A digital twin ELC-TCL virtual-real interaction method for a carbon-based material manufacturing workshop

Abstract

This invention relates to the fields of intelligent manufacturing and digital twin technology, proposing an edge-distributed ledger collaborative trustworthy closed-loop digital twin virtual-physical interaction method (ELC-TCL) for carbon-based material manufacturing workshops. Specifically, it describes an ELC-TCL virtual-physical interaction method for carbon-based material manufacturing workshops. For high-priority critical data, this invention superimposes digital signatures and hash digest verification in the transmission and storage links to reduce the risks of data forgery, tampering, and replay, ensuring the authenticity and integrity of the data upon which the twin model and on-site control decisions are based. Through an integrated design of "edge perception—data governance—trustworthy ledger—security enhancement," this invention achieves end-to-end trustworthiness and highly reliable collaboration in workshop virtual-physical interaction, from data collection, transmission, and storage to traceability. This significantly improves the transparency of carbon-based material manufacturing processes, enhances process optimization efficiency, and improves anomaly tracing capabilities, demonstrating significant engineering application value and innovation.

Description

Technical Field

[0001] This invention relates to the fields of intelligent manufacturing and digital twin technology, specifically to a virtual-real interaction method for a digital twin ELC-TCL in a carbon-based material manufacturing workshop. Background Technology

[0002] Carbon-based materials are widely used in aerospace, new energy batteries, and high-end electronic devices due to their high strength, high thermal conductivity, corrosion resistance, and lightweight properties. Their manufacturing process typically includes several key steps such as pre-oxidation, high-temperature carbonization, surface treatment, calcination and desulfurization, and precision molding. The high-temperature carbonization stage requires inert atmosphere at 1200-3000℃, which places extremely high demands on the real-time stability of parameters such as furnace temperature gradient and gas flow rate. The precision molding stage relies on tension control and synchronous traction, and any response lag or command deviation may lead to product performance deterioration, deformation, or even scrapping.

[0003] Traditional manufacturing workshops rely on local PLC control and manual inspection, making it difficult to achieve global collaboration across equipment and processes. While digital twin technology has been introduced into manufacturing in recent years, it still faces significant challenges in the field of carbon-based materials.

[0004] 1. High latency in virtual-real interaction: The existing centralized data architecture cannot meet the millisecond-level response requirements for events such as temperature changes and gas leaks during high-temperature carbonization;

[0005] 2. Data synchronization distortion: Common communication protocols (such as standard MQTT) have high overhead, which leads to inaccurate restoration of key process parameters on the virtual side;

[0006] 3. Unreliable command execution: Control commands issued by the virtual model are easily affected by actuator inertia and environmental disturbances at the physical device end, causing action deviations;

[0007] 4. Lack of forward-looking decision-making ability: Existing systems are mostly "state mapping + post-event alarm", which cannot simulate the effects of process adjustment or predict the trend of equipment degradation on the virtual side;

[0008] 5. Failure to optimize deployment for the characteristics of carbon-based manufacturing processes: The requirements for sensing density and response speed vary significantly among process areas such as carbonization zone, molding zone, and roasting and desulfurization zone, making a uniform deployment strategy inefficient.

[0009] Therefore, there is an urgent need for a digital twin approach to virtual-real interaction in carbon-based material manufacturing workshops. This approach requires innovative design across the entire chain, from data acquisition, transmission protocols, model mapping, instruction feedback to predictive maintenance, to achieve five-dimensional collaboration of "perception, transmission, mapping, feedback, and prediction". Summary of the Invention

[0010] (a) Technical issues

[0011] The technical problem solved by this invention: In view of the shortcomings of the prior art, this invention provides a method for virtual-real interaction of digital twin ELC-TCL in a carbon-based material manufacturing workshop. It has the advantages of low virtual-real interaction latency, high data synchronization accuracy, reliable execution of control commands, predictable process adjustment, and predictable equipment status. It effectively solves the technical problems caused by high-temperature and rapid change processes in carbon-based material manufacturing, such as lag in virtual-real response, inaccurate synchronization of key parameters, large deviation in command execution, reliance on trial and error for process debugging, and difficulty in early warning of sudden equipment failures. It realizes efficient, accurate, adaptive, and collaborative management and control of carbon-based material manufacturing workshops under complex working conditions.

[0012] (II) Technical Solution

[0013] To achieve the aforementioned objectives of "low latency in virtual-real interaction, high data synchronization accuracy, reliable execution of control commands, predictable process adjustments, and predictable equipment status," this invention provides the following technical solution: a method for digital twin virtual-real interaction in a carbon-based material manufacturing workshop, comprising the following steps:

[0014] (a) Edge sensing units are deployed in the reactor, traction device, gas control system, temperature control unit and material conveying line in the carbon-based material manufacturing workshop to collect equipment operating status, process parameters and material flow information, and to perform timestamp alignment and redundancy removal on the collected data to generate structured sensing data packets; supplementary constraints: the structured sensing data packets satisfy the data field definition of equation (1); the time synchronization accuracy of its timestamp alignment satisfies equation (3), and the upper limit of the synchronization error is preferably no more than 100μs;

[0015] For the removal of obvious outliers, the threshold for the rate of change of tension satisfies equation (4) (preferred threshold is 20 N / s), and the threshold for the rate of change of temperature satisfies equation (5).

[0016] The elimination of redundant information with multiple sources and the same parameters satisfies the calculation of the "average value of other sensors" within the window of equation (6) and the determination of the deviation threshold of equation (7) (temperature example threshold ±2℃).

[0017] (b) Divide the structured sensing data packet into high-priority data and regular data, wherein the high-priority data includes temperature change signal, tension abnormal signal and gas leakage signal; transmit the high-priority data to the digital twin platform in real time through a dedicated low-latency communication channel, and upload the regular data to the digital twin platform after compression and encoding by the edge computing node according to a preset period; supplementary constraint: the classification determination of the high-priority data and the regular data satisfies the data classification determination function of equation (8);

[0018] The definition of the high-priority data end-to-end delay satisfies equation (9) and satisfies the requirement that the end-to-end transmission delay of claim 3 does not exceed 50ms;

[0019] The batch upload cycle of the regular data satisfies the cycle constraint of equation (10).

[0020] (c) The high-priority data and regular data are synchronously written into a distributed ledger jointly maintained by multiple edge nodes and a digital twin platform. The distributed ledger adopts a lightweight consensus mechanism. The lightweight Byzantine fault-tolerant consensus mechanism requires that the number of faulty nodes among the nodes participating in the consensus does not exceed one-third of the total number of nodes. After receiving the data transaction, each node broadcasts it to other consensus nodes. After a majority of nodes verify the signature and data integrity, a consensus is reached and the data is written into a block. The distributed ledger has no less than 5 consensus nodes and covers at least three different process regions and a digital twin platform. Supplementary constraint: The scale of the consensus nodes satisfies equation (11). And according to claim 1, the distributed ledger has no less than 5 consensus nodes and covers at least three different process regions and a digital twin platform.

[0021] The lightweight Byzantine fault-tolerant mechanism satisfies the upper bound constraint of fault nodes in equation (12) (the number of fault nodes does not exceed one-third of the total number of nodes).

[0022] The write confirmation threshold satisfies the legal number threshold of equation (13) (the main scheme is that most nodes agree, and the preferred implementation is that no less than two-thirds confirm).

[0023] The hash digest and digital signature generation satisfy equation (14), and the signature verification and hash consistency verification satisfy equation (15).

[0024] (d) After receiving the data, the digital twin platform drives the virtual workshop model, which includes the equipment geometric model, the process logic model and the material state model, to be dynamically updated. The three are synchronized in state based on a unified time reference and spatial coordinate system. Supplementary constraints: The state vectors of the three models satisfy equation (16); The synchronization of the three models based on a unified time and space reference satisfies equation (17).

[0025] The relative deviation between the virtual and real states satisfies equation (18), with the preferred deviation being less than 0.5%.

[0026] (e) The edge computing nodes of each process area maintain local virtual sub-models and exchange key status data with adjacent process areas through point-to-point communication protocols;

[0027] (f) Establish a two-way collaborative interaction interface between the virtual workshop model and the physical equipment, perform permission verification, historical behavior comparison and digital signature verification before the control command is issued, and record the execution result back to the distributed ledger; Supplementary constraint: the permission verification, historical behavior comparison and digital signature verification of the control command shall satisfy the triple verification constraint of equation (19).

