battery pack production line multi-terminal monitoring system provided in an embodiment of the present invention. Figure 2 This illustrates the workflow of the multi-user dynamic management unit provided in an embodiment of the present invention. Figure 3 This demonstrates the workflow of the virtual-real interaction control unit provided in an embodiment of the present invention. Figure 4 This demonstrates the workflow of the multi-mode simulation unit provided in this embodiment of the invention. Figure 5 The system deployment architecture provided by the embodiments of the present invention is shown.

[0060] The following combination Figure 1 A detailed description of the overall system architecture is provided. Figure 1 The overall architecture of the battery pack production line multi-terminal monitoring system based on digital twins, as shown, consists of two main modules: a digital twin data processing hub and a visual interactive terminal.

[0061] The twin data processing hub is responsible for multi-source data acquisition, fusion storage, and access management, specifically including:

[0062] The multi-source device access unit is used to collect real-time equipment operation, process parameters, environmental and product quality testing data from multiple heterogeneous data sources deployed on various physical equipment and supporting facilities on the battery pack production line. It is compatible with multiple industrial communication protocols, ensures data validity through a built-in preprocessing module, and realizes core data breakpoint resume transmission through an improved sliding window method.

[0063] The data mapping and storage unit, which is connected to the multi-source device access unit, is used to establish a precise mapping relationship between physical entities and virtual nodes. It uses a time-series database to persistently store and manage the historical status of production line operation data, supporting the value classification of production line data and multi-dimensional data query.

[0064] The multi-user dynamic management unit, connected to the data mapping and storage unit, is used to dynamically allocate resources based on user needs, provide dedicated access channels and services for multiple visual interactive terminals with data isolation, and build a three-dimensional permission matrix of role-scenario-task to achieve fine-grained permission management and ensure the security of multi-terminal collaborative access.

[0065] The visual interactive terminal provides intuitive visualization of production line status, virtual-real interactive control, and multi-scenario simulation, specifically including:

[0066] The virtual model presentation unit is used to drive the visualization of digital twin models based on mapping data. It integrates geometric, physical and behavioral models to replicate the structure, operating characteristics and motion laws of production line equipment. It can achieve dynamic rendering of multi-precision models through process feature triggering methods, and combine finite element, volume cloud and particle system rendering methods to visualize battery pack process features and realize intuitive monitoring of production line status.

[0067] The virtual-physical interaction control unit is connected to the multi-user dynamic management unit and the virtual model presentation unit to enable users to interact and control the physical production line through the virtual interface. It supports switching between three modes: production line monitoring, offline debugging, and online control. Secure interaction is achieved through a four-level verification of syntax, semantics, process interlock, and security.

[0068] The multi-mode simulation unit, connected to the virtual model presentation unit and the virtual-real interaction control unit, is based on the simulation hub-scene engine-inference sandbox architecture. It is used to realize different simulation modes based on real-time production line data or user input parameters, adapt to the needs of all scenarios such as production line operation, optimization, and debugging, and provide data support for decision-making.

[0069] The simulation hub, acting as a model and knowledge repository, is responsible for acquiring real-time or historical production line data and maintaining the three core elements required for simulation: battery pack product models, equipment behavior models, and process rules. The battery pack product model library stores parameterized 3D geometric models and bills of materials; the equipment behavior model library contains kinematic, dynamic, and control logic models of production line equipment (such as six-axis welding robots, laser galvanometers, and AGVs); and the process rule library stores process constraints and interlocking logic for battery pack production. As a further preferred implementation, the simulation hub and the scenario engine achieve bidirectional data exchange through a standardized API interface, ensuring efficient invocation and synchronous updates of simulation models and rules. The scenario engine, acting as a simulation scheduler, dynamically loads the corresponding models and rules from the simulation hub based on the user-selected mode or system-triggered events, configures and generates specific simulation scenarios, including initial parameters, boundary conditions, and simulation objectives, and outputs this scenario initialization parameter package to the simulation sandbox.

