Method and system for adjusting and optimizing digital twin model for automobile production line monitoring

Abstract

The invention belongs to the technical field of digital twinning, and provides an adjustment and optimization method and system of a digital twinning model for automobile production line monitoring. The method comprises the following steps: collecting real-time operation data of each entity device of an automobile production line, and extracting working condition fluctuation characteristics from the real-time operation data; inputting the working condition fluctuation characteristics into a digital twinborn model, and calculating a comprehensive deviation value between a simulation result of the digital twinborn model and real-time operation data through a pre-constructed deviation evaluation model; when the comprehensive deviation value exceeds a preset threshold value, an equipment characteristic parameter library is called, and a parameter adjustment strategy for the corresponding entity equipment is generated based on a matching result of the working condition fluctuation characteristics and the equipment characteristic parameter library; and dynamically updating the virtual parameters of the digital twin model according to the parameter adjustment strategy within the duration corresponding to the update window length. According to the method, the adaptive capacity of the digital twin model to the working condition fluctuation can be enhanced.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and more specifically, to a method and system for adjusting and optimizing a digital twin model used for monitoring automotive production lines. Background Technology

[0002] The entire process chain of automobile production, including stamping, welding, painting, and final assembly, involves complex dynamic coupling of thousands of pieces of equipment and tens of thousands of processes. Digital twins, by constructing virtual models that completely correspond to the physical production line, can realize functions such as real-time perception of equipment status, full traceability of process parameters, and early warning of quality defects, providing intelligent solutions for precise control, intelligent detection, and efficient sorting of automated production lines.

[0003] However, in practical applications, a significant contradiction exists between the simulation realism of digital twin models and the dynamic changes of physical entities. The core issue lies in the imbalance between the virtual and physical mapping relationship caused by fluctuations in operating conditions. Fluctuations in the operating conditions of automotive production lines are characterized by normalization and complexity, primarily manifested as gradual performance degradation due to equipment aging, abrupt attribute changes caused by batch variations in materials, and dynamic drift of process parameters due to environmental and load influences. These fluctuations directly cause the actual operating state of the physical entity to continuously deviate from the initial design baseline, but digital twin models struggle to achieve synchronous adaptation, specifically in at least the following aspects: (1) The fixed nature of model parameters limits real-time matching capabilities. The simulation logic of a digital twin model relies on preset physical parameters, equipment characteristic parameters, and process baseline parameters, which are statically set based on historical data or design standards. When operating conditions fluctuate, the actual state of the physical entity changes, but if the model does not update the corresponding parameters in time, it will still perform simulation calculations based on the initial parameters. For example, when a welding robot's trajectory deviates due to wear on the robotic arm, if the model does not adjust the motion accuracy parameters synchronously, its simulated trajectory will still use the initial standard value, directly causing the accumulation of deviations between the virtual simulation results and the actual physical state.

[0004] (2) The dynamic complexity of the operating condition fluctuations exceeds the model's preset range. The multi-dimensional coupling of equipment aging, material changes and parameter drift forms a nonlinear and complex fluctuation scenario. The initial parameter system of the digital twin model is difficult to cover the dynamic correlation of all fluctuation combinations. For example, when a stamping machine tool encounters both die wear and abnormal material hardness at the same time, if the model only presets the simulation logic of a single factor, it cannot reflect the real stress state under dual fluctuations, resulting in simulation distortion.

[0005] (3) The synchronization between the model and the entity lacks an automated maintenance mechanism. The accurate mapping of digital twins is not naturally achieved; it requires continuous data interaction and parameter calibration. Existing digital twin models generally lack adaptive adjustment capabilities. When the physical entity deviates from the standard baseline due to fluctuations in operating conditions, the model still operates based on the initial baseline, leading to a continuous increase in the deviation between the virtual simulation results and the actual production data. For example, if the drying temperature of the coating line drifts due to heater aging, and the model does not update the temperature conduction parameters, the simulated paint film curing effect will deviate significantly from the actual quality, resulting in a loss of monitoring effectiveness.