[0028] (g) In response to the process parameter adjustment request, the process parameter pre-simulation feedback module is called in the virtual workshop model to simulate the impact of parameter adjustment on the manufacturing process and output a pre-simulation report containing product quality indicators, energy consumption indicators and equipment load indicators; the parameter change operation, operator, time, simulation basis and approval record are written into the distributed ledger as a structured transaction; supplementary constraint: the mapping of process parameter pre-simulation and indicator output satisfy equation (20), and the risk judgment satisfies the threshold inequality of equation (21);

[0029] The pre-evolutionary, traceable on-chain transactions satisfy the structured transaction constraints of equation (22).

[0030] (h) Calculate the virtual-real consistency evaluation index, which includes the daily average prediction error, the weekly average prediction error, the state deviation distribution entropy, and the model response lag index; when any index exceeds the preset tolerance for multiple consecutive periods, trigger the model recalibration or online fine-tuning process, and store the evaluation log in the distributed ledger; supplementary constraints: the definition of the model response lag index satisfies equation (23); the triggering criterion when any index exceeds the preset tolerance for multiple consecutive periods satisfies equation (24).

[0031] The daily average prediction error is the average absolute error between the daily predicted values ​​and actual measured values ​​of the physical device state variables by the virtual model. The weekly average prediction error is the arithmetic mean of the daily average prediction errors over seven consecutive days. The state deviation distribution entropy is calculated based on the probability distribution of the deviations of each state variable. The model response lag index is the average time offset between the output of the virtual model and the actual feedback from the physical device. When any indicator exceeds the preset tolerance for multiple consecutive periods, the model recalibration or online fine-tuning process is triggered, and the evaluation log is stored in the distributed ledger.

[0032] (i) The remaining service life of heating elements, tension control rollers, and gas flow regulating valves is calculated by combining the predictive maintenance module integrated into the digital twin platform with the results of thermo-mechanical coupling simulation and the actual operating data of physical equipment. When a quality abnormality or equipment failure is detected, a causal graph is constructed based on the product batch ID, time window, and equipment operation log stored in the distributed ledger. The causal graph has nodes of process, equipment, parameter, and operator, and edges of time sequence and logical dependency. It supports traceability queries by process, parameter type, or responsible person, and can be exported as a structured audit data package containing node attributes, edge relationships, and digital signatures. Supplementary constraint: The definition of the remaining service life of key components satisfies equation (25) (no specific algorithm is limited).

[0033] Preferably, the edge sensing unit is deployed according to the carbon-based manufacturing process region; the carbon-based manufacturing process region includes a raw material processing region, a calcination oxidation and carbonization region, a forging coke mixing region, a surface treatment and compression molding region, and a roasting desulfurization region; each process region is configured with independent edge computing resources, and the edge computing resources serve as consensus nodes of the distributed ledger.

[0034] Preferably, the end-to-end transmission delay of the high-priority data from the edge sensing unit to the digital twin platform does not exceed 50 milliseconds; the high-priority data is appended with a digital signature and hash digest during transmission.

[0035] Preferably, the key status data exchanged by the point-to-point communication protocol includes material batch identifier, outlet tension value, calcining furnace temperature gradient, and gas flow rate setpoint.

[0036] Preferably, the model response hysteresis index is the average time offset between the virtual model output and the actual feedback from the physical device.

[0037] A computer-readable storage medium storing a computer program corresponding to the aforementioned digital twin virtual-real interaction method for a carbon-based material manufacturing workshop, wherein the program, when executed by a processor, implements the method steps for digital twin virtual-real interaction in a carbon-based material manufacturing workshop.

[0038] An electronic device includes at least one processor and a memory, the memory storing a computer program corresponding to the above-described digital twin virtual-real interaction method for a carbon-based material manufacturing workshop, the program being executed by the processor to implement the steps of the digital twin virtual-real interaction method for a carbon-based material manufacturing workshop.

[0039] (III) Beneficial Effects

[0040] Compared with existing technologies, this invention provides a method for digital twin virtual-real interaction in a carbon-based material manufacturing workshop, which has the following beneficial effects:

[0041] 1. The digital twin ELC-TCL virtual-physical interaction method in this carbon-based material manufacturing workshop synchronously writes sensing data and control commands into a distributed ledger jointly maintained by multiple edge nodes and the digital twin platform. It adopts a lightweight Byzantine fault-tolerant consensus mechanism to ensure that all manufacturing data is immutable, verifiable in origin, and traceable throughout its entire lifecycle. High-priority data is appended with digital signatures and hash digests during transmission, effectively preventing data from being forged or tampered with during transmission or storage. This solves the data trust risks caused by single points of failure or internal operations under a centralized architecture, and provides a technically guaranteed trustworthy data foundation for quality auditing, accident accountability, and compliance supervision.

[0042] 2. The digital twin ELC-TCL virtual-physical interaction method in this carbon-based material manufacturing workshop deploys independent edge computing resources in each process area and maintains local virtual sub-models. It realizes the direct exchange of key status data (such as material batch identifiers, outlet tension values, and calcining furnace temperature gradients) between adjacent areas through point-to-point communication protocols. This breaks the traditional "edge → center → edge" information loop, reduces cross-process collaboration delays, and eliminates information asymmetry caused by data version lag in the central platform. Different process teams can carry out collaborative optimization based on consistent and real-time shared data, significantly improving production transparency and response consistency throughout the workshop.

[0043] 3. The digital twin ELC-TCL virtual-physical interaction method of this carbon-based material manufacturing workshop stores process parameter change records, control command execution logs, abnormal events, and model update evaluation results as structured transactions in a distributed ledger, and automatically constructs a causal graph based on product batch ID, time window, and equipment operation logs. When product quality abnormalities or equipment failures occur, multi-dimensional rapid tracing can be performed by process, parameter type, or responsible person to accurately locate the root cause. This avoids the difficulty of accountability caused by the risk of missing or tampered logs in traditional centralized databases, and supports the generation of legally valid audit evidence.

[0044] 4. The digital twin ELC-TCL virtual-real interaction method in this carbon-based material manufacturing workshop introduces a virtual-real consistency quantitative evaluation mechanism to continuously calculate indicators such as daily / weekly average prediction error, state deviation distribution entropy, and model response lag index. When the indicators continuously exceed the limits, the model recalibration or online fine-tuning process is automatically triggered. This mechanism shifts from passive threshold correction to active model health monitoring, effectively identifying model drift caused by changes in raw material batches, environmental disturbances, or equipment aging. This ensures the prediction accuracy and control reliability of the digital twin during long-term operation, providing sustainable high-fidelity virtual-real mapping capabilities for the manufacturing of high-value carbon-based materials. Attached Figure Description

[0045] Figure 1 This is an overall flowchart of the digital twin virtual-real interaction method for carbon-based material manufacturing workshop of the present invention;

[0046] Figure 2 This is a diagram of the regionalized edge perception and distributed ledger architecture for the carbon-based material manufacturing workshop of this invention. Detailed Implementation

[0047] 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.

[0048] A digital twin ELC-TCL virtual-physical interaction method for a carbon-based material manufacturing workshop includes the following steps:

[0049] (a) In the calcination reactor, vibration molding machine traction device, gas control system, temperature control unit and material conveying line key production units in the carbon-based material manufacturing workshop, collect equipment operating status, process parameters and material flow information, and perform timestamp alignment and redundancy removal on the collected data to generate structured sensing data packets; supplementary constraints: the structured sensing data packets satisfy the data field definition of formula (1); the lightweight binary protocol frame structure constraint (carrying high priority data and regular data) satisfies formula (2), its timestamp alignment time synchronization accuracy satisfies formula (3), and the upper limit of synchronization error preferably does not exceed 100μs; for the removal of obvious abnormal values, the tension change rate threshold satisfies formula (4) (preferred threshold 20N / s), and the temperature change rate threshold satisfies formula (5); the removal of redundant information of multiple sources with the same parameters satisfies the calculation of "average value of other sensors" in the window of formula (6) and the deviation threshold judgment of formula (7) (temperature example threshold ±2℃);

[0050] Equation (1) Structured Sensing Data Packet Format Constraints

[0051] ;

[0052] in: This is the k-th structured sensing data packet; For timestamps; For device / node identification; For parameter type; The parameter value; Confidence level; This is a verification field (CRC checksum), used for transmission integrity verification;

[0053] Equation (2) Lightweight binary protocol frame structure constraints (carrying high-priority data and regular data)

[0054] ;

[0055] in: This is the k-th protocol frame; Priority identifier (0x01 = high priority, 0x02 = normal data); The value is in milliseconds (32-bit unsigned integer); type is a 2-byte data type code (e.g., 0x1001 = furnace temperature, 0x1002 = tension, 0x1003 = gas flow rate); payload is the effective payload. For CRC-8 / ATM check bits;

[0056] Equation (3) Time synchronization accuracy constraint

[0057] ;

[0058] in: This refers to the k-th sampling / processing time. Indicates that edge nodes u and v are in Clock deviation; In this embodiment, 100 μs is taken as the upper bound for the synchronization error.