[0070] Specifically, the multi-source device access unit supports the access of various heterogeneous data sources, including PLCs, sensors, and vision recognition devices, collecting production line equipment operation data, process parameter data, environmental data, and product quality inspection data. The built-in data preprocessing module removes redundant and invalid data through data filtering, deduplication, format standardization, and noise reduction. This unit adopts a protocol adapter design pattern, developing dedicated adapter components for different communication protocols. The OPCUA adapter uses the Basic256Sha256 security strategy to establish an encrypted channel, while the Modbus adapter collects register data with a 50ms polling cycle. In the data processing stage, an improved sliding window algorithm is first used for data filtering to remove outliers exceeding the normal fluctuation range. This improved sliding window uses an adaptive window length mechanism, specifically calculating the network transmission jitter coefficient with a 100ms statistical cycle. (σ is the standard deviation of transmission delay, μ is the mean of transmission delay), the dynamic adjustment formula for window length L is: ,in =50 is the base window length (unit: data frame), α=0.8 is the jitter adjustment coefficient, when When the value is ≥0.3, the maximum window length can be expanded to... =100, when J≤0.1 the window length shrinks to a minimum. =20, ensuring that the data filtering effect is adapted to the network transmission status; at the same time, it realizes the breakpoint resume transmission of core process parameters such as battery pack terminal welding and cell assembly, and the retransmission priority of core parameters is higher than that of regular parameters when resuming the transmission; then, the hash deduplication algorithm is used to delete duplicate data, and then the format is standardized and converted according to the industry data standard preset by the battery pack production line. Finally, the wavelet denoising algorithm is used to reduce the signal noise caused by environmental interference, ensuring the accuracy and consistency of the output data, and providing high-quality data input for subsequent processing.

[0071] Specifically, the data mapping and storage unit uses a time-series database to achieve multi-dimensional data querying and long-term historical data management based on production line workstations, equipment types, and time ranges. It adopts a four-level storage structure: production line, workstation, equipment, and parameters. The main storage directory is divided according to production line areas, and then further subdivided into subdirectories according to workstation functions. Core operating parameters are stored as data blocks in timestamp order, achieving standardized data storage. The database has a built-in index optimization mechanism, establishing composite indexes for workstations, equipment types, and time ranges to meet multi-dimensional query needs. It supports historical data backtracking queries at time granularities such as milliseconds, seconds, minutes, hours, and days, ensuring a write throughput of ≥100,000 records / second and a single query response time of ≤50ms. This allows for rapid response to data call requests during data analysis, fault tracing, and process optimization, ensuring efficient reading, writing, and management of massive amounts of time-series data. This unit constructs a dynamic evaluation mechanism based on a data value decay function, using timeliness, process attributes, and access characteristics as independent variables to build a dynamic evaluation model, achieving value grading of production line data. The specific implementation of the dynamic evaluation mechanism is as follows: a data value scoring function is defined for the characteristics of battery pack production line data. ,in The benchmark value score; For time-dependent decay functions, a hyperbolic decay model is adopted. , For the value halving cycle, The time elapsed since the data was first published; These are process attribute coefficients, taken from core process parameters such as electrode welding current, cell pressing pressure, and welding temperature. Auxiliary parameters such as equipment operating status and ambient temperature and humidity are taken. For log-type data retrieval ; The access characteristic coefficient is segmented based on the access frequency over the past 24 hours: The value of data decreases dynamically over time. When the value falls below a preset threshold, a differentiated storage strategy is automatically triggered. The data is determined to be core data and is stored in a dual-storage mode of time-series database and local disk. The local disk is partitioned by hour in Parquet format to ensure data traceability when the production line is offline. Data that is deemed important is stored using a time-series database with a single copy. If the data is determined to be regular data, it will be migrated to object storage. Data is identified as archived and transferred to a tape library (LTFS format) to ensure the security and traceability of data storage. Furthermore, the core data after classification will be prioritized and pushed to the multi-mode simulation unit and the virtual-real interaction control unit to provide data support for simulation and instruction verification.

[0072] Specifically, the multi-user dynamic management unit supports the dynamic creation, cancellation, and resource reclamation of users. The allocation and release of user-specific channels can be completed automatically without manual intervention. After receiving a user login request, the system first completes identity authentication and authorization to verify the legitimacy of the user's credentials. After successful authentication, a dedicated data channel is established for the corresponding user, and dedicated resources such as message queues and database connection pools are allocated. Subsequently, the data push phase begins, continuously pushing real-time production line status data to the corresponding user terminal. During this process, the user session status is continuously monitored. If the system detects that the user has actively cancelled or the session has timed out (default 30 minutes), the resource release and cleanup process is triggered, the data channel is securely closed, and the allocated resources are reclaimed, ultimately ending the user session service process and realizing automated management of the entire lifecycle of user access resources.