[0006] Therefore, how to solve the problem of dynamic parameter adaptation of digital twin models when operating conditions fluctuate, eliminate the accumulation of deviations between virtual simulation and physical reality, and realize the autonomous optimization and adjustment of the model are the technical problems that digital twin technology used for monitoring automotive production lines urgently needs to solve. Summary of the Invention

[0007] In response, the present invention provides a method, system, electronic device, computer storage medium, and computer program product for adjusting and optimizing a digital twin model for monitoring an automotive production line, in order to solve at least one of the aforementioned technical problems.

[0008] In a first aspect, the present invention provides a method for adjusting and optimizing a digital twin model for monitoring an automotive production line, applied to an edge computing node, comprising the following steps: Real-time operating data of various physical equipment in the automobile production line is collected, and operating condition fluctuation characteristics are extracted from the real-time operating data, including performance degradation characteristics caused by equipment aging, attribute differences caused by material batch changes, and drift characteristics of process parameters affected by environment and load. The operating condition fluctuation characteristics are input into the digital twin model, and the comprehensive deviation value between the simulation results of the digital twin model and the real-time operating data is calculated through a pre-built deviation evaluation model. When the overall deviation value exceeds the preset threshold, the equipment characteristic parameter library is invoked. Based on the matching result between the operating condition fluctuation characteristics and the equipment characteristic parameter library, a parameter adjustment strategy for the corresponding physical equipment is generated. The equipment characteristic parameter library contains operating principle parameters, loss law parameters and sensitive parameter ranges for different types of physical equipment. The parameter adjustment strategy includes parameter adjustment items, adjustment range and adjustment priority. Within the time period corresponding to the update window length, the virtual parameters of the digital twin model are dynamically updated according to the parameter adjustment strategy to complete the mapping calibration between the digital twin model and the corresponding physical device, and the adjustment and optimization record is fed back to the edge computing node to optimize the extraction strategy of the operating condition fluctuation characteristics; wherein, the update window length is determined based on the parameter adjustment strategy.

[0009] A second aspect of the present invention provides an adjustment and optimization system for a digital twin model used for monitoring an automotive production line, applied to an edge computing node, the system comprising: The working condition feature extraction unit collects real-time operating data of various physical equipment in the automobile production line and extracts working condition fluctuation features from the real-time operating data, including performance degradation features caused by equipment aging, attribute differences caused by material batch changes, and drift features of process parameters affected by environment and load. The comprehensive deviation analysis unit inputs the operating condition fluctuation characteristics into the digital twin model and calculates the comprehensive deviation value between the simulation results of the digital twin model and the real-time operating data through a pre-built deviation evaluation model. The parameter adjustment strategy generation unit calls the equipment characteristic parameter library when the comprehensive deviation value exceeds a preset threshold. Based on the matching result between the operating condition fluctuation characteristics and the equipment characteristic parameter library, it generates a parameter adjustment strategy for the corresponding physical equipment. The equipment characteristic parameter library contains operating principle parameters, loss law parameters and sensitive parameter ranges for different types of physical equipment. The parameter adjustment strategy includes parameter adjustment items, adjustment range and adjustment priority. The virtual parameter update unit dynamically updates the virtual parameters of the digital twin model according to the parameter adjustment strategy within the time period corresponding to the update window length, completes the mapping calibration between the digital twin model and the corresponding physical device, and feeds back the adjustment and optimization record to the edge computing node to optimize the extraction strategy of the operating condition fluctuation characteristics; wherein, the update window length is determined based on the parameter adjustment strategy.

[0010] A third aspect of the present invention provides an electronic device for use in an edge computing node, the electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, the computer program, when executed by the processor, implementing the method as described in any of the preceding claims.

[0011] A fourth aspect of the present invention provides a computer storage medium for use in an edge computing node, the computer storage medium storing a computer program that can be executed by a processor to implement the methods described in any of the preceding claims.

[0012] A fifth aspect of the invention provides a computer program product applied to an edge computing node, the computer program product comprising a computer program executable by a processor to implement the methods described in any of the preceding claims.

