Information physical fusion driven digital twin model real-time linkage method

Abstract

The invention discloses a digital twinborn model real-time linkage method driven by information physics fusion, particularly relates to the technical field of digital twinborn cooperative control of an industrial automation production line, and is used for solving the problems of instruction conflict and control failure caused by mismatching of virtual model parameters and dynamic capability of physical equipment in the prior art. According to the method, equipment state and material flow data are collected in real time, equipment dynamic degradation parameters are extracted to generate capability attenuation feature vectors, hidden process conflict path detection and resource preemption probability simulation are combined, a collaborative optimization model of equipment health and process scheduling is constructed, and a control instruction set containing dynamic capability constraints is generated. After the screening instruction is verified through physical constraint matching, parameters are corrected in a closed loop mode based on an execution result, a model is iterated, and self-adaptive matching of the virtual instruction and the physical equipment capacity is achieved; the reliability of the digital twinning control instruction and the stability of a production line are remarkably improved, and the risk of abnormal shutdown caused by an overrun instruction is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of digital twin collaborative control of industrial automation production lines. More specifically, the present invention relates to a real-time linkage method for digital twin models driven by information-physical fusion. Background Art

[0002] In the field of industrial automation, digital twin technology can be used to build virtual mapping models of physical production lines, enabling real-time monitoring and optimization of the production process. In existing technologies, cyber-physical systems (CPS) drive digital twin models by collecting data such as equipment status and material flow, and generate control instructions based on model simulation results to optimize production line operations. However, when implementing closed-loop control between virtual models and physical equipment, one-way or simplified two-way interaction mechanisms are typically relied upon. Specifically, the optimization instructions generated by the virtual model (such as equipment operating parameter adjustments and path planning) are often based on idealized assumptions and do not fully consider the dynamic state changes of the physical equipment (such as mechanical wear and loss of execution accuracy). This leads to a deviation between the control instructions and the actual execution capabilities of the equipment.

[0003] Against this backdrop, existing technologies have the following flaws: when a digital twin model sends optimization instructions to a physical device through closed-loop feedback, the instructions may exceed the device's current physical conditions (such as mechanical structure limits and dynamic response thresholds) due to a dynamic mismatch between the model parameters and the device's actual state, leading to control instruction conflicts. For example, the virtual model requires a robotic arm to perform a task at a specific acceleration, but the actual device cannot respond to the instruction due to aging, resulting in control failure or even abnormal shutdown of the device. Such problems are particularly prominent in bidirectional control scenarios requiring high frequency and strong real-time performance, seriously affecting the reliability and practicality of digital twin technology in industrial control. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a real-time linkage method of a digital twin model driven by information-physical fusion to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A real-time linkage method for digital twin models driven by cyber-physical fusion includes the following steps:

[0007] S1. Real-time collection of equipment status data and material flow data from the physical production line and synchronization to the digital twin model;

[0008] S2. Extracting dynamic degradation parameters of equipment in the physical production line based on equipment status data, and generating equipment capability attenuation feature vectors based on the dynamic degradation parameters;

[0009] S3. Based on the dynamic degradation parameters and material flow data, identify the implicit process conflict paths through a constraint satisfaction problem model, and generate the material scheduling priority coefficients by combining Monte Carlo tree search to simulate the resource preemption probability;

[0010] S4. Input the equipment capacity attenuation feature vector and the material scheduling priority coefficients into the digital twin model to generate an optimized control instruction set including equipment dynamic capacity constraints;

[0011] S5. Perform physical constraint matching verification on the optimized control instruction set. When the verification passes, screen out the effective control instructions and send them to the physical production line;

[0012] S6. Update the dynamic degradation parameters and the material scheduling priority coefficients according to the execution results of the effective control instructions by the physical production line.

[0013] In a preferred embodiment, the equipment status data and material flow data of the physical production line are collected in real time and synchronized to the digital twin model, including:

[0014] Collect the equipment status data of the physical production line in real time, including the equipment vibration amplitude, temperature gradient, and working current parameters;

[0015] Collect the material flow data of the physical production line in real time, including the material position coordinates, movement speed vector, and transmission direction angle;

[0016] Synchronize the equipment status data and the material flow data to the digital twin model, where the synchronization process includes timestamp alignment of multi-source data and unified spatial coordinate system conversion.

[0017] In a preferred embodiment, extract the dynamic degradation parameters of the equipment in the physical production line based on the equipment status data, and generate an equipment capacity attenuation feature vector based on the dynamic degradation parameters, including:

[0018] Based on the equipment vibration amplitude, temperature gradient, and working current parameters, extract the peak value of the equipment vibration time-domain feature, the temperature gradient change rate, and the current fluctuation amplitude-frequency characteristic as the dynamic degradation parameters;

[0019] Calculate the vibration degradation weight coefficient according to the difference between the peak value of the equipment vibration time-domain feature and the preset vibration reference value;

[0020] Determine the temperature degradation influence factor according to the correlation between the temperature gradient change rate and the equipment heat dissipation efficiency;

[0021] Generate the current load attenuation index by combining the current fluctuation amplitude-frequency characteristic and the ratio of the equipment rated load;

[0022] Perform data dimensionality reduction processing on the vibration degradation weight coefficient, the temperature degradation influence factor, and the current load attenuation index to generate an equipment capacity attenuation feature vector.

[0023] In a preferred embodiment, based on the dynamic degradation parameters and the material flow data, the implicit process conflict paths are identified through a constraint satisfaction problem model, and the material scheduling priority coefficient is generated by combining the Monte Carlo tree search to simulate the resource preemption probability, including:

[0024] Based on the dynamic degradation parameters and the material flow data, a process conflict detection model including equipment health constraints and material spatio-temporal constraints is constructed;

[0025] Through the constraint satisfaction problem model, traverse the production line process topology network to identify the implicit process conflict paths that violate the equipment health constraints or the material spatio-temporal constraints;

[0026] Based on the Monte Carlo tree search, simulate the resource preemption actions in the implicit process conflict paths, and count the resource preemption success probabilities of each implicit process conflict path;

[0027] Fuse the resource preemption success probability and the severity weight of the implicit process conflict path through weighted fusion to generate the material scheduling priority coefficient.

[0028] In a preferred embodiment, the severity weight is dynamically adjusted based on the correlation strength between the equipment capacity attenuation feature vector and the conflict path.

[0029] In a preferred embodiment, the equipment capacity attenuation feature vector and the material scheduling priority coefficient are input into the digital twin model to generate an optimized control instruction set including equipment dynamic capacity constraints, including:

[0030] Input the equipment capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model;

[0031] Define the equipment dynamic capacity constraints based on the equipment capacity attenuation feature vector, including the upper limit of the peak value of the equipment vibration time domain feature, the tolerance range of the temperature gradient change rate, and the threshold of the current load attenuation index;

[0032] Combine the equipment dynamic capacity constraints and the material scheduling priority coefficient, and generate the equipment operation parameter adjustment instruction and the material path planning instruction through a multi-objective optimization algorithm;

[0033] Resolve the conflicts of the equipment operation parameter adjustment instruction and the material path planning instruction. The conflict resolution includes dynamically adjusting the instruction priority and eliminating the instructions that violate the material spatio-temporal constraints;

[0034] Screen the instruction combinations that meet the equipment dynamic capacity constraints and the production line beat synchronization requirements to form an optimized control instruction set.

[0035] In a preferred embodiment, physical constraint matching verification is performed on the optimized control instruction set. When the verification passes, valid control instructions are screened and sent to the physical production line, including:

[0036] Performing physical constraint matching verification on the optimized control instruction set includes calculating the trajectory deviation rate under dynamic load sensitivity based on dynamic degradation parameters, and analyzing whether the safety margin of the kinematic chain joint angle meets the standard based on the production line layout topology;

[0037] When the trajectory deviation rate is lower than the trajectory deviation rate threshold and the safety margin meets the standard, it is determined that the physical constraint matching verification passes, and valid control instructions are screened and sent to the physical production line;

[0038] During the screening process, instructions with an excessive trajectory deviation rate or a safety margin that does not meet the standard are eliminated, and instructions that meet the dynamic capacity constraints of the equipment are retained.