[0073] Specifically, the isolation mechanism of the multi-user dynamic management unit includes data isolation, permission isolation, and access isolation. These three mechanisms work together to ensure the security of multi-terminal access and guarantee that data between different users is not leaked. Its core role-scenario-task three-dimensional permission matrix execution logic is as follows: Figure 2 As shown. Data isolation achieves physical separation of different user data through database table partitioning, and a multi-tenant architecture is adopted to create independent data nodes for each user, thus achieving physical data isolation; permission isolation is implemented based on a three-dimensional permission matrix of role-scenario-task, P. rst ∈{0,1} NR×NS×NT Dimensional cardinality N R N S N T These represent the number of characters, scenes, and task categories, respectively, and matrix element P. rst This is a binary discrete value, where 0 represents no permission and 1 represents permission. It uniquely corresponds to the permission status of the r-th role executing the t-th task in scenario s. Based on the user's login scenario and the currently executed task, the corresponding permission item is automatically matched. The permission status of the corresponding user's role is accurately queried in the three-dimensional permission matrix to determine the valid permission item and grant it in real time. Differentiated operation permissions are assigned to different roles such as production personnel, process engineers, and system administrators. The permission determination function is... ,in For users The role to which they belong. This is a temporary authorization marker; the temporary permissions are automatically revoked upon completion of the task. Access isolation implements access restrictions through dedicated communication ports and access control lists. A unique communication port is assigned to each user, and only legitimate users are included in the access control list, enabling precise control over access behavior.

[0074] Specifically, the virtual model presentation unit is built on the Unity3D engine to create a high-fidelity digital twin model, integrating the 3D geometric model, physical model, and behavioral model of the physical production line. It accurately replicates the structural features, physical characteristics, and behavioral logic of the production line equipment. This unit supports three multi-view visualization modes: a global view, a local workstation focus view, and an equipment detail view. Users can flexibly switch between these modes using the view control buttons on the interactive interface: the global view provides a bird's-eye view, allowing users to view the overall operating status of the entire production line; the local workstation focus view automatically zooms in to display the equipment layout and operational details of the selected workstation; and the equipment detail view displays the internal structure and operational status of key components. The ability to switch between these three views adapts to different scenarios, such as overall monitoring, focused observation, and troubleshooting, enhancing the flexibility of production line monitoring.

[0075] Specifically, the virtual model presentation unit has a multi-precision model with three levels of detail: geometric, process, and fault. The process-triggered dynamic rendering adopts a dual triggering mechanism of view frustum culling and process hot zone. This mechanism enables the precise removal of invalid rendering content, effectively reducing video memory usage by ≥50%, while ensuring a real-time rendering frame rate of ≥40FPS for the model, thus balancing the smoothness of production line monitoring with the completeness of process detail display.

[0076] Specifically, the virtual model presentation unit features real-time visualization and annotation of production line operating status. Equipment operating status is coded in three colors: green, yellow, and red correspond to normal operation, warning status, and fault shutdown, respectively. The status switching delay is less than 100ms. Through intuitive color differentiation and text overlay, the system clearly displays equipment operating status, process parameter values, and abnormal warning information, facilitating users' rapid capture of key operating data. Furthermore, for core processes such as battery pack terminal welding and cell assembly, the finite element method is used to accurately calculate the temperature, pressure, and stress fields. To achieve real-time finite element simulation, the finite element model is optimized using an intrinsic orthogonal decomposition model. Specifically, this involves first collecting finite element simulation results under all operating conditions of the core battery pack process as snapshot samples, and then constructing a snapshot matrix. (n is the number of nodes in the finite element mesh, and m is the number of snapshot samples), for the snapshot matrix Centralized processing is required to obtain ( (mean of the samples); then the centered snapshot matrix Perform Singular Value Decomposition (SVD): ,in It is a left singular matrix. It is a singular value diagonal matrix. The matrix is ​​a right singular matrix; the top K principal modes with a cumulative singular value ratio of ≥99% are extracted to form a reduced-order basis matrix. Projecting the original high-dimensional finite element space to a low-dimensional modal space, the reduced-order physical field reconstruction formula is as follows: ( Reconstruct the physical field at time t. (These are low-dimensional modal coefficients). In real-time simulation, only the low-dimensional modal coefficients need to be solved. It can quickly reconstruct the temperature field, pressure field, and stress field, compared to the original full-dimensional finite element calculation (the number of mesh nodes is usually 10). 4 ~10 5 The computational dimension is compressed by a factor of 100, and the computational speed is increased by 8 to 15 times to meet the needs of real-time rendering. Furthermore, the physical field is visualized through volumetric cloud rendering and particle system rendering, which transforms abstract process physical parameters into intuitive visual images, allowing users to accurately grasp the physical field change patterns of core processes.