[0013] The beneficial technical effects of this invention are as follows: (1) By adopting a data acquisition architecture that combines multiple types of sensors with edge computing, the characteristics of working condition fluctuations such as equipment aging, material changes, and parameter drift are accurately extracted, which solves the problem of insufficient perception of complex working conditions by traditional digital twin models and provides reliable input for mapping calibration. (2) By matching multi-dimensional deviation assessment with equipment characteristic parameter library, a differentiated parameter adjustment strategy is generated to realize the hierarchical dynamic update of virtual parameters, effectively eliminate the accumulation of mapping deviation, and improve the dynamic mapping accuracy of the model to physical entities; (3) By relying on structured log feedback and incremental learning to optimize feature extraction strategies, a closed-loop optimization mechanism is formed, which can continuously enhance the adaptability of digital twin models to working condition fluctuations. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating a method for adjusting and optimizing a digital twin model for monitoring an automobile production line, as disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of the deviation evaluation model disclosed in the embodiments of the present invention; Figure 3 This is a schematic diagram of the structure of a digital twin model adjustment and optimization system for monitoring an automobile production line, as disclosed in an embodiment of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

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

[0018] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the embodiments of the present invention.

[0019] like Figure 1As shown in the figure, this invention discloses an adjustment and optimization method for a digital twin model used for monitoring an automotive production line, applied to edge computing nodes, including the following method steps: 100. Collect real-time operating data of each physical equipment in the automobile production line, and extract the operating condition fluctuation characteristics from the real-time operating data, including the performance degradation characteristics caused by equipment aging, the attribute difference characteristics caused by material batch changes, and the drift characteristics of process parameters affected by the environment and load.

[0020] This invention employs a distributed sensing and edge processing architecture. Specifically, it deploys multiple types of sensors at key equipment nodes such as the hydraulic system of a stamping machine, the mechanical joints of a welding robot, and the drying oven of a coating line. These sensors include: vibration sensors (sampling frequency up to 1kHz) to capture minute vibration signals from equipment operation; temperature sensors (accuracy ±0.5℃) to monitor equipment and ambient temperature in real time; pressure sensors to record pressure changes in the hydraulic system and pneumatic devices; image data such as material dimensions and assembly posture acquired through machine vision equipment; and process parameter sensors to directly acquire key process values ​​such as welding current and coating viscosity. The collected raw data is transmitted to edge computing nodes via industrial Ethernet.

[0021] First, data cleaning is performed: outliers caused by sensor malfunctions are removed (using the 3σ principle), missing values ​​in data transmission gaps are filled (using linear interpolation), and different types of data are normalized (physical quantities such as vibration and temperature are converted into 0-1 intervals). Then, feature mining is performed using feature extraction algorithms. For example, for equipment aging, time-domain analysis is used to extract attenuation features such as the root mean square value and peak factor of vibration signals, and trend analysis is used to obtain performance degradation curves; for batch changes in materials, image recognition is used to extract differences in material surface texture, dimensional deviations, and other attributes, and batch feature vectors are established by combining material composition detection data; for process parameter drift, frequency domain analysis (Fourier transform) is used to extract the frequency features of parameter fluctuations and identify the slope and acceleration of the drift trend.

[0022] 200. Input the operating condition fluctuation characteristics into the digital twin model, and calculate the comprehensive deviation value between the simulation results of the digital twin model and the real-time operating data through the pre-constructed deviation evaluation model.

[0023] After receiving the operational fluctuation characteristics output in step 100, the digital twin model starts the real-time simulation engine. That is, it calculates the kinematic parameters of the equipment (such as robot joint angles and conveyor line speed) based on the physics engine, calculates the quality prediction values ​​(such as welding strength and paint film thickness) based on the process simulation model, and generates a virtual operation result dataset that corresponds one-to-one with the physical entity.

[0024] Simultaneously, a pre-built deviation assessment model is invoked to calculate the deviation between the simulation results of the digital twin model and the real-time operating data. This includes: in the parameter dimension, calculating the deviation between the virtual and actual values ​​of individual parameters such as vibration, temperature, and pressure (absolute deviation = |virtual value - actual value|, relative deviation = |virtual value - actual value| / actual value); in the time dimension, analyzing the trend of deviation over time (such as the rate of deviation accumulation and duration); and in the spatial dimension, calculating shape deviation using Hausdorff distance for spatial parameters such as welding trajectory and coating coverage. Subsequently, the Analytic Hierarchy Process (AHP) is used to assign weights to the deviations in different dimensions (after normalization) (e.g., the weight of process parameter deviations is higher than that of environmental parameters), and the results are then fused to obtain a comprehensive deviation value.