[0059] Equation (4) Clear outlier removal (tension change rate threshold)

[0060] ;

[0061] in: This is the tension measurement value; The time interval between adjacent sampling; left-hand side The rate of change of tension (in N / s); the threshold is 20 N / s.

[0062] Equation (5) Clear outlier removal (temperature change rate threshold)

[0063] ;

[0064] in: This is a temperature measurement value; The time interval between adjacent sampling; left-hand side The temperature change rate (unit: °C / s); the threshold is... It is set by process experience or online calibration;

[0065] Equation (6) Redundancy Removal: Average value of other sensors within the window (implementation details not limited)

[0066] ;

[0067] in: For parameter type; Same region, same parameters A collection of multi-source sensors; For sensors The parameters at time τ The reading; A sliding window (which can be a time window or a sample window, as specified in the manual); This is the "Average Value of Other Sensors" within the window;

[0068] Equation (7) Redundancy Removal: Deviation Threshold Determination (Temperature Example ±2℃)

[0069] ;

[0070] in: For sensors exist At any time, the parameters The reading; The average value is as described in equation (6); The threshold is used; when p = T (temperature), the example threshold is taken as... It can also be set according to the sensor accuracy and process fluctuations;

[0071] (b) The structured sensing data packet is divided into high-priority data and regular data, wherein the high-priority data includes temperature change signals, tension abnormal signals and gas leakage signals; the high-priority data is transmitted to the digital twin platform in real time through a dedicated low-latency communication channel, and the regular data is compressed and encoded by the edge computing node and uploaded to the digital twin platform at a preset period.

[0072] Supplementary constraints: The classification of high-priority data and regular data satisfies the data classification judgment function of equation (8); the definition of the end-to-end delay of high-priority data satisfies equation (9), and meets the requirement that the end-to-end transmission delay does not exceed 50ms in the further specific implementation of this method; the batch upload period of regular data satisfies the period constraint of equation (10);

[0073] Equation (8) Data Classification Judgment Function

[0074] ;

[0075] in: This is a hierarchical function; H represents high-priority data; N represents regular data;

[0076] Equation (9) End-to-end delay definition (high priority)

[0077] ;

[0078] in: The end-to-end delay for the k-th high-priority data; Timestamps for generating data packets for edge sensing units; Analyze and update the timestamps for the digital twin platform to complete model updates;

[0079] Equation (10) Constraints on the Batch Upload Cycle of Conventional Data

[0080] ;

[0081] in: This is a sequence of upload trigger times for regular data; The time interval and upload cycle are defined as the interval between the m-th trigger and the (m+1)-th trigger.

[0082] This is the maximum upload cycle for regular data; This refers to the set of regular data packets generated / arrived between two adjacent upload trigger times; For structured sensing data packets. It is timestamped; thus enabling regular data to be stored within a specified timeframe. Uploading under periodic constraints can result in both batch uploads and short-term periodic uploads.

[0083] (c) The high-priority data and regular data are synchronously written into a distributed ledger jointly maintained by multiple edge nodes and a digital twin platform. The distributed ledger adopts a lightweight consensus mechanism. The lightweight Byzantine fault-tolerant consensus mechanism requires that the number of faulty nodes among the nodes participating in the consensus does not exceed one-third of the total number of nodes. After receiving a data transaction, each node broadcasts it to other consensus nodes. After a majority of nodes verify the signature and data integrity, a consensus is reached and the data is written into a block. The distributed ledger has no less than 5 consensus nodes and covers at least three different process regions and digital twin platforms.

[0084] Supplementary constraints: The consensus node size satisfies equation (11); and according to the above equation (10) of this method, the number of consensus nodes of the distributed ledger is not less than 5, and covers at least three different process regions and digital twin platforms; the lightweight Byzantine fault tolerance mechanism satisfies the upper bound constraint of fault nodes in equation (12) (the number of fault nodes does not exceed one-third of the total number of nodes); the write confirmation threshold satisfies the quorum threshold in equation (13) (the main scheme is majority node consensus, and the preferred implementation is not less than two-thirds confirmation); the hash digest and digital signature generation satisfy equation (14), and the signature verification and hash consistency verification satisfy equation (15);

[0085] Consensus node size constraint in equation (11)

[0086] ;

[0087] Where: n is the number of distributed ledger consensus nodes; and it should cover at least three different process regions and a digital twin platform;

[0088] Equation (12) Byzantine fault tolerance upper bound (faulty nodes do not exceed one-third)

[0089] ;

[0090] in: The number of faulty / abnormal consensus nodes is the tolerable number; n is the total number of consensus nodes; this constraint is equivalent to "the number of faulty nodes does not exceed one-third of the total number of nodes";

[0091] Equation (13) Write confirmation threshold (main scheme: majority of nodes consistent)

[0092] ;

[0093] Write confirmation threshold (preferred implementation: at least two-thirds confirmations)

[0094] ;

[0095] in: For transactions to be written; The number of valid acknowledgment messages received; Let be the number of consensus nodes; the right-hand side of this equation... The quorum threshold for "majority consensus";

[0096] Equation (14) Hash Digest and Digital Signature Generation

[0097] ;

[0098] in: For example, a hash function (such as SHA-256); For data packets The hash digest; For signature functions; This is the private key for node i; This is the signature value;

[0099] Equation (15) Signature verification and hash consistency verification

[0100] ;

[0101] in: This is the signature verification function; Let i be the public key of node i; the right-hand side of the equation Used to recalculate the hash and compare it to ensure that the data has not been tampered with;

[0102] (d) After receiving the data, the digital twin platform drives the virtual workshop model, which includes the equipment geometric model, the process logic model and the material state model, to be dynamically updated. The three are synchronized in state based on a unified time reference and spatial coordinate system.

[0103] Additional constraints: The state vectors of the three models are defined to satisfy equation (16); the unified spatiotemporal reference of the three models is synchronously satisfied to satisfy equation (17); the relative deviation between the virtual and real states satisfies equation (18), preferably less than 0.5%;

[0104] Equation (16) defines the state vector of the three models.

[0105] ;

[0106] in: The state of the equipment's geometric model; This represents the state of the process logic model; The state is the material state model; t is time. Represents geometric model, Representative process model and Represents a material state model;

[0107] Equation (17) Unified Spatiotemporal Reference Synchronization Constraints for the Three Models

[0108] ;

[0109] in: These are the time bases used for updating the three models; The three models use a spatial coordinate system. Unified coordinate system;

[0110] Equation (18) Relative deviation constraint between virtual and real states (deviation < 0.5%)

[0111] ;

[0112] in: These are measured state quantities of physical equipment. The left-hand side represents the predicted value corresponding to the virtual model. This represents a relative deviation; 0.005 corresponds to a 0.5% upper limit.

[0113] (e) The edge computing nodes of each process area maintain local virtual sub-models and exchange key status data with adjacent process areas through point-to-point communication protocols.

[0114] (f) Establish a two-way collaborative interaction interface between the virtual workshop model of the carbon-based material and the physical equipment, perform permission verification, historical behavior comparison and digital signature verification before the control command is issued, and record the execution result back to the distributed ledger; Supplementary constraint: the permission verification, historical behavior comparison and digital signature verification of the control command shall satisfy the triple verification constraint of equation (19);

[0115] Equation (19) commands the issuance of triple verification constraints

[0116] ;

[0117] in: For the first Control commands; For the operator / main entity; This is the permission verification function; For comparing historical behavior; A collection of ledger audit logs; Sign the instruction; For signature verification function, In response to a request to adjust process parameters, the process parameter pre-simulation feedback module is invoked in the virtual workshop model to simulate the impact of parameter adjustments on the manufacturing process and output a pre-simulation report containing product quality indicators, energy consumption indicators, and equipment load indicators; the parameter change operation, operator, time, simulation basis, and approval records are written as structured transactions into the distributed ledger.