[0039] In a preferred embodiment, the trajectory deviation rate under dynamic load sensitivity is calculated by the difference between the real-time equipment load rate and the expected load of the instruction, and the trajectory deviation rate threshold is dynamically adjusted according to the equipment capacity attenuation characteristic vector.

[0040] In a preferred embodiment, the safety margin of the kinematic chain joint angle is calculated by the difference between the current joint angle and the mechanical structure limit angle, and the condition for the safety margin to meet the standard is that the difference is greater than the preset safety threshold.

[0041] In a preferred embodiment, according to the execution results of the valid control instructions on the physical production line, the dynamic degradation parameters and the material scheduling priority coefficient are updated, including:

[0042] Obtaining the execution result data of the valid control instructions on the physical production line, including the actual equipment load rate, the joint angle deviation value, and the process beat synchronization error;

[0043] Updating the dynamic degradation parameters based on the execution result data, including correcting the peak value of the vibration time domain characteristic by the actual equipment load rate, correcting the temperature gradient change rate by the joint angle deviation value, and correcting the current load attenuation index by the process beat synchronization error;

[0044] Regenerating the equipment capacity attenuation characteristic vector based on the updated dynamic degradation parameters;

[0045] Combining the process beat synchronization error and the historical execution effect, updating the material scheduling priority coefficient through a regression model;

[0046] Synchronizing the updated equipment capacity attenuation characteristic vector and the material scheduling priority coefficient to the digital twin model to generate the optimized control instruction set for the next cycle.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. By deeply integrating the device status and material flow data, a dynamic optimization system with virtual-real linkage is constructed to achieve the precise matching of production line control instructions and the actual execution capabilities of devices; the identification process of device health degradation parameters and process conflict paths is deeply coupled to generate an optimized instruction set containing dynamic capacity constraints in real time; through the multi-dimensional characterization of the device capacity attenuation eigenvector, the comprehensive effects of mechanical wear, heat dissipation degradation, and load attenuation are quantified to ensure that the current performance boundaries of devices are fully considered when generating instructions; combined with a two-dimensional verification mechanism of physical constraints, feasible instructions are screened based on dynamic load sensitivity and kinematic safety margins, effectively avoiding the issuance of over-limit instructions caused by insufficient device capabilities, and significantly improving the reliability of virtual-real interaction and the success rate of instruction execution.

[0049] 2. Through the cross-correction of multi-source execution data (such as load deviation, joint angle error), the dynamic degradation parameters reflect the evolution trend of device performance in real time, and the material scheduling priority coefficient is adaptively optimized based on historical scheduling effects to ensure the deep fit of the next-cycle instruction set with the real-time state of the production line; enabling the system to maintain efficient and stable operation even when the devices age, the working conditions fluctuate, or the production line layout is adjusted, reducing the frequency of manual intervention, and providing a highly robust closed-loop control paradigm for digital twin applications in complex industrial scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a flowchart of a real-time linkage method for a cyber-physical fusion-driven digital twin model of the present invention;

[0051] Figure 2 is a flowchart of the physical constraint matching verification of the optimized control instruction set of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0053] Embodiment: Figure 1 A real-time linkage method for a cyber-physical fusion-driven digital twin model of the present invention is given, including the following steps:

[0054] S1. Real-time collect the device status data and material flow data of the physical production line and synchronize them to the digital twin model;

[0055] S2. Extract the dynamic degradation parameters of the equipment in the physical production line based on the equipment status data, and generate the equipment capacity attenuation feature vector based on the dynamic degradation parameters;

[0056] S3. Based on the dynamic degradation parameters and the material flow data, identify the implicit process conflict paths through the constraint satisfaction problem model, and generate the material scheduling priority coefficient by combining the Monte Carlo tree search to simulate the resource preemption probability;

[0057] S4. Input the equipment capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model to generate an optimized control instruction set including the equipment dynamic capacity constraint;

[0058] S5. Perform physical constraint matching verification on the optimized control instruction set. When the verification passes, screen out the effective control instructions and send them to the physical production line;

[0059] S6. Update the dynamic degradation parameters and the material scheduling priority coefficient according to the execution results of the effective control instructions by the physical production line.

[0060] S1. Real-time collect the equipment status data and the material flow data of the physical production line and synchronize them to the digital twin model. The specific implementation is as follows:

[0061] Real-time collect the equipment status data of the physical production line, including the equipment vibration amplitude, temperature gradient and working current parameters. The equipment vibration amplitude is obtained by the acceleration sensors installed on the key moving parts of the equipment. For example, three-axis acceleration sensors are deployed at the joints of the robotic arm to collect vibration signals at a preset sampling frequency, and the maximum amplitude of the vibration signals is used as the equipment vibration amplitude. The temperature gradient is measured by the thermocouple array distributed on the surface of the equipment. Specifically, it is to calculate the temperature difference between adjacent thermocouples on the surface of the equipment. For example, thermocouples are arranged at intervals of 10 cm on the motor housing, and the real-time temperature difference between adjacent thermocouples is used as the local temperature gradient. The working current parameters are obtained in real time by the current transformers connected in series in the equipment power supply circuit. For example, Hall current sensors are installed on the power cables of the servo motors to collect the effective current value at a millisecond-level frequency and record the instantaneous peak current.

[0062] Real-time collect the material flow data of the physical production line, including the material position coordinates, the movement speed vector, and the transmission direction angle. The material position coordinates are obtained by the combined positioning of laser range sensors and RFID tags deployed on both sides of the production line conveyor belt. For example, laser range sensors are installed at the entrance and exit of the conveyor belt, combined with the unique identification of the RFID tags on the material carrier, and the real-time position coordinates of the material in the production line plane coordinate system are calculated by the triangulation method. The movement speed vector is obtained by dividing the difference between the material position coordinates of two consecutive frames by the time interval. For example, the material position coordinates are collected at a frequency of 10 frames per second, and the plane movement speed vector is obtained by dividing the coordinate change amount between adjacent frames by 0.1 second. The transmission direction angle is calculated by the tangent direction of the material movement trajectory. For example, the movement trajectory curve is fitted by taking the material position coordinates of three consecutive frames, and the angle between the tangent direction of the current frame position and the production line reference coordinate system is calculated as the transmission direction angle.

[0063] Synchronize the device status data and the material flow data to the digital twin model. The synchronization process includes the timestamp alignment of multi-source data and the conversion of the unified space coordinate system. The timestamp alignment is achieved by the time synchronization protocol deployed between the sensor nodes and the digital twin model. For example, the IEEE 1588 Precision Time Protocol (PTP) is used to synchronize the local clocks of the acceleration sensor, thermocouple, current transformer, and laser range sensor with the global clock of the digital twin model server, ensuring that the timestamp deviation of all data is less than 1 millisecond.

[0064] The conversion of the unified space coordinate system is achieved by the pre-calibrated production line reference coordinate system. Specifically, the local coordinate system data collected by each sensor is converted into the global coordinate system data with the starting point of the production line as the origin, the conveyor belt extension direction as the X-axis, and the vertical direction as the Y-axis. For example, the relative relationship between the installation position of the laser range sensor and the production line reference coordinate system is calibrated by a laser tracker, a coordinate transformation matrix is established, and the laser range data is multiplied by the transformation matrix to obtain the global coordinates.