[0077] Specifically, the virtual-physical interaction control unit supports switching between three modes: production line monitoring, offline debugging, and online control. After receiving and recognizing the mode command input by the user, the system determines the working mode and executes the corresponding branch process based on the result: In production line monitoring mode, the system feeds back the real-time status of the physical production line to the virtual model, achieving synchronization between virtual and physical states; in offline debugging mode, the system receives debugging commands and verifies their validity. Commands that pass verification only drive the virtual model to run and provide feedback, without sending any control signals to the physical production line; in online control mode, the system sequentially executes the steps of receiving control commands, verifying their validity, sending compliant commands to the corresponding equipment, and providing production line status feedback, achieving closed-loop control from the virtual interface to the physical equipment. The three modes can be flexibly switched through clear decision branches to meet the production line interaction and management needs in different scenarios.

[0078] Specifically, the virtual-real interaction control unit has a built-in command validity verification mechanism, whose core four-level verification process includes syntax-semantics-process interlocking-security checks, as follows: Figure 3As shown, the four-level verification works in concert, and instructions must pass the full-process verification before they can be sent to the physical production line, maximizing the safety of production line operation. The syntax parsing stage uses regular expressions to verify whether the instruction format conforms to industrial communication protocol specifications such as OPC UA and Modbus, and outputs the syntax verification confidence score. The semantic verification stage checks whether the instruction parameter values ​​are within the preset process threshold range of each process in the battery pack production line, and uses the previous stage syntax confidence score as input weight to output the semantic verification confidence score. The process interlock verification stage uses a rule matching algorithm to verify the matching of instruction parameters with the battery pack process interlock rule base, and verifies the compliance of the linkage between parameters, using the product of the previous stage confidence scores as input weight to output the process interlock verification confidence score. The safety review stage uses a dynamic Bayesian network risk assessment method, using historical operation sequences, current equipment status, and environmental parameters as evidence nodes to calculate the probability of failure after the instruction is issued, and outputs the safety review confidence score. After the four levels of verification are completed, a comprehensive risk score is calculated. Only when the score is lower than the safety threshold can the instruction pass the verification. If the verification fails, the instruction is intercepted, and an error code and detailed description are returned to the operation terminal through a message queue. Only instructions that pass the verification are encapsulated into industrial control messages and sent to the PLC for execution.

[0079] For safety risk assessment of battery pack production lines, a dynamic Bayesian network with three layers of nodes was constructed. The evidence layer nodes E = {E1, E2, E3, E4} represent real-time observed variables. For operation history nodes, count the number of abnormal user operations (such as parameter over-limit adjustment, frequent mode switching) in the past hour, with values ​​of {0, 1, 2, ≥3}. For equipment operating condition nodes, the health score of key equipment such as laser welding machine and press machine (based on vibration, temperature and current characteristics) is comprehensively evaluated, and the value status is {good, average, deteriorated}. As an environmental factor node, it monitors the deviation of production line temperature, humidity, and dust concentration from standard operating conditions, and sets the value status as {normal, deviation, danger}. For personnel status nodes, assess the current operator's fatigue level (continuous working hours) and skill level (historical operation pass rate), with values ​​of {reliable, average, high risk}. Intermediate layer nodes I = {I1, I2} represent the overall system status. The system stability index. The personnel reliability index is obtained through training on historical failure cases using a conditional probability table. Parameters are estimated using the EM algorithm, and the prior distribution is set based on the FTA (Fault Tree Analysis) results from the battery pack production line. The target layer node H represents the failure probability, indicating the probability that issuing an instruction will result in equipment damage, product quality defects, or a safety accident. Network inference uses a connection tree algorithm; given evidence e = {e1, e2, e3, e4}, the marginal probabilities are calculated forward. ,in , These represent the state values ​​of intermediate layer nodes I1 and I2, and their security audit confidence levels. The network structure is customized for battery pack production lines. The node state division and conditional probability parameters are determined based on historical operating data and fault case databases of key processes such as electrode welding and cell pressing, ensuring that the risk assessment is highly consistent with the actual operating conditions of the production line.