[0118] Supplementary constraints: The mapping and index output of the process parameter pre-simulation satisfy equation (20), the risk judgment satisfies the threshold inequality of equation (21), and the pre-simulation change trace on-chain transaction satisfies the structured transaction constraint of equation (22);

[0119] Equation (20) Process parameter pre-mapping and index output

[0120] ;

[0121] in: For the vector of process parameters to be adjusted (such as carbonization temperature and traction speed); This represents the initial state of the simulation (derived from the current operating conditions of the virtual workshop); For multiphysics simulation engine / pre-model; For product quality indicators; Energy consumption indicators; For equipment load indicators;

[0122] Equation (21) Risk Judgment Inequality (Exceeding the Threshold is High Risk)

[0123] ;

[0124] in: This is the lower limit for quality and safety. This is the upper limit of energy consumption; Maximum load; This is a risk level function; "High" indicates high risk.

[0125] Equation (22) Pre-evolution, evolution, traceability, and on-chain transaction constraints

[0126] ;

[0127] in: For the preview report transaction; For parameter changes; As the operator; For timestamps; Hash the simulation report; For approval information; It is a distributed ledger;

[0128] (h) Calculate the virtual-real consistency evaluation index, which includes the daily average prediction error, the weekly average prediction error, the state deviation distribution entropy, and the model response lag index; when any index exceeds the preset tolerance for multiple consecutive periods, trigger the model recalibration or online fine-tuning process, and store the evaluation log in the distributed ledger; supplementary constraints: the definition of the model response lag index satisfies equation (23); the triggering criterion when any index exceeds the preset tolerance for multiple consecutive periods satisfies equation (24);

[0129] Equation (23) Model response lag exponent (mean time offset)

[0130] ;

[0131] in: The model response lag index; For the evaluation time set The number of samples; For virtual model output and physical device feedback at time Time offset;

[0132] Equation (24) Triggering criterion: K consecutive cycles of exceeding the limit trigger recalibration / online fine-tuning (general expression)

[0133] ;

[0134] in: This is an indicator function (1 if the condition is true, 0 otherwise); These are, in order, daily error, weekly error, entropy, and hysteresis tolerance; The number of cycles in "multiple consecutive cycles" (this method can be further refined in specific implementation) (It can also be implemented in scenarios such as four consecutive cycles); For the first The daily average forecast error corresponding to each cycle For the first The weekly average forecast error corresponding to each cycle For the first Entropy of state deviation distribution over a period of time For the first Model response lag index for each period;

[0135] The daily average prediction error is the average absolute error between the daily predicted values ​​and actual measured values ​​of the physical device state variables by the virtual model. The weekly average prediction error is the arithmetic mean of the daily average prediction errors over seven consecutive days. The state deviation distribution entropy is calculated based on the probability distribution of the deviations of each state variable. The model response lag index is the average time offset between the output of the virtual model and the actual feedback from the physical device. When any indicator exceeds the preset tolerance for multiple consecutive periods, the model recalibration or online fine-tuning process is triggered, and the evaluation log is stored in the distributed ledger.

[0136] (i) By integrating the predictive maintenance module into the digital twin platform, and combining the results of thermo-mechanical coupling simulation with the actual operating data of physical equipment, the remaining service life of heating elements, tension control rollers, and gas flow regulating valves is calculated; when a quality abnormality or equipment failure is detected, a causal graph is constructed based on the product batch ID, time window, and equipment operation log stored in the distributed ledger; the causal graph has nodes of process, equipment, parameter, and operator, and edges of time sequence and logical dependency, supports traceability query by process, parameter type, or responsible person, and can be exported as a structured audit data package containing node attributes, edge relationships, and digital signatures;

[0137] Additional constraints: The definition of the remaining service life of critical components satisfies equation (25) (without limiting the specific algorithm);

[0138] Equation (25) Definition of Remaining Useful Life (RUL) (without limiting the specific algorithm) ;

[0139] in: The current moment; The component health index (a degradation characterization quantity obtained by fusing thermo-mechanical coupling simulation and measured data); This is the failure threshold; Indicates the earliest time when the failure condition is met; This represents the remaining service life.

[0140] The edge sensing units are deployed according to the carbon-based manufacturing process regions; the carbon-based manufacturing process regions include raw material processing regions, calcination oxidation and carbonization regions, post-forging coke mixing regions, surface treatment and compression molding regions, and roasting desulfurization regions; each process region is configured with independent edge computing resources, and the edge computing resources serve as consensus nodes of the distributed ledger.

[0141] The end-to-end transmission delay of the high-priority data from the edge sensing unit to the digital twin platform does not exceed 50 milliseconds; the high-priority data is appended with a digital signature and hash digest during transmission.

[0142] The key status data exchanged by the point-to-point communication protocol includes material batch identifier, outlet tension value, furnace temperature gradient, and gas flow rate setpoint.

[0143] The model response hysteresis index is the average time offset between the output of the virtual model and the actual feedback from the physical device.

[0144] A computer-readable storage medium storing a computer program corresponding to the method of the present invention, wherein the computer program, when executed by a processor, implements the steps of the method of the present invention.

[0145] An electronic device includes at least one processor and a memory, the memory storing a computer program corresponding to the method of the present invention, the computer program implementing the steps of the method of the present invention when executed by the processor.

[0146] Example 1

[0147] Please see Figure 1-2 A digital twin ELC-TCL virtual-physical interaction method for a carbon-based material manufacturing workshop includes the following steps:

[0148] (a) Regionalized edge sensing and trusted data encapsulation

[0149] Additional constraints: The field definitions of the structured sensing data packet satisfy equation (1); the time synchronization accuracy satisfies equation (3), and the upper bound of the synchronization error is preferably ≤100μs; outlier removal satisfies equations (4) and (5); redundancy removal satisfies equations (6) and (7).

[0150] Edge sensing units are deployed in key equipment such as calcining furnaces, traction devices, gas control systems, temperature control units, and material conveying lines according to the carbon-based manufacturing process area (the carbon-based manufacturing process area includes the raw material processing area, calcination oxidation and carbonization area, post-forging coke mixing area, surface treatment and compression molding area, and roasting desulfurization area), and independent edge computing resources are configured for each area to achieve localized data acquisition and preprocessing.

[0151] It should be noted that each edge sensing unit includes a sensor array (such as a temperature sensor, tension sensor, pressure sensor, flow meter, encoder, etc.) and an edge computing module.

[0152] The data collection process is as follows:

[0153] Each sensor acquires raw signals in real time at a fixed sampling frequency (e.g., 10Hz);

[0154] The edge computing module timestamps data from different devices according to a unified time base (such as time synchronization based on the IEEE 1588 protocol);

[0155] For multi-source data with the same parameter (such as multiple temperature sensors monitoring the same area), a sliding window mean filtering method is used to remove redundancy. That is, when the reading of a certain sensor deviates from the average value of other sensors in the window by more than a preset threshold (e.g., ±2℃), it is judged as abnormal and removed.

[0156] The processed valid data is encapsulated into structured sensing data packets according to a predefined data structure;

[0157] Its format includes:

[0158] {timestamp, device_id, parameter_type, value, confidence_level, checksum}, where checksum is a CRC checksum used for transmission integrity verification.

[0159] (b) Hierarchical transmission and distributed evidence storage

[0160] Additional constraints: Data classification judgment satisfies equation (8); high priority end-to-end delay definition satisfies equation (9) and is ≤50ms; regular data upload cycle satisfies equation (10).

[0161] If a lightweight binary protocol is used for transmission, the protocol frame structure of high-priority / regular data satisfies equation (2).

[0162] Sensing data is divided into high-priority data (such as temperature changes, tension anomalies, and gas leaks) and regular data. The former is uploaded in real time through a dedicated low-latency channel, and digital signatures and hash digests are attached during transmission to ensure its integrity and verifiable source. All data is simultaneously broadcast to a distributed ledger network composed of edge nodes of each process area and the platform while being uploaded to the digital twin platform. A lightweight Byzantine fault-tolerant consensus mechanism is used to complete the on-chain evidence storage. The distributed ledger has no fewer than 5 consensus nodes and covers at least three different process areas and the digital twin platform.

[0163] The specific process of digital twin interaction is as follows:

[0164] (1) Data acquisition and preprocessing: Each edge sensing unit acquires the equipment operating status, process parameters and material flow information, and performs timestamp alignment and redundancy removal to generate structured sensing data packets;

[0165] (2) Data classification: The data is divided into high-priority data (such as temperature change signals, tension abnormal signals, and gas leakage signals) and routine data;

[0166] (3) Differentiated transmission:

[0167] High-priority data is uploaded to the digital twin platform in real time via a dedicated low-latency communication channel (such as UDP over 5G or industrial Ethernet), with an end-to-end latency of no more than 50 milliseconds;

[0168] Regular data is compressed and encoded at the edge nodes (such as the LZ4 algorithm) and then uploaded in batches every 10 seconds.