[0065] The data formats of the device vibration amplitude, temperature gradient, working current parameter, material position coordinates, movement speed vector, and transmission direction angle are unified during the synchronization process. For example, the vibration amplitude is converted to the unit of millimeters per second squared, the temperature gradient is converted to degrees Celsius per centimeter, the current parameter is converted to the unit of amperes, the position coordinates are converted to the unit of millimeters, the speed vector is converted to the unit of millimeters per second, and the transmission direction angle is converted to the unit of radians, and then encapsulated into a structured data frame and transmitted to the digital twin model. The structured data frame contains a data category identifier, a timestamp, a numerical field, and a check code. For example, it is encapsulated in the JSON format, the data category identifiers are "vibration", "temperature_gradient", "current", etc., and the check code is the CRC-32 cyclic redundancy check code, which is used to detect data errors during the transmission process.

[0066] During the timestamp alignment process, if it is detected that the clock offset of the sensor node exceeds the preset threshold, the clock resynchronization mechanism is triggered. For example, when the clock deviation of a thermocouple node exceeds 5 milliseconds, the digital twin model server sends a clock calibration instruction to this node to realign its local clock with the global clock. During the spatial coordinate system conversion process, if it is detected that the installation position of the sensor has changed, the calibration process is re-executed. For example, when the installation angle of a laser range sensor is offset due to mechanical vibration, the calibration program is manually triggered, and the transformation matrix parameters between it and the production line reference coordinate system are re-measured using a laser tracker.

[0067] The acquisition and synchronization process of device status data and material flow data is executed in a real-time stream processing manner. For example, using Apache Kafka as the data middleware, the sensor node pushes the data to the Kafka message queue, and the digital twin model subscribes to the message queue and processes the data stream according to the time window. The time window of the real-time stream processing is set to 100 milliseconds. Every 100 milliseconds, timestamp alignment and spatial coordinate system conversion are performed on the received device vibration amplitude, temperature gradient, working current parameters, material position coordinates, motion speed vector, and transmission direction angle, and the processed data is batch-written into the real-time database of the digital twin model.

[0068] S2. Extract the dynamic degradation parameters of the devices in the physical production line based on the device status data, and generate the device capacity attenuation feature vector based on the dynamic degradation parameters. The specific implementation is as follows:

[0069] Based on the device vibration amplitude, temperature gradient, and working current parameters, extract the peak value of the device vibration time-domain feature, the temperature gradient change rate, and the current fluctuation amplitude-frequency characteristic as the dynamic degradation parameters. The peak value of the device vibration time-domain feature is obtained through the time-domain waveform analysis of the device vibration amplitude signal. Specifically, the maximum absolute value of the vibration signal is calculated within a continuous time window, and the length of the time window is set according to an integer multiple of the device operation cycle. For example, for a rotating device, the window length is set based on the single-cycle period. The temperature gradient change rate is calculated by dividing the difference in temperature gradients between adjacent time points by the time interval, and the time interval is dynamically adjusted according to the production line working conditions. For example, the interval is shortened during the high-load stage of the device to improve the monitoring frequency. The current fluctuation amplitude-frequency characteristic is extracted by performing spectrum analysis on the current signal to obtain the amplitude of the main frequency component. The spectrum analysis is implemented using the fast Fourier transform, and the main frequency component is the frequency component with an energy ratio exceeding the preset threshold in the spectrum.

[0070] Calculate the vibration degradation weight coefficient based on the difference between the peak value of the time-domain vibration characteristics of the device and the preset vibration reference value. The preset vibration reference value is calibrated by the vibration data under the initial healthy state of the device. The calibration process includes collecting the peak values of the time-domain vibration characteristics under the no-load and rated-load conditions of the device and taking the average value. The vibration degradation weight coefficient is obtained through the normalization process of dividing the difference by the reference value. The normalization process confines the range of the weight coefficient between 0 and 1, where 0 indicates no degradation and 1 indicates reaching the preset degradation critical state.

[0071] Determine the temperature degradation influence factor according to the correlation between the temperature gradient change rate and the heat dissipation efficiency of the device. The heat dissipation efficiency of the device is calculated by the product of the radiator surface area and the flow rate of the cooling medium. The radiator surface area is obtained according to the structural design parameters of the device, and the flow rate of the cooling medium is monitored in real time by a flow sensor. The temperature degradation influence factor is calculated by the ratio of the temperature gradient change rate to the heat dissipation efficiency. Before the ratio calculation, a moving average filter is applied to the temperature gradient change rate to eliminate noise interference.

[0072] Generate the current load attenuation index by combining the amplitude-frequency characteristics of the current fluctuation and the ratio of the rated load of the device. The rated load of the device is determined according to the nameplate parameters of the device or the statistical peak value of the historical operation data. For example, the 99% quantile of the effective current value in the operation data for 30 consecutive days is taken as the rated load. The current load attenuation index is obtained by dividing the amplitude-frequency characteristics of the current fluctuation by the rated load and taking the logarithmic transformation. The logarithmic transformation is used to linearize the attenuation trend.

[0073] Perform data dimensionality reduction on the vibration degradation weight coefficient, the temperature degradation influence factor, and the current load attenuation index to generate the device capacity attenuation feature vector. The data dimensionality reduction process uses the principal component analysis method, which specifically includes the following steps: standardize the vibration degradation weight coefficient, the temperature degradation influence factor, and the current load attenuation index. The standardization formula is to subtract the mean value from each parameter and then divide by the standard deviation; calculate the covariance matrix of the standardized parameters; extract the eigenvector corresponding to the largest eigenvalue of the covariance matrix as the principal component direction; project the original parameters onto the principal component direction to generate a one-dimensional feature vector. The update period of the principal component analysis direction is synchronized with the device maintenance period. For example, recalculate the covariance matrix every quarter to adapt to the change of the device degradation trend.

[0074] The mapping relationship between the device capacity attenuation feature vector and the dynamic degradation parameter is established through offline calibration. The calibration process includes collecting the dynamic degradation parameters and generating the feature vector under different healthy states of the device, and fitting the functional relationship between the feature vector and the actual load capacity through a regression model. The regression model uses the least squares method for fitting, and robust weighting is performed on the abnormal data points during the fitting process to reduce the influence of outliers.

[0075] During the data dimensionality reduction process, if it is detected that there are inconsistent dimensions or numerical overflows in the input parameters, a data standardization review mechanism is triggered. For example, when the current load decay index exceeds the effective range of logarithmic conversion, the acquisition accuracy of the current fluctuation amplitude-frequency characteristics is recalibrated. The update trigger condition of the equipment capacity decay eigenvector is bound to the production line task switching event. For example, when changing the processed workpiece or adjusting the production rhythm, the eigenvector is immediately updated to ensure model synchronization.

[0076] A fault tolerance mechanism is introduced in the calculation process of the vibration degradation weight coefficient, temperature degradation impact factor, and current load decay index. For example, when the temperature gradient change rate returns an invalid value due to a sensor failure, the historical mean value is used to calculate the temperature degradation impact factor. The status flag of the fault tolerance mechanism is linked to the digital twin model to ensure that invalid data does not trigger misjudgment.

[0077] S3. Based on the dynamic degradation parameters and material flow data, identify the implicit process conflict paths through a constraint satisfaction problem model, and generate the material scheduling priority coefficient by combining Monte Carlo tree search to simulate the resource preemption probability. The specific implementation is as follows:

[0078] Based on the dynamic degradation parameters and material flow data, construct a process conflict detection model that includes equipment health constraints and material space-time constraints. The equipment health constraints are defined by the dynamic degradation parameters, specifically including that the peak value of the equipment vibration time-domain characteristics does not exceed the preset vibration threshold, the temperature gradient change rate does not exceed the allowable range of heat dissipation efficiency, and the current load decay index is not lower than the minimum load capacity of the equipment. Among them, the preset vibration threshold is dynamically adjusted according to the equipment factory calibration value, the allowable range of heat dissipation efficiency is calculated by the product of the equipment heat dissipation surface area and the cooling medium flow rate, and the minimum load capacity of the equipment is determined by the statistical quantile of the historical operation data.