[0025] 300. When the comprehensive deviation value exceeds the preset threshold, the equipment characteristic parameter library is invoked. Based on the matching result between the operating condition fluctuation characteristics and the equipment characteristic parameter library, a parameter adjustment strategy for the corresponding physical equipment is generated. The equipment characteristic parameter library contains operating principle parameters, loss law parameters and sensitive parameter ranges for different types of physical equipment. The parameter adjustment strategy includes parameter adjustment items, adjustment range and adjustment priority.

[0026] When the overall deviation value exceeds the upper limit of the preset threshold range, the parameter adjustment process is triggered. The preset threshold adopts a dynamic threshold library design, that is, threshold ranges are set for different equipment types (such as stamping machines and welding robots) and different operating conditions (such as equipment break-in period, stable operation period, and aging period). The initial threshold value is determined based on the equipment's factory standard and historical stable operation data statistics, and is subsequently dynamically optimized through the feedback mechanism in step 400.

[0027] The equipment characteristic parameter library comprises a body attribute layer and a rule engine layer. Specifically, the body attribute layer stores basic equipment parameters (such as rated power and operating accuracy level), loss pattern parameters (such as the aging coefficient change curve over operating time), and sensitive parameter ranges (such as the optimal range of welding current and the safe threshold of stamping pressure), and is categorized and indexed by equipment type (stamping, welding, painting, assembly) and functional module (drive system, control system, execution system). The rule engine layer contains preset adjustment logic, such as "when the vibration deviation exceeds the threshold and bearing aging characteristics are extracted, the equipment damping coefficient parameter is adjusted first," and "when the material size deviation triggers a quality warning, the conveyor line positioning parameters are adjusted in conjunction."

[0028] The extracted operating condition fluctuation features (such as aging feature vectors and drift feature vectors) are compared with typical feature templates pre-stored in the equipment characteristic parameter library. The cosine similarity or Euclidean distance is calculated, and the equipment type and parameter adjustment rules corresponding to the template with the highest matching degree are activated. For example, when the feature combination of "welding current fluctuation + abnormal arc intensity" is detected, the typical template of "electrode wear" of the welding robot is matched, and the corresponding current compensation parameter adjustment rules are called.

[0029] The parameter adjustment strategy includes parameter adjustment items, adjustment ranges, and adjustment priorities. The parameter adjustment items include specific parameters to be optimized, such as robot joint stiffness parameters and coating flow control parameters. The adjustment range is dynamically calculated based on the deviation value; for example, the adjustment range increases by 5% for every 10% deviation exceeding a threshold, while also being constrained by the safety range in the equipment characteristic parameter library. The adjustment priority is ranked according to the parameters' impact on mapping accuracy; for example, core process parameters take precedence over auxiliary environmental parameters.

[0030] 400. Within the time period corresponding to the update window length, the virtual parameters of the digital twin model are dynamically updated according to the parameter adjustment strategy to complete the mapping calibration between the digital twin model and the corresponding physical device, and the adjustment and optimization record is fed back to the edge computing node to optimize the extraction strategy of the operating condition fluctuation characteristics; wherein, the update window length is determined based on the parameter adjustment strategy.

[0031] For the parameter adjustment items in the parameter adjustment strategy obtained in step 300, implement layered updates according to the device functional modules: The basic physics layer focuses on the physical parameters of the equipment. For example, based on the adjustment requirements of the robot joint stiffness parameters in the adjustment strategy, it updates the physical properties of the corresponding joints in the virtual model, such as the elastic coefficient and damping coefficient, to ensure that the kinematic characteristics of the virtual equipment are consistent with the physical entity. The process simulation layer optimizes the simulation algorithm parameters. For example, based on the adjustment range of the coating flow control parameters, it corrects the process parameters such as the paint adhesion coefficient and drying rate in the virtual coating model to improve the accuracy of quality prediction values ​​such as paint film thickness and uniformity. The control logic layer updates the logic parameters. For example, it adjusts the PID parameters of the virtual controller according to the welding current compensation rules to make the virtual control logic match the actual execution effect.