[0169] (4) Blockchain writing: All data is submitted to the distributed ledger by the digital twin platform and edge nodes as transaction records. A lightweight Byzantine fault-tolerant consensus mechanism (such as the HotStuff variant) is adopted, requiring that the number of faulty nodes among the nodes participating in the consensus does not exceed one-third of the total number of nodes.

[0170] (5) Model update and feedback: After receiving the data, the digital twin platform drives the virtual workshop model to update dynamically. The three (geometric model, logical model and material model) maintain state synchronization based on a unified spatiotemporal benchmark. Before the control command is issued, the platform performs permission verification, historical behavior comparison and digital signature verification. After the execution result is returned, it is recorded in the ledger.

[0171] (c) Construction of virtual workshops with multi-model fusion

[0172] Constraints: The state vectors of the three models satisfy equation (16), the unified spatiotemporal reference synchronously satisfies equation (17), and the virtual-real relative deviation satisfies equation (18) (preferably <0.5%).

[0173] The digital twin platform is based on a unified spatiotemporal reference and synchronously drives the dynamic updates of equipment geometric models, process logic models and material state models. At the same time, edge nodes of each process area maintain local virtual sub-models and exchange key state data such as material batch identifiers, outlet tension values ​​and furnace temperature gradients with adjacent areas through point-to-point communication protocols to achieve local collaborative optimization.

[0174] The lightweight Byzantine fault-tolerant consensus mechanism adopts a simplified version of the HotStuff protocol, and its core process is as follows:

[0175] Upon receiving a new data transaction, each consensus node first verifies its digital signature (based on the SM2 or ECDSA algorithm) and hash digest.

[0176] If the verification passes, the transaction is broadcast to other nodes;

[0177] After a node collects confirmation messages from at least two-thirds of the nodes, it packages the transaction into a block and writes it to its local ledger.

[0178] After a block is written, it is broadcast to all nodes, thus achieving consensus.

[0179] The digital signature and hash digest are appended in the following way:

[0180] Each sensing data packet is digitally signed by the edge node using its private key before being sent.

[0181] The data packet content is hashed using SHA-256 to generate a digest, which is then transmitted along with the data.

[0182] The recipient verifies the signature using the public key and recalculates the hash value for comparison to ensure that the data has not been tampered with.

[0183] (d) Two-way interaction and instruction auditing

[0184] Constraint Supplement: The control command issuance is subject to triple verification to satisfy equation (19).

[0185] Establish a two-way collaborative interaction interface between the virtual model and the physical device to perform permission verification, historical behavior comparison and digital signature verification before the control command is issued; after the execution result is returned, it is written into the distributed ledger along with the command content, operator and timestamp to form an immutable closed-loop audit log.

[0186] (e) Process parameter simulation and change tracking

[0187] Constraints: Pre-drill mapping and indicator output satisfy equation (20), risk judgment satisfies equation (21), and pre-drill change trace on-chain satisfies equation (22).

[0188] When a process parameter adjustment request is received, the digital twin platform invokes the process parameter pre-simulation feedback module to perform the following steps:

[0189] Parse request parameters: Obtain the process variables to be adjusted (such as carbonization temperature and traction speed) and their target values;

[0190] Load current operating conditions: Extract the current material status (such as material location, degree of carbonization) and equipment status (such as furnace temperature distribution, tension value) from the virtual workshop model as the initial conditions for simulation;

[0191] Driven by a multiphysics simulation engine:

[0192] Use a heat conduction model to predict the evolution of the temperature field;

[0193] Use mechanical models to predict tension distribution and fracture risk;

[0194] Use energy consumption models to estimate electricity and gas consumption;

[0195] Generate quantitative indicators:

[0196] Product quality indicators: predicted tensile strength (MPa), carbonization uniformity (%);

[0197] Energy consumption indicators: total electricity (kWh), natural gas consumption (m³);

[0198] Equipment load indicators: maximum thermal stress of heating element (MPa), peak power of traction motor (kW);

[0199] Risk assessment: If any indicator exceeds the preset safety threshold (e.g., intensity <3.5GPa), it will be marked as "high risk" and an alternative parameter suggestion will be provided;

[0200] Generate a structured simulation report, including the above indicators, simulation basis, and risk level, and send it to the approver;

[0201] Once approved, the parameter change operation, operator, time, simulation report hash value, etc., are encapsulated into a JSON format transaction and written into the distributed ledger as the basis for subsequent execution and auditing.

[0202] (f) Proactive assessment of consistency between virtual and real data

[0203] Constraints: The model response lag exponent satisfies equation (23), and the over-limit trigger recalibration / online fine-tuning for K consecutive cycles satisfies equation (24).

[0204] The system continuously calculates the daily / weekly average prediction error, state deviation distribution entropy, and model response lag index. The model response lag index is the average time offset between the virtual model output and the actual feedback from the physical equipment. When the index exceeds the limit for three consecutive cycles, the system automatically triggers the model recalibration process and saves the evaluation log to the ledger.

[0205] (g) Predictive maintenance and fault tracing using integrated simulation

[0206] Constraint Supplement: The Remaining Useful Life (RUL) is defined to satisfy Equation (25).

[0207] An integrated predictive maintenance module is used to predict the lifespan of key components by combining thermo-mechanical coupling simulation and measured data. When a quality anomaly or equipment failure is detected, a causal graph is constructed based on the product batch ID, time window, and equipment operation log stored in the distributed ledger. The causal graph supports traceability queries by process, parameter type, or responsible person, and can be exported as a structured audit data package.

[0208] To ensure the immutability, verifiability of origin, and full lifecycle traceability of manufacturing data, the system adopts the following technical approach:

[0209] Trustworthy data source: Each edge sensing unit is configured with a unique identity (ID) and pre-installed with a digital certificate; all uploaded data is accompanied by a digital signature generated by the node's private key;

[0210] Data integrity guarantee: Each data packet is appended with an SHA-256 hash digest, which the receiver can verify by comparison whether it has been modified;

[0211] Ledger consistency maintenance: The distributed ledger consists of no fewer than 5 consensus nodes, covering at least three process areas such as raw material processing, pre-oxidation, carbonization, surface treatment, and molding, as well as a digital twin platform, forming a cross-regional consensus network;

[0212] Anti-tampering mechanism: Any modification to data already written to the ledger will cause the subsequent block hash chain to break, thus being detected as invalid by the system;

[0213] Audit trail: All operations (such as parameter adjustment, instruction issuance, and model update) are written into the ledger as structured transactions, supporting queries by time, device, operator, and other dimensions.

[0214] This design ensures that the entire process from data acquisition, transmission, storage to use is transparent, reliable, and auditable.

[0215] Example 2

[0216] To make the technical solution of this invention clearer, the invention will be further described in detail below in conjunction with the actual application scenario of a carbon-based material manufacturing workshop. This embodiment takes a carbon-based material manufacturing production line as an example and fully demonstrates the entire process from sensing deployment, data transmission, model building to interactive control and predictive maintenance, including the following:

[0217] I. System Deployment and Data Acquisition

[0218] In the carbon-based material manufacturing workshop, the carbon-based manufacturing process area is divided according to the typical process flow, including the raw material processing area, the calcination oxidation and carbonization area, the forging coke mixing area, the surface treatment and compression molding area, and the roasting desulfurization area. Each process area is equipped with independent edge computing resources for localized processing of the perceived data within the area, realizing an edge intelligent architecture of "regional autonomy and global collaboration".

[0219] The specific equipment deployment is as follows:

[0220] Calcination reactor (carbonization zone): Infrared temperature sensor array (accuracy ±10℃), tension sensor (range 0-500N, resolution 1N) and natural gas flow and leakage detection module are installed at the inlet, middle and outlet of the 900-1300℃ high temperature carbonization furnace, respectively.

[0221] Traction device (traction and forming area): High-resolution encoders and vibration accelerometers are deployed at the compression force of the main shaft of the vibration forming machine, the traction roller, and the guide roller to monitor the real-time status and mechanical status of the carbon-based material block.

[0222] Gas control system: Pressure transmitters and solenoid valve status monitors are installed on the main pipeline of the protective gas (high-purity natural gas) to ensure atmosphere stability;

[0223] Temperature control unit: integrates a PID temperature controller and a solid-state relay actuator to precisely regulate the temperature of each process section;

[0224] Material conveying line: UHF RFID tags and readers are used to track the location, batch, and flow status of each batch of carbon-based raw materials.