[0079] The material space-time constraints are defined by the material flow data, including that the material position coordinates are within the tolerance range of the process topology network nodes, and the motion speed vector and the transmission direction angle meet the requirements of the inter-process connection time window. Among them, the node tolerance range calibrates the physical boundary of the process nodes through a laser range finder sensor, and the connection time window is calculated by the matching relationship between the material transmission speed and the process beat time. The constraint variables of the process conflict detection model include equipment health state variables and material space-time state variables, and the constraint condition is the logical AND combination of equipment health constraints and material space-time constraints. The logical AND combination means that both types of constraints need to be satisfied simultaneously to pass the verification.

[0080] Traverse the production line process topology network through the constraint satisfaction problem model to identify the implicit process conflict paths that violate the equipment health constraints or material spatio-temporal constraints. The production line process topology network consists of process nodes and connecting edges. The nodes represent processing operations, and the edges represent the material flow direction. The attributes of the nodes and edges include process duration, equipment load rate, and material buffer capacity. The traversal process starts from the starting process node and performs a depth-first search along the connecting edges. For each path, verify the constraint satisfaction of the equipment health status variables and material spatio-temporal status variables. The verification process includes checking whether the peak value of the vibration time-domain feature exceeds the threshold, whether the temperature gradient change rate is within the allowable range of heat dissipation, whether the current load decay index is higher than the minimum load capacity, whether the material position coordinates are within the node tolerance range, and whether the velocity vector and direction angle satisfy the time window. The implicit process conflict path is defined as a path where at least one node or edge violates the equipment health constraints or material spatio-temporal constraints. For example, in a certain path, the peak value of the vibration time-domain feature of the robotic arm process node exceeds the preset vibration threshold, or the material transfer direction angle corresponding to a certain edge exceeds the process connection time window requirement, resulting in the material being unable to reach the next process node on time.

[0081] Based on Monte Carlo tree search, simulate resource preemption actions in the implicit process conflict paths and statistically calculate the resource preemption success probability for each implicit process conflict path. The resource preemption actions include equipment priority adjustment, material buffer preemption, and processing queue reordering. The simulation process is iteratively executed in the digital twin model. The simulation strategy of Monte Carlo tree search includes a selection phase, an expansion phase, a simulation phase, and a backtracking phase. In the selection phase, select the node to be explored based on the path conflict severity weight. In the expansion phase, generate possible combinations of resource preemption actions. In the simulation phase, execute the actions and record whether the resources are successfully preempted. In the backtracking phase, update the node statistical information. The resource preemption success probability is the ratio of the number of successful simulations to the total number of simulations. For example, for a certain implicit process conflict path, 1000 simulations are performed, and 800 of them successfully resolve the conflict by adjusting the equipment priority or preempting the buffer, so the success probability is 0.8. The number of simulations is adaptively adjusted according to the number of conflict paths. For example, when 5 conflict paths are detected, each is simulated 200 times; when 10 conflict paths are detected, each is simulated 100 times, and the total number is maintained within 1000 to ensure real-time performance.

[0082] Fuse the resource preemption success probability and the severity weight of the implicit process conflict path through weighted fusion to generate the material scheduling priority coefficient. The severity weight is dynamically adjusted based on the correlation strength between the equipment capacity decay feature vector and the conflict path. The correlation strength is determined by calculating the Pearson correlation coefficient between the equipment capacity decay feature vector and the health constraint parameters of the conflict path. For example, the value of the equipment capacity decay feature vector is 0.65, and the ratio of the peak value of the vibration time-domain feature of a certain conflict path to the preset vibration threshold is 1.5. Then the Pearson correlation coefficient is 0.8, and the severity weight is 0.8×1.5 = 1.2.

[0083] The material scheduling priority coefficient is equal to the resource preemption success probability multiplied by the severity weight. For example, if the resource preemption success probability is 0.8 and the severity weight is 1.2, then the material scheduling priority coefficient is 0.8×1.2 = 0.96. The dynamic adjustment process of the severity weight is synchronized with the update period of the equipment capacity attenuation eigenvector. For example, the equipment capacity attenuation eigenvector is received every 10 seconds, and the correlation strength with each conflict path is recalculated and the weight is updated. The weight allocation strategy for weighted fusion is dynamically optimized according to the real-time load status of the production line. For example, the weight of the resource preemption success probability is increased during the peak period, and the proportion of the severity weight is increased during the low period. The dynamic optimization strategy is obtained through training with historical data. The training process includes collecting the material scheduling priority coefficient and the actual scheduling effect data under different load statuses, and determining the optimal weight allocation ratio through regression analysis.

[0084] It should be noted that before calculating the Pearson correlation coefficient between the equipment capacity attenuation eigenvector and the conflict path health constraint parameter, standardization processing needs to be performed on both to eliminate the dimension difference. The standardization processing is achieved through the following steps: collect the numerical distribution of the equipment capacity attenuation eigenvector in historical data, and calculate its mean and standard deviation; collect the numerical distribution of the conflict path health constraint parameter (such as the ratio of the vibration time-domain feature peak value to the preset threshold) in historical data, and calculate its mean and standard deviation; subtract the historical mean from the current equipment capacity attenuation eigenvector value and divide by the historical standard deviation to obtain the standardized eigenvector value; subtract the historical mean from the conflict path health constraint parameter value and divide by the historical standard deviation to obtain the standardized parameter value. For example, if the historical mean of the equipment capacity attenuation eigenvector is 0.5, the standard deviation is 0.2, and the current value is 0.65, after standardization it is (0.65 - 0.5) / 0.2 = 0.75; if the mean of the ratio of the vibration time-domain feature peak value of the conflict path is 1.0, the standard deviation is 0.3, and the current value is 1.5, after standardization it is (1.5 - 1.0) / 0.3≈1.67. The standardized data is used to calculate the Pearson correlation coefficient to ensure the comparability of parameters with different dimensions.

[0085] The constraint update mechanism of the process conflict detection model is linked with the equipment maintenance records. For example, when a key component of the equipment is replaced, the preset vibration threshold is reset according to the health benchmark value of the new component, and the new benchmark value is obtained through no-load and rated load tests. The tolerance range of the material spatio-temporal constraint is dynamically corrected according to the production line layout. For example, after adding a new process node, the tolerance range of the material position coordinates is recalibrated by a laser range finder sensor, and the calibration process includes scanning the physical boundary of the node in the absence of materials and calculating the safety margin. The simulation results of Monte Carlo tree search are stored in the historical database of the digital twin model for optimizing subsequent simulation strategies. For example, if the success rate of a certain resource preemption action is less than 30% in historical simulations, its selection priority is reduced in subsequent simulations. The cleaning cycle of the historical database is synchronized with the production line scheduling cycle. For example, after the daily production ends, the data of the current day is cleared and the weekly statistical average value is retained for initializing the next day's simulation.

[0086] The severity weight anomaly detection mechanism for implicit process conflict paths is achieved by monitoring the weight change rate. For example, when the weight of a certain path increases by more than 50% within 1 minute, an artificial review process is triggered to ensure that the weight mutation is not caused by non-equipment degradation factors. The anomaly detection results are fed back to the digital twin model, and the digital twin model adjusts the correlation strength calculation parameters or excludes abnormal data points according to the review results. For example, when the review confirms that the weight mutation is caused by sensor noise, the input value of the equipment capacity attenuation eigenvector in the correlation strength calculation is reset. The timeout interruption mechanism is introduced into the simulation execution process of resource preemption actions. For example, if a single simulation exceeds the preset time threshold, it is forced to terminate and marked as failed to avoid system resource exhaustion caused by deadlocks. The timeout threshold is set to 100 milliseconds according to the real-time requirements of the production line.