[0080] Level 4 verification works collaboratively through a weighted geometric mean confidence propagation operator. The comprehensive risk score is calculated using the following formula: ,in For syntax validation confidence, To verify the confidence level for semantics, To verify the confidence level of process interlocking, To ensure the confidence level for safety reviews, this embodiment simplifies the weighting of confidence level transfer to a fixed value to meet the safety control requirements of battery pack production lines. The weighting configuration... This reflects the principle that security becomes increasingly critical as time progresses. The confidence levels are calculated as follows: Syntax validation The value is 1 if the instruction format is a perfect match, 0.5 if the node exists but the type does not match, and 0 if the format is incorrect; semantic validation. ,in For command parameter values, This is the nominal value of the process. For process standard deviation, when parameters exceed the threshold range Process interlock verification ,in To meet the process rule number, Total number of relevant rules; security review Output by a dynamic Bayesian network. Final risk score. Only when The instruction passed the verification.

[0081] Specifically, the multi-mode simulation unit adapts to different application scenarios in battery pack production, supporting four core modes: real-time simulation, process optimization simulation, fault simulation, and production line debugging simulation. In real-time simulation mode, the system subscribes to real-time data streams from the physical production line, using a time-stepping algorithm to synchronize the virtual model with the physical world, enabling real-time monitoring and mirroring of the current battery pack production status. In process optimization simulation mode, after the user sets optimization goals (such as improving yield and reducing energy consumption) and parameter ranges, the system calls a built-in improved genetic algorithm to automatically perform multiple iterations in the virtual environment, evaluating the production effects under different combinations of process parameters and ultimately outputting the optimal parameter configuration scheme. In fault simulation mode, the user can select or the system can preset fault types (such as motor overload or sensor malfunction) and their locations. Faults are injected into the virtual model using a parameter overlay method, deduce the fault propagation chain and its impact on the overall production line status, and are used for fault emergency drills and emergency plan verification. In the production line debugging simulation mode, users can modify the production line layout and adjust equipment parameters through drag-and-drop UI. The system performs collision detection based on axis-aligned bounding boxes (equipment spacing safety threshold ≥500mm) and simulates and verifies the control logic and cycle time of equipment such as robots and conveyors, thereby significantly shortening the on-site debugging cycle of new or modified production lines.

[0082] Specifically, the multi-mode simulation unit implements the aforementioned multi-mode simulation capabilities based on a three-layer collaborative architecture of simulation hub, scene engine, and inference sandbox. Their collaborative relationship is as follows: Figure 4 As shown, the simulation hub, acting as a model and knowledge repository, is responsible for acquiring real-time or historical production line data and maintaining the three core elements required for simulation: battery pack product models, equipment behavior models, and process rules. The battery pack product model library stores parameterized 3D geometric models and bills of materials; the equipment behavior model library contains kinematic, dynamic, and control logic models of production line equipment (such as six-axis welding robots, laser galvanometers, and AGVs); and the process rule library stores process constraints and interlocking logic for battery pack production. The scenario engine, acting as a simulation scheduler, dynamically loads the corresponding models and rules from the simulation hub based on the user-selected mode or system-triggered events, configures and generates specific simulation scenarios, including initial parameters, boundary conditions, and simulation objectives, and outputs this scenario initialization parameter package to the simulation sandbox. The simulation sandbox, as a safe and isolated virtual testbed, has a built-in timing actuator. After receiving the scenario parameter package, it drives the models in the simulation hub to run in a virtual spacetime. This timing actuator supports multi-rate simulation: for the millisecond-level thermal process of electrode welding, a variable-step Runge-Kutta method is used with a time step Δτ = 0.001s, and 1 second of physical time corresponds to a simulation calculation time of <50ms; for the second-level mechanical process of cell pressing, an implicit Euler method is used with Δτ = 0.01s; for the hour-level long-term process of equipment performance degradation, an accelerated simulation strategy is adopted with a time scaling factor r. fast=3600. The simulation sandbox outputs all process and result data to the virtual model presentation unit in real time, realizing the visualization of the simulation process.

[0083] As a further preferred implementation, the simulation hub and the scene engine achieve bidirectional data exchange through a standardized API interface, ensuring efficient invocation and synchronous updating of simulation models and rules.