[0225] All edge sensing units are connected to the local edge computing node via industrial gigabit Ethernet. The raw data is timestamped (using the IEEE 1588PTP precision time protocol with a synchronization error of ≤100μs), and obvious outliers are removed (such as tension mutations >20N / s or temperature change rates >10℃ / s are considered as interference noise), generating structured sensing data packets.

[0226] II. Data Hierarchical Transmission Mechanism

[0227] Structured-aware data packets are divided into two categories based on data timeliness and task priority:

[0228] High-priority data includes signals of sudden temperature changes, abnormal tension, and gas leaks, requiring millisecond-level response.

[0229] Routine data: such as periodic equipment status, material location, etc., with a certain delay allowed.

[0230] High-priority data is transmitted to the digital twin platform in real time via a dedicated UDP low-latency communication channel. Before being encapsulated into a blockchain transaction, a digital signature generated by the device's private key and a SHA-256 hash digest are attached to prevent tampering and forgery. Regular data is encoded by the edge computing nodes using the LZ4 compression algorithm and uploaded via TCP at 100ms intervals. Both types of data are encapsulated into blockchain transactions at the edge nodes and broadcast to at least 5 ledger nodes (covering more than 3 process areas and the central platform). After passing a lightweight BFT consensus, the data is written into a block.

[0231] III. Lightweight Communication Protocol Format Design

[0232] To address the stringent requirements of low latency and low overhead in carbon-based manufacturing, this embodiment designs a dedicated lightweight binary communication protocol, the frame structure of which is shown in Table 1 below:

[0233] Table 1

[0234]

[0235] The protocol header consists of the first three fields, with a total length of 7 bytes (1+4+2), which is much shorter than standard MQTT (≥20 bytes) or HTTP (≥100 bytes), effectively improving transmission efficiency.

[0236] IV. 50ms End-to-End Delay Test Verification

[0237] To verify the performance of high-priority data transmission, a sudden temperature event (a jump from 1100℃ to 1150℃) was simulated at the carbonization furnace outlet, and the time taken from the occurrence of the event to the completion of the virtual model update was recorded:

[0238] The edge sensing unit detects a sudden temperature change and generates a structured data packet: 8ms;

[0239] Transmission to the digital twin platform via a dedicated UDP channel (physical distance 50 meters, gigabit industrial switch): 32ms;

[0240] The platform receives, parses, and drives the virtual workshop model update (including synchronization of the three models): 7ms;

[0241] The total end-to-end latency is 47ms, which meets the technical specification of "not exceeding 50 milliseconds".

[0242] This delay level is sufficient to support real-time intervention in high-risk events such as temperature runaway and deformation of carbon-based materials during carbonization, thus preventing the scrapping of the entire batch of products.

[0243] V. Virtual Workshop Model Construction and Dynamic Synchronization

[0244] The digital twin platform is built on the Unity3D engine and a self-developed process simulation kernel to create a 1:1 virtual workshop model, which includes three core sub-models:

[0245] Equipment geometric model: accurately reproduces the internal structure of the carbonization furnace, the layout of heating elements, and the diameter (Φ150mm) and spacing of the traction rollers;

[0246] Process logic model:

[0247] Material state model: Real-time tracking of the carbonization progress, current tension, and spatial coordinates (X / Y / Z) of each roll of carbon-based material.

[0248] The three are based on a unified spatiotemporal benchmark: the time axis is synchronized in milliseconds, and the spatial coordinate system takes the workshop floor as the origin, ensuring that the actions of physical equipment (such as the acceleration of traction rollers) are presented synchronously on the virtual side, and the deviation between the virtual and real states is controlled within 0.5%.

[0249] VI. Two-way collaborative interaction and process parameter simulation

[0250] A two-way collaborative interaction interface is established between the virtual workshop model and the physical equipment to support the secure issuance and execution feedback loop of control commands. When an operator requests to increase the carbonization temperature from 1100℃ to 1150℃, the system generates a pre-simulation report. After the operator confirms, the parameter combination, operator ID, timestamp, and simulation basis are serialized into a JSON transaction and submitted to the distributed ledger. The transaction can only take effect after being jointly verified by the edge nodes of the carbonization zone and the forming zone.

[0251] The specific execution flow of the pre-rehearsal feedback module is as follows:

[0252] Parse request parameters: Obtain the process variables to be adjusted (such as carbonization temperature and traction speed) and their target values;

[0253] Load current working conditions: Extract the current material status (such as the location of carbon-based materials and the length of carbonized material) and equipment status (such as furnace temperature distribution and tension value) from the virtual workshop model as the initial conditions for simulation;

[0254] Driven by a multiphysics simulation engine:

[0255] Use a heat conduction model to predict the evolution of the temperature field;

[0256] Use mechanical models to predict tension distribution and fracture risk;

[0257] Use energy consumption models to estimate electricity and gas consumption;

[0258] Generate quantitative indicators:

[0259] Product quality indicators: predicted tensile strength (MPa), carbonization uniformity (%);

[0260] Energy consumption indicators: total electricity (kWh), natural gas consumption (m³);

[0261] Equipment load indicators: maximum thermal stress of heating element (MPa), peak power of traction motor (kW);

[0262] Risk assessment: If any indicator exceeds the preset safety threshold (e.g., intensity <3.5GPa), it will be marked as "high risk" and an alternative parameter suggestion will be provided;

[0263] Generate a structured simulation report, including the above indicators, simulation basis, and risk level, and send it to the approver;

[0264] Once approved, the parameter change operation, operator, time, simulation report hash value, etc., are encapsulated into a JSON format transaction and written into the distributed ledger as the basis for subsequent execution and auditing.

[0265] VII. Abnormal Interaction Validation and Dynamic Command Correction

[0266] During a traction speed adjustment, due to aging of the traction roller bearings, the actual rotational speed lagged behind the target value by 8%. The system continuously compared the virtual predicted rotational speed with the actual feedback. Once the deviation exceeded a 5% threshold, an abnormal interactive verification process was triggered.

[0267] Analysis of the device's historical execution records over the past 30 days revealed an average response delay of 120ms, a steady-state error of +6.5%, and high sensitivity to environmental disturbances.

[0268] Dynamically adjust the next instruction: increase the target speed by 7% and send it 120ms in advance;

[0269] At the same time, update the behavior logic of the device in the virtual model and add an inertia compensation factor.

[0270] After correction, the deviation of the subsequent three instruction executions was controlled within ±1%, significantly improving control robustness and demonstrating the effectiveness of the "dynamic correction based on historical response characteristics" mechanism.

[0271] VIII. Predictive Maintenance Based on Thermo-Mechanical Coupled Simulation

[0272] The predictive maintenance module integrates two types of data sources:

[0273] Virtual side: Thermo-mechanical coupling simulation of molybdenum heating element of carburizing furnace is performed using ANSYS Workbench, inputting real-time temperature field and outputting thermal stress distribution cloud map;

[0274] Physical side: Collect operational data such as heating current, surface infrared temperature, and vibration spectrum.

[0275] By combining a pre-stored library of typical failure modes in carbon-based manufacturing (including 12 types of failure modes such as "calcining furnace damage", "tension roller bearing wear", and "gas valve seal failure"), and using an LSTM neural network to fuse simulation and measured data, the remaining service life of a certain heating element was predicted to be 86 hours. When the service life falls below the 100-hour threshold, the system automatically generates a maintenance work order and pushes it to the MES system, arranging a shutdown for replacement in advance. This successfully avoided an unplanned downtime accident and verified the life prediction method that combines "data-driven and model-driven" approaches.

[0276] If the tensile strength of a certain batch of products fails to meet the standard, the system automatically queries the distributed ledger and finds that the tension setting value of that batch in the pre-oxidation zone has been modified without simulation. Based on the product batch ID, time window (±2 hours) and equipment operation log, the system constructs a cause-effect graph, locates the root cause as the operator in the pre-oxidation zone unauthorizedly adjusted the parameters, and exports an audit data package containing a complete chain of evidence (including digital signature, timestamp, and operation record), which can be used for internal accountability or third-party audit.