[0087] The conflict priority classification mechanism for equipment health constraints and material spatio-temporal constraints is implemented through a preset rule library. For example, the priority of equipment health conflicts is higher than that of material spatio-temporal conflicts. The rule library is customized according to the equipment type and production line process requirements. For example, in a precision machining scenario, the weight of equipment health constraints is increased by 50%. The priority classification results are used to optimize the selection strategy of Monte Carlo tree search. For example, the number of simulations of high-priority conflict paths is increased by 20%. The update of the rule library is synchronized with the equipment maintenance log. For example, after each maintenance, the priority classification ratio is adjusted according to the latest equipment status.

[0088] Step S3 identifies potential process conflict paths through a constraint satisfaction problem model, combines Monte Carlo tree search to simulate resource preemption probabilities, and generates material scheduling priority coefficients. Compared with traditional scheduling methods that rely on static rules or single equipment status parameters and are difficult to dynamically adapt to the coupled conflicts between equipment degradation and material flow, this step precisely locates conflict paths through multi-dimensional constraint modeling (equipment health + material space-time), and simulates preemption actions based on Monte Carlo tree search to solve instruction conflicts caused by equipment capacity attenuation and resource competition; deeply integrates dynamic degradation parameters with material space-time data, and reflects the influence intensity of equipment health on conflict paths in real time through dynamic weights (severity weights), avoiding scheduling deviations caused by static thresholds; when equipment ages or production line loads fluctuate, the material scheduling priority coefficients can more accurately match the actual execution capabilities of physical equipment, significantly reducing the risk of instruction conflicts and enhancing the reliability and real-time performance of virtual-real synchronization optimization.

[0089] S4. Input the equipment capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model to generate an optimized control instruction set that includes equipment dynamic capacity constraints. The specific implementation is as follows:

[0090] Input the equipment capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model. The equipment capacity attenuation feature vector is generated by step S2 and represents the comprehensive degradation degree of the equipment; the material scheduling priority coefficient is generated by step S3 and represents the scheduling priority of the process conflict path. The input process is implemented through the data interface of the digital twin model. The data interface maps the feature vector and the priority coefficient to internal optimization variables of the model. For example, the equipment capacity attenuation feature vector is mapped to the equipment health status variable, and the material scheduling priority coefficient is mapped to the process node weight parameter.

[0091] Define equipment dynamic capacity constraints based on the equipment capacity attenuation feature vector, including the upper limit of the peak value of the time-domain vibration feature of the equipment, the tolerance range of the temperature gradient change rate, and the threshold of the current load attenuation index. The upper limit of the peak value of the time-domain vibration feature of the equipment is dynamically adjusted according to the current equipment capacity attenuation feature vector. For example, when the feature vector value is 0.65, the upper limit is set to 85% of the initial threshold; the tolerance range of the temperature gradient change rate is determined by the difference between the equipment heat dissipation efficiency and the current temperature gradient change rate. For example, when the heat dissipation efficiency is 1000 and the current change rate is 2 degrees Celsius per centimeter per minute, the tolerance range is ±10%; the threshold of the current load attenuation index is set according to the statistical quantile of the equipment rated load and historical operation data. For example, the 95% quantile of the load attenuation index in historical data is taken as the threshold.

[0092] Combined with the dynamic capacity constraints of the equipment and the material scheduling priority coefficient, adjustment instructions for equipment operating parameters and material path planning instructions are generated through a multi-objective optimization algorithm. The optimization objectives of the multi-objective optimization algorithm include maximizing the production line efficiency, minimizing the equipment degradation rate, and balancing the process conflict priorities. The optimization variables include the equipment processing speed, the cooling system power, and the selection of material path nodes. For example, adjust the processing speed of the robotic arm so that the peak value of its vibration time-domain characteristics does not exceed the upper limit, and at the same time allocate shorter material paths for high-priority process nodes. During the optimization process, a constraint relaxation strategy is adopted to handle conflicting objectives. For example, allow the equipment to operate beyond the limit for a short period within a certain time to ensure the beat synchronization of key processes.

[0093] Conflict resolution is performed on the adjustment instructions for equipment operating parameters and the material path planning instructions. Conflict resolution includes dynamically adjusting the instruction priority and eliminating instructions that violate the material spatio-temporal constraints. The dynamic adjustment of the instruction priority is updated in real time based on the material scheduling priority coefficient. For example, when the priority coefficient of a certain process node is 0.96, the priority of its corresponding path planning instruction is increased by 20%. Eliminating instructions that violate the material spatio-temporal constraints is achieved by verifying whether the material position coordinates are within the node tolerance range and whether the transmission direction angle meets the time window. For example, when an instruction causes the material position to exceed the tolerance range by 5 millimeters or the direction angle deviation exceeds 10 degrees, the instruction is directly eliminated.

[0094] Filter out instruction combinations that meet the equipment dynamic capacity constraints and the production line beat synchronization requirements to form an optimized control instruction set. The filtering conditions include that the peak value of the equipment vibration time-domain characteristics does not exceed the upper limit, the temperature gradient change rate is within the tolerance range, the current load attenuation index is higher than the threshold, and the material path planning conforms to the process beat time window. The generation frequency of the optimized control instruction set is synchronized with the production line beat. For example, an instruction set is generated every 5 seconds to adapt to the dynamic changes of the production line. The encoding format of the instruction set is compatible with the physical device control protocol. For example, the Modbus protocol is used to encapsulate the instruction parameters to ensure that they can be directly issued to the PLC for execution.

[0095] The update period of the equipment dynamic capacity constraints is synchronized with the generation period of the equipment capacity attenuation eigenvector. For example, receive the eigenvector every 10 seconds and recalculate the constraint threshold. The tolerance range of the material spatio-temporal constraints is dynamically corrected according to the real-time state of the production line. For example, when the conveyor belt speed fluctuation is detected, the tolerance range of the material position coordinates is automatically expanded to 8 millimeters to reduce the false rejection rate. The weight parameters of the multi-objective optimization algorithm are adaptively adjusted according to the production line load status. For example, during the peak period, the production line efficiency objective is emphasized, and during the low period, the equipment degradation objective is emphasized. The weight adjustment rule is achieved by combining offline training and online reinforcement learning.

[0096] Introduce an instruction rollback mechanism during conflict resolution. For example, when the generated instruction set fails to meet the constraints for three consecutive times, automatically switch to a conservative strategy, limit the equipment load rate to 70% of the rated value, and extend the spacing of material path nodes. The status flag of the instruction rollback mechanism is associated with the exception log of the digital twin model for subsequent maintenance and analysis. The historical data of the optimized control instruction set is stored in the digital twin database for training the weight model of the multi-objective optimization algorithm. For example, collect the instruction set and actual execution effect data in the past 24 hours and optimize the target weight allocation through regression analysis.

[0097] Step S4 generates an optimized control instruction set by integrating the equipment capacity decay feature vector and the material scheduling priority coefficient. Compared with traditional instruction generation methods that rely on static equipment parameters or single optimization objectives and are difficult to adapt to the coupled requirements of equipment dynamic degradation and real-time production line scheduling, define the instruction feasible region based on the dynamic capacity constraints of the equipment (vibration, temperature, current threshold), and balance the production line efficiency, equipment health, and process priority through a multi-objective optimization algorithm to ensure that the instruction set not only meets the current capacity limit of the equipment but also optimizes the material scheduling logic; deeply integrate the equipment health degradation parameters and the process conflict priority, and solve the problem of instruction-capability mismatch of the equipment through a dynamic constraint and conflict resolution mechanism; when the equipment ages or the production line load fluctuates, the generated instruction set accurately matches the execution ability of the physical equipment, avoiding the issuance of over-limit instructions, significantly reducing the risk of abnormal shutdown, and at the same time improving the real-time performance and reliability of virtual-real linkage.