[0084] The following combination Figure 5 Provide a detailed description of the system deployment architecture. Figure 5 This diagram illustrates the deployment architecture of the digital twin battery pack production line multi-terminal monitoring system provided in this embodiment of the invention. The architecture adopts a three-layer model of "acquisition-processing-presentation". The leftmost layer is the physical production line layer, which acts as the physical entity. PLCs, various sensors, and visual recognition devices deployed on-site collect raw data in real time, including production line operating status, process parameters, and images / videos. Production line data is aggregated and transmitted to the data processing layer located at the center of the architecture via standard industrial communication cables. The twin data processing hub is deployed on an industrial computer. This hub is the core computing unit of the system, responsible for receiving and integrating real-time data streams from all underlying devices. After data fusion, mapping, and storage, the data is distributed to the visual interactive terminals. The rightmost layer consists of multiple independent visual interactive terminals. The data processing hub establishes parallel data distribution channels to multiple terminals simultaneously through the enterprise network. A single data source can support real-time monitoring and interaction for multiple terminal users located in different physical locations, constructing a centralized processing and distributed access digital monitoring architecture for the battery pack production line.

[0085] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0086] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0087] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

[0088] This invention is not limited to the preferred embodiment described above. Anyone inspired by this invention can derive various other forms of a multi-terminal monitoring system for battery pack production lines based on digital twins. All equivalent variations and modifications made within the scope of the claims of this invention shall fall within the scope of this invention.

Claims

1. A multi-terminal monitoring system for a battery pack production line based on digital twins, comprising a twin data processing hub and a visual interactive terminal interconnected by communication; the twin data processing hub is used to collect data from multi-source heterogeneous equipment on the battery pack production line and establish a mapping relationship between physical equipment and virtual nodes in the digital twin model; the visual interactive terminal is used to drive the visual presentation of the digital twin model and receive user operation commands for the physical production line; characterized in that: The twin data processing hub also includes a multi-user dynamic management unit, which provides data isolation access services for multiple visual interactive terminals, constructs a three-dimensional permission matrix based on roles, scenarios, and tasks, and realizes dynamic and refined permission control for multiple users. The visualization interactive terminal also includes a virtual-real interaction control unit, which is communicatively connected to the multi-user dynamic management unit and the virtual model presentation unit. It is used to sequentially execute a four-level verification process of syntax verification, semantic verification, process interlock verification and security review for user operation commands that have passed the permission verification. It calculates a comprehensive risk score through a confidence transfer mechanism and controls the issuance of commands to the physical production line based on the comprehensive risk score. The visualization interactive terminal also includes a multi-mode simulation unit, which is communicatively connected to the virtual model presentation unit, the virtual-real interaction control unit, and the twin data processing center. Based on the three-layer architecture of simulation center-scene engine-deduction sandbox, it realizes multi-mode multi-rate simulation based on production line data or user input parameters.

2. The multi-terminal monitoring system for battery pack production lines based on digital twins according to claim 1, characterized in that: The multi-user dynamic management unit is configured with a triple isolation mechanism: data isolation, permission isolation, and access isolation. Data isolation achieves physical separation of different user data through database table partitioning; access isolation restricts access behavior through dedicated communication ports and access control lists. The three-dimensional permission matrix is ​​a binary three-dimensional matrix, expressed as follows: The three dimensions of the matrix correspond to the number of categories of roles, scenes, and tasks, respectively. Matrix elements take values ​​of 0 or 1, corresponding to the role's permission status (no permission or permission granted) when performing the corresponding task in the corresponding scene. The system permission determination function is... ,in For user u's role, This is a temporary authorization marker; the temporary permissions will be automatically revoked upon completion of the task.

3. The multi-terminal monitoring system for battery pack production lines based on digital twins according to claim 1, characterized in that: In the four-level verification process, syntax verification is used to verify the instruction protocol format, semantic verification is used to verify whether the instruction parameters are within the process threshold range, process interlock verification is used to verify the matching between the instruction and the process interlock rule base, and the security review uses a dynamic Bayesian network risk assessment method to comprehensively assess operational risks; the confidence transfer mechanism calculates a comprehensive risk score using a weighted geometric mean operator, and the comprehensive risk score calculation formula is as follows: ,in The confidence level for each level of verification. To correspond to the weighting coefficient, control commands can only be issued to the physical production line when the overall risk score is lower than the preset safety threshold.