[0277] IX. Virtual-to-Real Consistency Assessment and Model Recalibration

[0278] The system continuously calculates the daily / weekly average prediction error, state deviation distribution entropy, and model response hysteresis index (defined as the average time offset between virtual output and physical response). When the weekly average error of the traction tension in the carbonization zone reaches 4.6N (exceeding the limit of 3.5N) and the hysteresis index is 92ms (exceeding the limit of 80ms), exceeding the limit for four consecutive cycles, the system automatically triggers model recalibration: the original model is frozen, fine-tuned online based on 72 hours of measured data, a new version v3.2 is generated, deployed after shadow verification, and the entire process log is written to the distributed ledger (block #31087).

[0279] 10. Causal Graph Construction and Audit Tracing

[0280] For the abnormal batch CF20251128-A7, the system extracts parameter changes, instructions, and operation logs from the distributed ledger using the batch ID and time window (±2 hours). Based on preset rules (such as "tension mutation without rehearsal >10% → tension out of control"), it automatically generates a causal graph. The nodes include processes, equipment, parameters, and responsible persons. It supports tracing by process, parameter type, or responsible person and exports the causal graph as a structured audit data package (JSON format), which includes transaction hashes, digital signatures, operation contexts, and graph topology, and has legal effect.

[0281] This embodiment fully implements the digital twin virtual-real interaction method for carbon-based material manufacturing workshops described in this invention. Through field testing, it verifies the core functions such as lightweight protocol (7-byte header), low-latency transmission (47ms), multi-model synchronization (deviation <0.5%), simulation pre-playback, adaptive correction, and predictive maintenance of fused simulation. For the first time, it has been applied in an industrial-grade carbon-based production line to implement a distributed ledger-based virtual-real consistency quantitative evaluation mechanism and an auditable causal traceability system, fully supporting the technical implementation of claims 1–7. This fully demonstrates that this invention has outstanding substantive features, significant technological progress, and good industrial application prospects.

[0282] This invention belongs to the field of intelligent manufacturing and digital twin technology, proposing an edge-distributed ledger collaborative trusted closed-loop digital twin virtual-real interaction method (ELC-TCL) for carbon-based material manufacturing workshops. This method deploys edge sensing units to collect real-time equipment operating status, process parameters, and material flow information for key production units such as calcining reactors, vibration molding machines, gas control systems, temperature control units, and material conveying lines. It performs timestamp alignment, noise suppression, and redundancy removal on multi-source collected data to form structured sensing data packets that can directly drive the twin model, thereby improving the real-time performance and consistency of virtual-real mapping. Furthermore, this invention constructs a distributed trusted ledger jointly maintained by multiple edge nodes and a digital twin platform, synchronously uploading sensing data and control commands to the blockchain for evidence storage, and introducing a lightweight Byzantine fault-tolerant consensus mechanism to achieve a trusted closed loop for manufacturing data that is "verifiable in origin, tamper-proof, and traceable throughout its entire lifecycle," solving the pain points of traditional twin systems in cross-node collaboration, data consistency, and responsibility definition. For high-priority critical data, this invention superimposes digital signatures and hash digest verification in the transmission and storage links to reduce the risks of data forgery, tampering, and replay, ensuring the authenticity and integrity of the data upon which the twin model and on-site control decisions are based. Through an integrated design of "edge sensing—data governance—trusted ledger—security enhancement," this invention achieves end-to-end trustworthiness and highly reliable collaboration in workshop virtual-physical interaction, from data collection, transmission, and storage to traceability. It can significantly improve the transparency of carbon-based material manufacturing processes, enhance process optimization efficiency, and improve anomaly tracing capabilities, demonstrating significant engineering application value and innovation.