[0098] Figure 2 The flowchart of the physical constraint matching verification of the optimized control instruction set of the present invention is given. Conduct the physical constraint matching verification on the optimized control instruction set. When the verification passes, screen the effective control instructions and send them to the physical production line. The specific implementation is as follows:

[0099] Perform physical constraint matching verification on the optimized control instruction set, including calculating the trajectory deviation rate under dynamic load sensitivity based on dynamic degradation parameters, and analyzing whether the safety margin of the kinematic chain joint angle meets the standard based on the production line layout topology. The dynamic degradation parameters are provided by the device capability attenuation feature vector generated in step S2, and the production line layout topology is defined by the process topology network in step S3. The trajectory deviation rate under dynamic load sensitivity is calculated by the difference between the real-time device load rate and the instruction expected load. The real-time device load rate is obtained by the effective value of the working current collected by the current transformer, and the instruction expected load is determined by parsing the device operation parameters in the optimized control instruction by the digital twin model. For example, if an instruction requires the robotic arm to operate at 90% of the rated load, and the real-time load rate is only 75% of the rated value due to device aging, then the trajectory deviation rate is (90% - 75%) = 15%. The trajectory deviation rate threshold is dynamically adjusted according to the device capability attenuation feature vector. The adjustment rule is: when the feature vector value is less than 0.5, the threshold decreases by 3% for every 0.1 eigenvalue; when the feature vector value is greater than or equal to 0.5, it decreases by 6% for every 0.1 eigenvalue. For example, when the feature vector value is 0.4, the threshold is 22%, and when the feature vector value is 0.6, the threshold is 14%.

[0100] The safety margin of the kinematic chain joint angle is calculated by the difference between the current joint angle and the mechanical structure limit angle. The current joint angle is collected in real time by the encoder, and the mechanical structure limit angle is set according to the device design parameters. The condition for the safety margin to meet the standard is that the difference is greater than the preset safety threshold. The preset safety threshold is determined by the product of the minimum distance between adjacent process nodes in the production line layout topology and the device mechanical error. For example, if the minimum distance between adjacent nodes is 500 millimeters and the device mechanical error is 0.5%, then the safety threshold is 500×0.5% = 2.5 millimeters, which is converted to a joint angle tolerance of 2 degrees. When calculating the difference, if the current joint angle is 85 degrees and the mechanical structure limit angle is 90 degrees, then the safety margin is 5 degrees. When the preset safety threshold is 2 degrees, it is determined to meet the standard.

[0101] When the trajectory deviation rate is lower than the trajectory deviation rate threshold and the safety margin meets the standard, it is determined that the physical constraint matching verification passes, and the effective control instructions are screened and sent to the physical production line. During the screening process, the instructions with an over-limit trajectory deviation rate or a safety margin not meeting the standard are eliminated. For example, when the trajectory deviation rate of an instruction is 18% and exceeds the threshold of 15%, or the safety margin of a certain joint is only 1 degree, which is lower than the threshold of 2 degrees, it is directly eliminated. The retained instructions need to simultaneously meet the device dynamic capability constraints, including that the peak value of the device vibration time-domain characteristic does not exceed the upper limit, the temperature gradient change rate is within the tolerance range, and the current load attenuation index is higher than the threshold. The digital twin model sorts the screened effective control instructions by priority and encapsulates them into an instruction frame that can be parsed by the physical device. For example, the instruction parameters are encoded using the Modbus protocol and sent to the PLC controller for execution through the industrial Ethernet.

[0102] The dynamic adjustment process of the trajectory deviation rate threshold is synchronized with the update period of the equipment capacity attenuation eigenvector. For example, the eigenvector is received every 10 seconds and the threshold is recalculated. The preset safety threshold of the safety margin is dynamically corrected according to the real-time working conditions of the production line. For example, when the production line switches to the high-precision processing mode, the safety thresholds of all process nodes are uniformly reduced by 30%. A command priority backtracking mechanism is introduced during the screening process. For example, when a high-priority command is excluded due to the safety margin not meeting the standard, the digital twin model is automatically triggered to regenerate a replacement command to ensure the continuity of critical processes.

[0103] The intermediate results of the physical constraint matching verification are stored in the digital twin database for optimizing subsequent verification strategies. For example, if it is statistically found that the trajectory deviation rate of a certain type of command is long-term lower than the threshold by 50%, its verification frequency is gradually reduced to once every 30 seconds to reduce the computing load. If abnormal sensor data is detected during the verification process, such as the reading mutation of the current transformer exceeding the historical fluctuation range, a data review process is triggered, and the command issuance is suspended until the abnormality is resolved.

[0104] A redundant verification mechanism is introduced in the calculation of the safety margin of the kinematic chain joint angles. For example, the joint angles and the position coordinates of the end effector are simultaneously collected, and the consistency of the angle data is verified through inverse kinematics. If the deviation of the conversion result between the angle and the coordinate exceeds the tolerance, the angle data is determined to be invalid and re-collected. The redundant verification results are fed back to the digital twin model, and the digital twin model adjusts the filtering parameters of the joint angle collection or triggers the sensor calibration program according to the deviation value.

[0105] Among them, the joint angle tolerance is converted into a linear distance according to the arm length of the equipment kinematic chain. For example, when the arm length is 1 meter and the angle tolerance is 2 degrees, the corresponding end linear tolerance is 1×sin(2°)≈34.9 mm, ensuring clear physical meaning for the spatial constraint verification.

[0106] Step S5 screens valid control commands through physical constraint matching verification. Compared with the traditional command verification that relies on fixed thresholds or single equipment parameters and is difficult to dynamically adapt to the execution ability fluctuations caused by equipment degradation and production line layout changes, this step dynamically adjusts the trajectory deviation rate threshold based on the equipment capacity attenuation eigenvector and sets the safety margin conditions according to the production line layout topology, realizing the collaborative verification of the equipment health state and spatial constraints, deeply integrating the dynamic degradation parameters with the production line space topology, and solving the accurate matching problem between commands and equipment dynamic capabilities through two-dimensional verification (load trajectory + kinematic safety). When the equipment performance decays or the production line layout is adjusted, the verification conditions are real-time adapted to the real capabilities of physical equipment, avoiding control failures caused by issuing over-limit commands, and at the same time improving the real-time and reliability of virtual-real linkage, ensuring that the optimized command set not only meets the current capacity limitations of the equipment but also satisfies the production line space safety requirements.

[0107] S6. Update the dynamic degradation parameters and the material scheduling priority coefficient according to the execution result of the effective control instruction on the physical production line. The specific implementation is as follows:

[0108] Obtain the execution result data of the effective control instruction on the physical production line, including the actual equipment load rate, the joint angle deviation value, and the process beat synchronization error. The actual equipment load rate is obtained from the effective value of the real-time working current collected by the current transformer. For example, if the expected load of a certain instruction is 80% of the rated value and the actual load rate is 75%, the load rate deviation in the execution result data is 5%. The joint angle deviation value is calculated from the difference between the actual joint angle recorded by the encoder and the expected angle of the instruction. For example, if the expected angle of the instruction is 90 degrees and the actual angle is 88 degrees, the deviation value is 2 degrees. The process beat synchronization error is calculated from the difference between the material arrival timestamp and the planned time of the process node. For example, if a certain process is planned to start at 10:00:00 and the actual material arrival time is 10:00:03, the synchronization error is 3 seconds.