[0032] In addition, a virtual simulation verification process can be initiated simultaneously during parameter updates. Based on the updated virtual parameters, key scenarios of the current production conditions are reproduced in the digital twin model, such as the welding trajectory simulation of a welding robot and the stamping process simulation of a stamping machine. By comparing the deviation changes between the simulation results and real-time operating data, the potential impact of parameter adjustments on downstream processes can be predicted. For example, it can be verified whether adjusting the welding current will lead to subsequent assembly interference, ensuring the safety and effectiveness of the adjustment plan. If the simulation verification finds abnormal deviations, the process will be traced back to the parameter adjustment strategy stage to re-optimize the adjustment range or priority.

[0033] The adjustments and optimizations recorded in this study can be stored and fed back using a structured log format. The log entries include key information such as the type of operating condition fluctuation characteristics (e.g., performance degradation characteristics due to equipment aging, and attribute differences between material batches), the original data of the comprehensive deviation value, parameter adjustment items and their values ​​before and after adjustment, the deviation improvement rate after adjustment, and simulation verification results. These records are retained through the local storage module of the edge computing node and simultaneously uploaded to the feature extraction strategy optimization module.

[0034] Edge computing nodes utilize historical adjustment and optimization records, employing incremental learning algorithms to optimize the extraction strategy for operational condition fluctuation features. For example, to address issues such as insufficient sensitivity in recognizing weak aging features and low discriminative power for material batch differences, key parameters of the feature extraction algorithm are dynamically adjusted: for equipment aging features, the window size and threshold settings for vibration signal time-domain analysis are optimized to improve the extraction accuracy of attenuation features such as root mean square value and peak factor; for material batch features, the weight coefficients of texture features in image recognition are adjusted to enhance the discriminative power of surface texture, dimensional deviation, and other attribute differences; for process parameter drift features, the frequency resolution of frequency domain analysis is optimized to improve the extraction efficiency of parameter fluctuation frequency features. Simultaneously, based on the correlation between operational condition fluctuation features and parameter adjustment effects in the adjustment records, the priority ranking of feature extraction is updated. For instance, core process parameter drift features with high impact on mapping accuracy are placed at the beginning of the extraction sequence to ensure that key features are captured first, thereby continuously improving the adaptability of the digital twin model to complex operational condition fluctuations.

[0035] As an example, the deviation evaluation model includes a parameter weight allocation module, a multi-dimensional deviation calculation module, and a comprehensive fusion module; then, calculating the comprehensive deviation value between the simulation results of the digital twin model and the real-time running data through the pre-constructed deviation evaluation model includes: The parameter weight allocation module assigns weight values ​​to core process parameters, equipment status parameters, and environmental auxiliary parameters according to the importance of the process. The multi-dimensional deviation calculation module calculates numerical parameter deviation, curvilinear parameter deviation, and spatial parameter deviation based on the simulation results of the digital twin model and the real-time running data. The integrated fusion module standardizes the numerical parameter deviation, curve parameter deviation, and spatial parameter deviation, and then weights and sums them with the corresponding weight values ​​to obtain the integrated deviation value.

[0036] like Figure 2 As shown, the deviation evaluation model of the present invention includes a parameter weight allocation module, a multi-dimensional deviation calculation module, and a comprehensive fusion module. The specific functions and processing procedures of each component module are as follows: Parameter weighting module: Weight assignment is based on the Analytic Hierarchy Process (AHP). First, a process importance judgment matrix is ​​constructed, and core process parameters (such as welding current and stamping pressure), equipment status parameters (such as vibration and temperature), and environmental auxiliary parameters (such as humidity and cleanliness) are compared pairwise according to their impact on production quality. The weight coefficients are calculated through eigenvalue decomposition.

[0037] For example, core process parameters directly determine product performance, with a weighting coefficient of 0.3-0.4; equipment status parameters are related to equipment operating stability, with a weighting of 0.2-0.3; and environmental auxiliary parameters have an indirect impact, with a weighting of 0.1-0.2, ensuring that the weighting allocation conforms to the process priority logic.