4. The multi-terminal monitoring system for battery pack production lines based on digital twins according to claim 1, characterized in that: The simulation hub of the multi-mode simulation unit is used to maintain the battery pack product model, equipment behavior model and process rule library. The scene engine is used to load and configure the corresponding simulation scene and output the scene initialization parameters. The simulation sandbox has a built-in time-series thruster, which is used to drive the model to run in virtual space-time, adapting to the multi-rate simulation requirements of millisecond-level transient process and hour-level long-term equipment aging process.

5. A multi-terminal monitoring system for a battery pack production line based on digital twins according to claim 1, characterized in that: The twin data processing hub also includes a multi-source device access unit, which is used to collect data in real time from various heterogeneous data sources such as PLCs, sensors, and visual recognition devices in the battery pack production line, and is compatible with multiple industrial communication protocols. The multi-source device access unit achieves breakpoint resumption of core process data transmission through an improved sliding window method. This improved sliding window method dynamically adjusts the window length based on network transmission quality, and the dynamic adjustment formula for the window length is as follows: ,in Based on the window length, This is the jitter adjustment coefficient. To account for network transmission jitter, when resuming transmission after a breakpoint, three core process parameters—post welding current, cell pressing pressure, and welding temperature—are prioritized for retransmission.

6. A multi-terminal monitoring system for a battery pack production line based on digital twins according to claim 1, characterized in that: The twin data processing hub also includes a data mapping and storage unit. This unit employs a four-level storage structure: production line, workstation, equipment, and parameters, supporting multi-dimensional data queries based on production line workstation, equipment type, and time range. The data mapping and storage unit uses a time-series database for persistent storage of production line operation data, implementing value grading of production line data based on a time-degradation mechanism. The data value scoring function is as follows: ,in As the benchmark value score, For time-dependent decay function, For the time since the data was generated, For process attribute coefficients, To access feature coefficients, a differentiated storage strategy is automatically triggered based on the scoring results.

7. A multi-terminal monitoring system for a battery pack production line based on digital twins according to claim 1, characterized in that: The visualization interactive terminal also includes a virtual model presentation unit, which is used to drive the visualization presentation of the digital twin model according to the mapping data. It supports switching between multiple perspective visualization modes, including global view, local workstation focus view, and equipment detail view. The digital twin model includes a multi-precision model with three levels of detail: geometric level, process level, and fault level. The virtual model presentation unit uses a dual triggering mechanism of frustum culling and process hot zone to achieve dynamic rendering, rendering high-precision detail levels only for the model within the user's field of view and the process hot zone.

8. A multi-terminal monitoring system for a battery pack production line based on digital twins according to claim 7, characterized in that: The virtual model rendering unit targets the core processes of battery pack terminal welding and cell assembly. It uses the finite element method (FEM) to calculate the temperature, pressure, and stress fields, and visualizes the physical fields through volumetric cloud rendering and particle system rendering. The FEM employs an intrinsic orthogonal decomposition model for order reduction to achieve real-time acceleration. The reconstructed physical field formula after order reduction is as follows: Reconstruct the physical field at time t. For a reduced-order basis matrix, These are low-dimensional modal coefficients.

9. A multi-terminal monitoring system for a battery pack production line based on digital twins according to claim 1, characterized in that: The virtual-physical interaction control unit supports seamless switching between three modes: production line monitoring, offline debugging, and online control. During the switching process, the work-in-process status is saved through a buffer. In the production line monitoring mode, the physical production line operating status is mapped to the digital twin model in real time. In the offline debugging mode, the control commands only drive the digital twin model to run. In the online control mode, the compliant control commands are sent to the physical production line to drive the physical equipment to run and synchronously update the status of the digital twin model.

10. A multi-terminal monitoring system for a battery pack production line based on digital twins according to claim 1, characterized in that: The multi-mode simulation unit supports simulation modes including real-time operation simulation, process optimization simulation, fault simulation simulation, and production line debugging simulation. In the process optimization simulation mode, the system completes multiple rounds of iterative optimization of process parameters based on intelligent optimization algorithms and outputs the optimal parameter configuration. In the fault simulation mode, the system injects faults into the virtual model through parameter coverage method and deduces the fault propagation chain and impact range. In the production line debugging simulation mode, the system completes equipment layout collision detection and equipment operation cycle simulation verification.

Images