[0109] Update the dynamic degradation parameters based on the execution result data, including correcting the peak value of the vibration time domain feature by the actual equipment load rate, correcting the temperature gradient change rate by the joint angle deviation value, and correcting the current load decay index by the process beat synchronization error. The correction method for the peak value of the vibration time domain feature is: when the actual load rate is lower than the expected load of the instruction, reduce the vibration peak threshold according to the load rate deviation ratio. For example, a 5% deviation corresponds to a 3% reduction in the threshold. The correction method for the temperature gradient change rate is: adjust the heat dissipation efficiency calculation weight according to the joint angle deviation value. For example, when the deviation is 2 degrees, the weight is reduced by 0.1, expanding the tolerance range of the temperature gradient change rate. The correction method for the current load decay index is: convert the process beat synchronization error into a load decay compensation coefficient. For example, an error of 3 seconds corresponds to a compensation coefficient of 0.95, and the current load decay index is adjusted to the original value multiplied by 0.95.

[0110] Regenerate the equipment capacity attenuation feature vector based on the updated dynamic degradation parameters. The regeneration process uses the same data dimensionality reduction method as in step S2. For example, input the corrected peak value of the vibration time domain feature, the temperature gradient change rate, and the current load decay index into the principal component analysis model to calculate the new equipment capacity attenuation feature vector. The covariance matrix of the principal component analysis model is recalculated based on the updated dynamic degradation parameters to ensure that the feature vector reflects the latest health state of the equipment. For example, when the corrected vibration peak is 1.0 mm / s², the temperature gradient change rate is 1.5 °C / cm / min, and the current load decay index is 0.8, the generated feature vector value is 0.7, indicating that the comprehensive degradation degree of the equipment is 70%.

[0111] Combining the process cycle synchronization error with the historical execution effect, update the material scheduling priority coefficient through a regression model. The regression model uses the historical process cycle synchronization error as the input variable and the priority coefficient adjustment amount as the output variable. The model coefficients are obtained by fitting historical data using the least squares method. For example, when the average historical synchronization error of a certain process node is 2 seconds, the regression model outputs an increase of 0.1 in the priority coefficient of this node; if the error is 5 seconds, the coefficient is decreased by 0.2. The model training data includes the execution errors and priority coefficient adjustment records of all process nodes within the past 24 hours, and is refitted every 6 hours to adapt to the dynamic changes of the production line.

[0112] Synchronize the updated equipment capacity attenuation feature vector and the material scheduling priority coefficient to the digital twin model to generate the optimized control instruction set for the next cycle. The synchronization process is achieved through the data interface of the digital twin model. For example, map the feature vector to the equipment health status variable and the priority coefficient to the process node weight parameter. Based on the synchronized parameters, the digital twin model immediately starts generating instructions for the next cycle. For example, update the instruction set every 10 seconds to ensure real-time matching with the production line cycle. The encoding format of the updated instruction set is compatible with the physical device protocol. For example, encapsulate the instruction parameters using the OPC UA protocol and transmit them to the PLC controller for execution via the industrial Ethernet.

[0113] The update cycle of the dynamic degradation parameters is bound to the production line task switching event. For example, when the production line switches the processed workpiece type, immediately trigger the parameter update to ensure model synchronization. The regression model of the material scheduling priority coefficient starts an emergency update when an abnormal error is detected. For example, if the synchronization error of a certain process node exceeds 10 seconds continuously for 3 times, force the model to be refitted and adjust the coefficient. During the update process, if abnormal sensor data is detected, such as the reading of the current transformer mutating beyond the historical fluctuation range, suspend the parameter update and trigger the data review process, and continue to execute after the abnormality is eliminated.

[0114] The redundant check mechanism is introduced in the process of regenerating the equipment capacity attenuation feature vector. For example, simultaneously use two dimensionality reduction methods, principal component analysis and linear discriminant analysis. If the results differ by more than 5%, start the manual review. The redundant check log is stored in the digital twin database for subsequent model optimization. The adjustment record of the material scheduling priority coefficient is associated with the production line maintenance log. For example, when the priority coefficient of a certain node is continuously decreased, automatically generate equipment maintenance suggestions and push them to the operation and maintenance terminal.

[0115] Step S6 maps the timing error data into a non - linear adjustment amount of the priority coefficient through a regression model based on the process - beat synchronization error and historical execution effects, breaking through the traditional static weight - allocation rules. At the same time, the update of the dynamic degradation parameters is not an independent correction. Instead, through the cross - influence of the joint - angle deviation value and the load - rate deviation, a correlation correction model for vibration, temperature, and current parameters is established. For example, the joint - angle deviation is not only used to adjust the temperature - gradient change rate but also affects the compensation coefficient of the current - load decay index through spatial tolerance conversion, forming a multi - physical - field coupling iteration mechanism for the equipment health state. The regeneration of the equipment - capacity decay eigenvector is not a simple repetition of the initial dimensionality - reduction process. Instead, a redundancy - check mechanism is introduced to compare the consistency of the results of principal - component analysis and linear - discriminant analysis, ensuring that the eigenvector after parameter update can not only reflect the real - time degradation degree of the equipment but also be compatible with the safety - margin requirements of the production - line topological layout. This collaborative design of multi - algorithm cross - verification and dynamic - parameter feedback solves the problem of feature distortion caused by data noise or model drift in traditional methods. In addition, the real - time binding mechanism of the priority - coefficient update and the production - line task - switching event enables the model to adaptively adjust parameters without manual intervention when the processing scenario changes. For example, in the high - precision processing mode, the safety threshold is automatically reduced and the priority weight of key process nodes is increased, while traditional methods usually require presetting multiple working - condition modes and manually switching parameters.

[0116] The solution involved in this embodiment constructs a real - time linkage mechanism with deep virtual - physical integration through a complete technical chain of multi - source data acquisition, dynamic degradation modeling, conflict - path identification, instruction - generation verification, and closed - loop feedback. Traditional digital - twin control methods usually rely on one - way instruction issuance or static model parameters, making it difficult to adapt to the coupled changes of equipment dynamic degradation and production - line real - time scheduling and lacking the ability of cross - domain parameter collaborative optimization. Based on the deep integration of multi - physical - field data such as equipment vibration, temperature, and current, this embodiment extracts the equipment - capacity decay eigenvector to replace the traditional single health index. Through the collaboration of the constraint - satisfaction problem model and Monte Carlo tree search, the equipment health state and material spatio - temporal constraints are embedded in the process conflict detection to realize the dynamic identification of hidden paths and the accurate simulation of resource - preemption probability.

[0117] Furthermore, the physical - constraint matching verification is not a simple threshold judgment. Instead, it is a two - dimensional verification combining dynamic - load sensitivity and kinematic - chain safety margin. Among them, the trajectory - deviation - rate threshold is dynamically adjusted according to the equipment health state, and the safety - margin condition is deeply related to the production - line layout topology, breaking through the limitations of traditional fixed thresholds or empirical rules. The instruction - generation and update mechanism, through the collaboration of multi - objective optimization algorithms and regression models, incorporates the equipment dynamic - capacity constraints, process priorities, and historical execution effects into a unified optimization framework to realize the adaptive iteration of the instruction set.

[0118] Particularly crucial is that the update of dynamic degradation parameters and priority coefficients in the closed-loop feedback is not an independent operation. Instead, through the cross-influence of multi-source data such as joint angle deviation and beat synchronization error, an associated correction model for vibration-temperature-current parameters is established, and the priority weight allocation strategy is optimized based on the historical data of process conflicts, realizing the dynamic calibration and adaptive optimization of the virtual model and the execution capabilities of physical devices. For example, when the joint angle deviation exceeds the limit, the vibration threshold is synchronously reduced and the weight of heat dissipation efficiency is increased to avoid system fragmentation caused by independent correction of a single parameter. At the same time, the priority weight allocation strategy constructed based on the historical data of process conflicts dynamically adjusts the material scheduling rules and the equipment load distribution ratio by analyzing the spatio-temporal characteristics of high-frequency conflict paths, enabling the optimization instruction set to adapt to complex working conditions such as production line processing beat fluctuations and equipment performance degradation. It effectively solves the defect that the model parameters in traditional virtual-real linkage lag behind the physical state changes, and through data closed-loop driven parameter iteration, ensures that the digital twin model always accurately maps the true ability boundary of the equipment, thereby avoiding execution risks at the control instruction generation stage and realizing the upgrade from the one-way instruction issuing ecosystem to the two-way dynamic collaboration mode.