[0038] Multi-dimensional deviation calculation module: Based on the simulation results of the digital twin model and the real-time running data, the following multi-dimensional parameter deviations are calculated respectively.

[0039] (1) Numerical parameter deviation: The basic statistical algorithm is adopted. For single-point numerical parameters such as welding voltage and temperature, when the actual value is non-zero, the relative deviation formula (|virtual value - actual value| / actual value) is used to eliminate the influence of dimensions; when the actual value is zero (such as initial pressure), the absolute deviation formula is used to avoid calculation abnormalities with zero denominator.

[0040] (2) Curved parameter deviation: Based on the dynamic time warping (DTW) algorithm, for time series data such as equipment vibration waveform and temperature change curve, the optimal matching path between the virtual curve and the actual curve is found through dynamic programming, and the sum of the distances of corresponding points on the path is calculated as the morphological deviation, which solves the problem that traditional Euclidean distance is sensitive to time offset.

[0041] (3) Spatial parameter deviation: The Hausdorff distance algorithm is used to calculate the minimum distance between the farthest point pairs in two sets of points (virtual trajectory points and actual trajectory points) for spatial parameters such as welding trajectory and material assembly position, so as to quantify the overall deviation of spatial form, which is especially suitable for non-rigid matching scenarios.

[0042] The comprehensive fusion module first maps the deviation values ​​of each dimension to the [0,1] interval using methods such as min-max standardization, where the standardized value = (deviation value - minimum value) / (maximum value - minimum value)). Then, based on the weight values ​​output by the parameter weight allocation module, a weighted summation algorithm is used to fuse the parameter deviation results from multiple dimensions, ultimately obtaining a comprehensive quantitative index that can be directly used for threshold comparison, ensuring the objectivity and comparability of the deviation assessment. Specifically, the comprehensive deviation value = Σ(standardized deviation value × weight coefficient).

[0043] As an example, based on the matching results between the operating condition fluctuation characteristics and the equipment characteristic parameter library, a parameter adjustment strategy for the corresponding physical equipment is generated, including: The operating condition fluctuation characteristics are matched with typical feature templates pre-stored in the equipment characteristic parameter library using vector similarity, and templates that meet the preset matching degree conditions are selected as matching results. Based on the matching results, the adjustment rule engine of the corresponding type of entity equipment is activated. The adjustment rule engine determines the parameter adjustment items according to the degree of deviation of the operating condition fluctuation characteristics. Specifically: for the performance degradation characteristics caused by equipment aging, physical attribute adjustment items are determined according to the loss law parameters; for the attribute difference characteristics caused by material batch changes, process adaptation adjustment items are determined according to the sensitive parameter range; for the drift characteristics of process parameters affected by the environment and load, control logic adjustment items are determined according to the operating principle parameters. Based on the degree of deviation of the operating condition fluctuation characteristics and the safety threshold in the equipment characteristic parameter library, the parameter adjustment range is generated; based on the degree of influence of the parameters on the mapping accuracy and production quality, the parameter adjustment items are prioritized to form a parameter adjustment strategy that includes parameter adjustment items, adjustment range and adjustment priority.

[0044] In this embodiment, firstly, the extracted operating condition fluctuation features (performance degradation features due to equipment aging, attribute differences in material batches, and drift features of process parameters) are converted into multi-dimensional feature vectors, which are then compared with typical feature template vectors stored in the equipment characteristic parameter library according to equipment type and fluctuation type. A cosine similarity algorithm is used to calculate the similarity between the multi-dimensional feature vectors and these typical feature template vectors. When the similarity reaches a preset matching condition (i.e., exceeds a set threshold), it is determined to be a valid match, and the corresponding typical feature template vector is taken as the matching result.

[0045] Then, each typical feature template vector is associated with a corresponding adjustment rule engine. Based on the matching results, the adjustment rule engine for the corresponding type of entity device can be located. This adjustment rule engine has a built-in logical judgment mechanism associated with the device characteristic parameter library: (1) In response to the performance degradation characteristics caused by equipment aging, the rule engine calls the loss law parameters (such as the model of the change of aging coefficient with running time) in the equipment characteristic parameter library to derive the physical attribute adjustment items (such as the stiffness parameters and damping coefficients of mechanical joints) corresponding to the degree of performance degradation.