[0283] It should be noted that, in this document, relational terms such as "first" and "second" are used merely 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 a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0284] 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 ELC-TCL virtual-physical interaction method for a carbon-based material manufacturing workshop, characterized in that: Includes the following steps: (a) In the calcination reactor, vibration molding machine traction device, gas control system, temperature control unit and material conveying line key production units in the carbon-based material manufacturing workshop, collect equipment operating status, process parameters and material flow information, and perform timestamp alignment and redundancy removal on the collected data to generate structured sensing data packets; supplementary constraints: the structured sensing data packets satisfy the data field definition of formula (1); the lightweight binary protocol frame structure constraint satisfies formula (2), its timestamp alignment time synchronization accuracy satisfies formula (3), and the upper limit of synchronization error preferably does not exceed 100μs; for the removal of obvious outliers, the tension change rate threshold satisfies formula (4), the temperature change rate threshold satisfies formula (5); the removal of redundant information of multiple sources with the same parameters satisfies the calculation of "average value of other sensors" in the window of formula (6) and the deviation threshold judgment of formula (7); Equation (1) Structured Sensing Data Packet Format Constraints ； in: This is the k-th structured sensing data packet; For timestamps; For device / node identification; For parameter type; The parameter value; Confidence level; This is a validation field used for transmission integrity verification; Equation (2) Lightweight binary protocol frame structure constraints ； in: This is the kth protocol frame; Priority identifier (); The value is in milliseconds (); type is a 2-byte data type code (); payload is the payload; For CRC-8 / ATM check bits; Equation (3) Time synchronization accuracy constraint ； in: This refers to the k-th sampling / processing time. Indicates that edge nodes u and v are in Clock deviation; In this embodiment, 100 μs is taken as the upper bound for the synchronization error. Equation (4) Noticeable outlier removal ； in: This is the tension measurement value; The time interval between adjacent sampling; left-hand side The rate of change of tension (in N / s); the threshold is 20 N / s. Equation (5) Removal of obvious outliers ； in: This is a temperature measurement value; The time interval between adjacent sampling; left-hand side The temperature change rate (unit: °C / s); the threshold is... It is set by process experience or online calibration; Equation (6) Redundancy Removal: Average value of other sensors within the window ； in: For parameter type; Same region, same parameters A collection of multi-source sensors; For sensors The parameters at time τ The reading; A sliding window (which can be a time window or a sample window, as specified in the manual); The "Average Value of Other Sensors" within the window; Equation (7) Redundancy Removal: Deviation Threshold Determination ； in: For sensors exist At any time, the parameters The reading; The average value is as described in equation (6); The threshold is used; when p = T (temperature), the example threshold is taken as... It can also be set according to the sensor accuracy and process fluctuations; (b) The structured sensing data packet is divided into high-priority data and regular data, wherein the high-priority data includes temperature change signals, tension abnormal signals and gas leakage signals; the high-priority data is transmitted to the digital twin platform in real time through a dedicated low-latency communication channel, and the regular data is compressed and encoded by the edge computing node and uploaded to the digital twin platform at a preset period. Supplementary constraints: The classification of high-priority data and regular data satisfies the data classification judgment function of equation (8); the definition of the end-to-end delay of high-priority data satisfies equation (9), and satisfies the requirement that the end-to-end transmission delay does not exceed 50ms in the further specific implementation of this method; the batch upload period of regular data satisfies the period constraint of equation (10); Equation (8) Data Classification Judgment Function ； in: This is a hierarchical function; H represents high-priority data; N represents regular data; Equation (9) End-to-end delay definition ； in: The end-to-end delay for the k-th high-priority data; Timestamps for generating data packets for edge sensing units; Analyze and update the timestamps for the digital twin platform to complete model updates; Equation (10) Constraints on the Batch Upload Cycle of Conventional Data ； in: This is a sequence of upload trigger times for regular data; The time interval and upload cycle are defined as the interval between the m-th trigger and the (m+1)-th trigger. This is the maximum upload cycle for regular data; This refers to the set of regular data packets generated / arrived between two adjacent upload trigger times; For structured sensing data packets. It is timestamped; thus enabling regular data to be stored within a specified timeframe. Uploading under periodic constraints can result in both batch uploads and short-term periodic uploads. (c) The high-priority data and regular data are synchronously written into a distributed ledger jointly maintained by multiple edge nodes and a digital twin platform. The distributed ledger adopts a lightweight consensus mechanism. The lightweight Byzantine fault-tolerant consensus mechanism requires that the number of faulty nodes among the nodes participating in the consensus does not exceed one-third of the total number of nodes. After receiving a data transaction, each node broadcasts it to other consensus nodes. After a majority of nodes verify the signature and data integrity, a consensus is reached and the data is written into a block. The distributed ledger has no less than 5 consensus nodes and covers at least three different process regions and digital twin platforms. Supplementary constraints: The scale of consensus nodes satisfies equation (11); and according to the above equation (10) of this method, the number of consensus nodes of the distributed ledger is not less than 5, and covers at least three different process regions and digital twin platforms; the lightweight Byzantine fault tolerance mechanism satisfies the upper bound constraint of fault nodes in equation (12); the write confirmation threshold satisfies the quorum threshold in equation (13); the hash digest and digital signature generation satisfy equation (14), and the signature verification and hash consistency verification satisfy equation (15); Consensus node size constraint in equation (11) ； Where: n is the number of distributed ledger consensus nodes; and it should cover at least three different process regions and a digital twin platform; Equation (12) Byzantine Fault Tolerance Upper Bound ； in: The number of faulty / abnormal consensus nodes is the tolerable number; n is the total number of consensus nodes; this constraint is equivalent to "the number of faulty nodes does not exceed one-third of the total number of nodes"; Equation (13) write confirmation threshold ； Write confirmation threshold ； in: For transactions to be written; The number of valid acknowledgment messages received; The number of consensus nodes; the right-hand side of this equation The quorum threshold for "majority consensus"; Equation (14) Hash Digest and Digital Signature Generation ； in: The hash function is (). For data packets The hash digest; For signature functions; This is the private key for node i; This is the signature value; Equation (15) Signature verification and hash consistency verification ； in: This is the signature verification function; Let i be the public key of node i; the right-hand side of the equation Used to recalculate the hash and compare it to ensure that the data has not been tampered with; (d) After receiving the data, the digital twin platform drives the virtual workshop model, which includes the equipment geometric model, the process logic model and the material state model, to be dynamically updated. The three are synchronized in state based on a unified time reference and spatial coordinate system. Additional constraints: The state vectors of the three models are defined to satisfy equation (16); the unified spatiotemporal reference of the three models is synchronously satisfied to satisfy equation (17); the relative deviation between the virtual and real states satisfies equation (18), preferably less than 0.5%; Equation (16) defines the state vector of the three models. ； in: The state of the equipment's geometric model; This represents the state of the process logic model; The state is the material state model; t is time. Represents geometric model, Representative process model and Represents a material state model; Equation (17) Unified Spatiotemporal Reference Synchronization Constraints for the Three Models ； in: These are the time bases used for updating the three models; The three models use a spatial coordinate system. Unified coordinate system; Equation (18) Relative Deviation Constraint between Virtual and Real States ； in: These are measured state quantities of physical equipment. The left-hand side represents the predicted value corresponding to the virtual model. This represents a relative deviation; 0.005 corresponds to a 0.5% upper limit. (e) The edge computing nodes of each process area maintain local virtual sub-models and exchange key status data with adjacent process areas through point-to-point communication protocols. (f) Establish a two-way collaborative interaction interface between the virtual workshop model of the carbon-based material and the physical equipment, perform permission verification, historical behavior comparison and digital signature verification before the control command is issued, and record the execution result back to the distributed ledger; Supplementary constraint: the permission verification, historical behavior comparison and digital signature verification of the control command shall satisfy the triple verification constraint of equation (19); Equation (19) commands the issuance of triple verification constraints ； in: For the first Control commands; For the operator / main entity; This is the permission verification function; For comparing historical behavior; A collection of ledger audit logs; Sign the instruction; For signature verification function, In response to a request to adjust process parameters, the process parameter pre-simulation feedback module is invoked in the virtual workshop model to simulate the impact of parameter adjustments on the manufacturing process and output a pre-simulation report containing product quality indicators, energy consumption indicators, and equipment load indicators; the parameter change operation, operator, time, simulation basis, and approval records are written as structured transactions into the distributed ledger. Supplementary constraints: The mapping and index output of the process parameter pre-simulation satisfy equation (20), the risk judgment satisfies the threshold inequality of equation (21), and the pre-simulation change trace on-chain transaction satisfies the structured transaction constraint of equation (22); Equation (20) Process parameter pre-mapping and index output ； in: This is the vector of process parameters to be adjusted; This represents the initial state of the simulation. For multiphysics simulation engine / pre-model; For product quality indicators; Energy consumption indicators; For equipment load indicators; Equation (21) Risk Judgment Inequality ； Where: is the lower limit of quality safety; is the upper limit of energy consumption; is the upper limit of load; is the risk level function; "High" indicates high risk; Equation (22) Pre-evolution, evolution, traceability, and on-chain transaction constraints ； in: For the preview report transaction; For parameter changes; As the operator; For timestamps; Hash for the simulation report; For approval information; It is a distributed ledger; (h) Calculate the virtual-real consistency evaluation index, which includes the daily average prediction error, the weekly average prediction error, the state deviation distribution entropy, and the model response lag index; when any index exceeds the preset tolerance for multiple consecutive periods, trigger the model recalibration or online fine-tuning process, and store the evaluation log in the distributed ledger; supplementary constraints: the definition of the model response lag index satisfies equation (23); the triggering criterion when any index exceeds the preset tolerance for multiple consecutive periods satisfies equation (24); Equation (23) Model Response Lag Index ； in: The model response lag index; For the evaluation time set The number of samples; For virtual model output and physical device feedback at time Time offset; Equation (24) Trigger criterion: K consecutive cycles of exceeding the limit trigger recalibration / online fine-tuning ； in: This is an indicator function (where 1 is true and 0 is false); These are, in order, daily error, weekly error, entropy, and hysteresis tolerance; The number of cycles in a series of consecutive cycles; For the first The daily average forecast error corresponding to each cycle For the first The weekly average forecast error corresponding to each cycle For the first Entropy of state deviation distribution over a period of time For the first Model response lag index for each period; The daily average prediction error is the average absolute error between the daily predicted values ​​and actual measured values ​​of the physical device state variables by the virtual model. The weekly average prediction error is the arithmetic mean of the daily average prediction errors over seven consecutive days. The state deviation distribution entropy is calculated based on the probability distribution of the deviations of each state variable. The model response lag index is the average time offset between the output of the virtual model and the actual feedback from the physical device. When any indicator exceeds the preset tolerance for multiple consecutive periods, the model recalibration or online fine-tuning process is triggered, and the evaluation log is stored in the distributed ledger. (i) By integrating the predictive maintenance module into the digital twin platform, and combining the results of thermo-mechanical coupling simulation with the actual operating data of physical equipment, the remaining service life of heating elements, tension control rollers, and gas flow regulating valves is calculated; when a quality abnormality or equipment failure is detected, a causal graph is constructed based on the product batch ID, time window, and equipment operation log stored in the distributed ledger; the causal graph has nodes of process, equipment, parameter, and operator, and edges of time sequence and logical dependency, supports traceability query by process, parameter type, or responsible person, and can be exported as a structured audit data package containing node attributes, edge relationships, and digital signatures; Supplementary constraint: The definition of the remaining service life of critical components satisfies equation (25); Equation (25) Definition of remaining useful life ; in: The current moment; The component health index (a degradation characterization quantity obtained by fusing thermo-mechanical coupling simulation and measured data); This is the failure threshold; Indicates the earliest time when the failure condition is met; This represents the remaining service life.

2. The digital twin ELC-TCL virtual-real interaction method for a carbon-based material manufacturing workshop according to claim 1, characterized in that: The edge sensing units are deployed according to the carbon-based manufacturing process regions; the carbon-based manufacturing process regions include raw material processing regions, calcination oxidation and carbonization regions, post-forging coke mixing regions, surface treatment and compression molding regions, and roasting desulfurization regions; each process region is configured with independent edge computing resources, and the edge computing resources serve as consensus nodes of the distributed ledger.

3. The digital twin ELC-TCL virtual-real interaction method for a carbon-based material manufacturing workshop according to claim 1, characterized in that: The end-to-end transmission delay of the high-priority data from the edge sensing unit to the digital twin platform does not exceed 50 milliseconds; the high-priority data is appended with a digital signature and hash digest during transmission.

4. The digital twin ELC-TCL virtual-real interaction method for a carbon-based material manufacturing workshop according to claim 1, characterized in that: The key status data exchanged by the point-to-point communication protocol includes material batch identifier, outlet tension value, furnace temperature gradient, and gas flow rate setpoint.

5. The digital twin ELC-TCL virtual-real interaction method for a carbon-based material manufacturing workshop according to claim 1, characterized in that: The model response hysteresis index is the average time offset between the output of the virtual model and the actual feedback from the physical device.

6. A computer-readable storage medium, characterized in that: The storage medium stores a computer program corresponding to the method of any one of claims 1-5, which, when executed by a processor, implements the steps of the method of any one of claims 1-5.

7. An electronic device, characterized in that: It includes at least one processor and a memory, the memory storing a computer program corresponding to the method of any one of claims 1-5, the computer program being executed by the processor to implement the steps of any one of claims 1 to 5.

Images