[0119] The calculations involved in the embodiments are all numerical calculations after removing the dimensions, and the preset parameters and threshold selections in the calculations are set by those skilled in the art according to the actual situation.

[0120] It should be noted that the present invention can be deployed on the device itself to achieve embedded applications, or can also run on a PC or other terminals with a user interface, so as to meet various hardware environments and usage requirements.

[0121] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0122] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and modules described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0123] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or modules can be in electrical, mechanical, or other forms.

[0124] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules. They can be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0125] In addition, in each embodiment of the present application, each functional module may be integrated into one processing module, may exist physically alone for each module, or two or more modules may be integrated into one module.

[0126] If the above-mentioned function is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0127] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

[0128] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A real-time linkage method for a cyber-physical fusion-driven digital twin model, characterized in that It includes the following steps: S1. Real-time collect the equipment status data and material flow data of the physical production line and synchronize them to the digital twin model; S2. Extract the dynamic degradation parameters of the equipment in the physical production line based on the equipment status data, and generate an equipment capacity attenuation feature vector based on the dynamic degradation parameters; S3. Based on the dynamic degradation parameters and the material flow data, identify the implicit process conflict paths through the constraint satisfaction problem model, and generate the material scheduling priority coefficient by combining the Monte Carlo tree search to simulate the resource preemption probability; S4. Input the equipment capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model to generate an optimized control instruction set including equipment dynamic capacity constraints; S5. Perform physical constraint matching verification on the optimized control instruction set, and when the verification passes, screen out the effective control instructions and send them to the physical production line; S6. Update the dynamic degradation parameters and the material scheduling priority coefficient according to the execution results of the effective control instructions by the physical production line.

2. The real-time linkage method of a cyber-physical fusion-driven digital twin model according to claim 1, characterized in that, Real-time collect the equipment status data and material flow data of the physical production line and synchronize them to the digital twin model, including: Real-time collect the equipment status data of the physical production line, including the equipment vibration amplitude, temperature gradient, and working current parameters; Real-time collect the material flow data of the physical production line, including the material position coordinates, movement speed vector, and transmission direction angle; Synchronize the equipment status data and the material flow data to the digital twin model, where the synchronization process includes timestamp alignment of multi-source data and unified spatial coordinate system conversion.

3. A real - time linkage method for a cyber - physical fusion - driven digital twin model according to claim 1, characterized in that, Extract the dynamic degradation parameters of the equipment in the physical production line based on the equipment status data, and generate an equipment capacity attenuation feature vector based on the dynamic degradation parameters, including: Based on the equipment vibration amplitude, temperature gradient, and working current parameters, extract the equipment vibration time-domain feature peak value, temperature gradient change rate, and current fluctuation amplitude-frequency characteristics as the dynamic degradation parameters; Calculate the vibration degradation weight coefficient according to the difference between the equipment vibration time-domain feature peak value and the preset vibration reference value; Determine the temperature degradation influence factor according to the correlation between the temperature gradient change rate and the equipment heat dissipation efficiency; Generate the current load attenuation index by combining the current fluctuation amplitude-frequency characteristics and the ratio of the equipment rated load; Perform data dimensionality reduction processing on the vibration degradation weight coefficient, temperature degradation influence factor, and current load attenuation index to generate an equipment capacity attenuation feature vector.

4. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 1, characterized in that Based on the dynamic degradation parameters and the material flow data, identify the implicit process conflict paths through the constraint satisfaction problem model, and generate the material scheduling priority coefficient by combining the Monte Carlo tree search to simulate the resource preemption probability, including: Based on the dynamic degradation parameters and the material flow data, construct a process conflict detection model including equipment health constraints and material space-time constraints; Traverse the production line process topology network through the constraint satisfaction problem model to identify the implicit process conflict paths that violate the equipment health constraints or the material space-time constraints; Based on the Monte Carlo tree search, simulate the resource preemption actions in the implicit process conflict paths, and count the resource preemption success probabilities of each implicit process conflict path; Perform weighted fusion on the resource preemption success probability and the severity weight of the implicit process conflict path to generate the material scheduling priority coefficient.

5. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 4, characterized in that, Among them, the severity weight is dynamically adjusted based on the correlation strength between the device capacity attenuation feature vector and the conflict path.

6. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 1, characterized in that, Input the device capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model to generate an optimized control instruction set including device dynamic capacity constraints, including: Input the device capacity attenuation feature vector and the material scheduling priority coefficient into the digital twin model; Define device dynamic capacity constraints based on the device capacity attenuation feature vector, including the upper limit of the peak value of the device vibration time-domain feature, the tolerance range of the temperature gradient change rate, and the threshold of the current load attenuation index; Combine the device dynamic capacity constraints and the material scheduling priority coefficient, and generate device operation parameter adjustment instructions and material path planning instructions through a multi-objective optimization algorithm; Resolve conflicts for the device operation parameter adjustment instructions and the material path planning instructions. Conflict resolution includes dynamically adjusting the instruction priority and eliminating instructions that violate the material space-time constraints; Screen the instruction combinations that meet the device dynamic capacity constraints and the production line beat synchronization requirements to form an optimized control instruction set.

7. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 1, characterized in that Perform physical constraint matching verification on the optimized control instruction set. When the verification passes, screen the effective control instructions and send them to the physical production line, including: Perform physical constraint matching verification on the optimized control instruction set, including calculating the trajectory deviation rate under the dynamic load sensitivity based on the dynamic degradation parameters, and analyzing whether the safety margin of the kinematic chain joint angle meets the standard based on the production line layout topology; When the trajectory deviation rate is lower than the trajectory deviation rate threshold and the safety margin meets the standard, it is determined that the physical constraint matching verification passes. Screen the effective control instructions and send them to the physical production line; During the screening process, eliminate the instructions with an excessive trajectory deviation rate or a safety margin that does not meet the standard, and retain the instructions that meet the device dynamic capacity constraints.

8. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 7, characterized in that The trajectory deviation rate under the dynamic load sensitivity is calculated by the difference between the real-time device load rate and the expected load of the instruction. The trajectory deviation rate threshold is dynamically adjusted according to the device capacity attenuation feature vector.

9. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 7, characterized in that The safety margin of the kinematic chain joint angle is calculated by the difference between the current joint angle and the mechanical structure limit angle. The condition for the safety margin to meet the standard is that the difference is greater than the preset safety threshold.

10. A real-time linkage method for a cyber-physical fusion-driven digital twin model according to claim 1, characterized in that, According to the execution results of the effective control instructions by the physical production line, update the dynamic degradation parameters and the material scheduling priority coefficient, including: Obtain the execution result data of the effective control instructions by the physical production line, including the actual device load rate, the joint angle deviation value, and the process beat synchronization error; Update the dynamic degradation parameters based on the execution result data, including correcting the peak value of the vibration time-domain feature through the actual device load rate, correcting the temperature gradient change rate through the joint angle deviation value, and correcting the current load attenuation index through the process beat synchronization error; Regenerate the device capacity attenuation feature vector based on the updated dynamic degradation parameters; Update the material scheduling priority coefficient through a regression model by combining the process beat synchronization error and the historical execution effect; Synchronize the updated device capacity attenuation feature vector and the material scheduling priority coefficient to the digital twin model to generate the optimized control instruction set for the next cycle.

Images