[0046] (2) Based on the characteristic differences in attributes caused by batch changes of materials, determine the process adaptation adjustment items (such as conveyor line positioning parameters and processing accuracy compensation parameters) according to the range of sensitive parameters (such as the process parameter range of material adaptation).

[0047] (3) Based on the drift characteristics of process parameters, determine the control logic adjustment items (such as PID controller parameters and dynamic correction coefficients of process parameters) in conjunction with the operating principle parameters (such as the control model of the equipment and energy conversion efficiency) to ensure that the adjustment items correspond precisely to the source of fluctuation.

[0048] Next, based on the degree of deviation of the operating condition fluctuation characteristics (such as the difference from standard characteristics) and combined with the safety thresholds in the equipment characteristic parameter library (such as the maximum range of parameter adjustment), a reasonable adjustment range is calculated through a preset algorithm. This ensures that deviations are effectively eliminated while avoiding exceeding the safe operating boundaries of the equipment. The priority ranking of parameter adjustment items is determined by the weight of the parameter's impact on mapping accuracy (the degree of matching between virtual and physical) and production quality. A multi-factor decision-making method is used to determine the ranking result, prioritizing the adjustment of core process parameters and key equipment status parameters. This ensures that resource investment is focused on the links that have the greatest impact on production monitoring effectiveness, ultimately forming a complete strategy that includes adjustment items, adjustment ranges, and adjustment priorities.

[0049] As an example, within the time period corresponding to the update window length, the virtual parameters of the digital twin model are dynamically updated according to the parameter adjustment strategy, including: The update window length is determined by considering the type, magnitude, and priority of the parameter adjustment items in the parameter adjustment strategy. Within the time period corresponding to the update window length, update operations are performed sequentially on each virtual parameter in the digital twin model, including: S1, for the adjustment items of the basic physical layer, update the kinematic and dynamic parameters of the virtual device; S2, for the adjustment items of the process simulation layer, correct the algorithm parameters of the process simulation model; S3, for the adjustment items of the control logic layer, update the logic parameters of the virtual controller. After each update window period ends, the digital twin model is run for overall simulation verification to confirm whether the updated virtual parameters make the corresponding dimension deviation meet the preset requirements. If not, the window length is recalibrated and the update is performed.

[0050] The basic physical layer, process simulation layer, and control logic layer of a digital twin model are strongly coupled. If each component updates its virtual parameters independently, it will lead to parameter asynchrony, damage to model consistency (e.g., the control logic still responds based on the old parameters after the physical parameters are updated), and may also cause resource contention and difficulties in tracing deviations. Therefore, a unified update window is needed to achieve coordinated updates of layered parameters to avoid the above problems.

[0051] However, during the dynamic update of virtual parameters in a digital twin model, the timeliness requirements for updates vary significantly among different parameter adjustment items: deviations in the basic physical layer parameters may distort the motion trajectory of the virtual device, abnormal parameters in the process simulation layer may affect the accuracy of quality prediction, and drift in the control logic layer parameters may cause control response delays. If a uniform fixed update window is used, either the mapping deviation will be amplified due to untimely updates of critical parameters, or the load on edge computing nodes will increase due to frequent updates of non-critical parameters. To address these technical problems, this invention further determines an appropriate update window length based on a parameter adjustment strategy.

[0052] Within the calculated unified window duration, parameter update operations are performed in an orderly manner according to the model hierarchy to ensure that the parameters of each layer are coordinated and adapted to the physical device status, avoiding mapping deviations caused by parameter asynchrony: For the basic physics layer adjustment items, update the kinematic parameters (such as joint range of motion and conveyor linear velocity curve) and dynamic parameters (such as friction coefficient and stiffness coefficient) of the virtual equipment to make the physical behavior (such as motion trajectory and force characteristics) of the virtual equipment consistent with the current state of the physical equipment, and provide accurate basic data support for the upper-level process simulation.

[0053] For the process simulation layer adjustment items, the process algorithm parameters (such as the welding penetration calculation coefficient and the stamping springback compensation value) are corrected based on the updated physical parameters to ensure that the quality prediction results of the virtual production process (such as weld strength and part size deviation) match the output characteristics of the physical equipment.

[0054] For the control logic layer adjustment items, the logic parameters of the virtual controller (such as control command trigger threshold and execution timing parameters) are adjusted in combination with the update results of the basic physical layer and the process simulation layer, so that the output pattern of the virtual control command is synchronized with the control response characteristics of the physical equipment.

[0055] After each update window period, a comprehensive simulation verification of the digital twin model is initiated. By comparing the virtual full-process operation data with the real-time data collected by the physical equipment, the overall deviation under the coupling effect of parameters at each layer is evaluated to determine whether it is within the preset allowable range (e.g., core parameter deviation ≤ 5%, general parameter deviation ≤ 10%). If the verification fails to meet the standard (i.e., there are dimensional deviations exceeding the preset range), it indicates that the current update window length does not match the parameter coupling adjustment requirements, and the three-dimensional calculation coefficients need to be recalibrated: for example, if the coupling deviation between the basic physical layer and the control logic layer exceeds the standard, the weight of the physical layer type can be increased to 0.5; for parameter items that still exceed the deviation after significant adjustments, the corresponding correction coefficient is increased to 1.8. After calibration, the window length is recalculated and the update operation is performed. Through the closed-loop mechanism of calculation-update-verification-calibration, it is ensured that the unified update window is always dynamically adapted to the status of the physical equipment and the parameter adjustment requirements, so as to maintain the long-term mapping accuracy of the digital twin model.

[0056] As an example, the determination of the update window length based on the type, magnitude, and priority of the parameter adjustment items in the parameter adjustment strategy includes: The Analytic Hierarchy Process (AHP) is used to perform a weighted analysis on the type, adjustment range, and adjustment priority of each parameter adjustment item to obtain the relative importance of each parameter adjustment item. Based on the relative importance, the weight ratio of each parameter adjustment item is assigned. Based on the weight ratio, the normalized values ​​corresponding to the type, adjustment range and adjustment priority of each parameter adjustment item are weighted and calculated to obtain a comprehensive score, which is then mapped to the initial update window length. The initial update window length is fine-tuned based on the real-time load rate of the digital twin model, and the boundary verification of the fine-tuned update window length is performed in combination with the historical adjustment response time in the device characteristic parameter library to ensure that the final update window length is not less than the minimum response threshold and not greater than the maximum delay threshold.

[0057] In this embodiment, the Analytic Hierarchy Process (AHP) is used to systematically weight the three elements of parameter adjustment for each parameter adjustment item in the parameter adjustment strategy, specifically including: The decision target layer is defined by the type, adjustment range, and adjustment priority of parameter adjustment items, while the specific attributes of each element are defined by the criterion layer (e.g., parameter type includes basic physics, process simulation, and control logic; adjustment range includes ≤20%, 20%-50%, and >50%; priority includes levels one, two, and three). Pairwise comparisons are performed on elements at the same level using a 1-9 scale (e.g., the importance of the basic physics layer to the process simulation layer is 2, indicating that the former is slightly more important), forming a judgment matrix. The judgment matrix is ​​normalized, and eigenvectors are calculated using summation or square root methods to obtain the relative importance of parameter type, adjustment range, and priority (e.g., 0.4, 0.35, and 0.25 respectively). The consistency index (CI) and consistency ratio (CR) are calculated. When CR < 0.1, the judgment matrix meets the consistency requirements, and the relative importance is directly used as the weight ratio; otherwise, the judgment matrix is ​​revised and recalculated to finally determine the weight allocation of each element.

[0058] Then, the initial window length is obtained based on the parameter feature mapping, as follows: The specific attributes (type, adjustment range, adjustment priority) of each parameter adjustment item are converted into normalized values ​​in the range of 0-1. For example, the basic physical layer, process simulation layer, and control logic layer are mapped to 1.0, 0.6, and 0.3, respectively; the adjustment range >50%, 20%-50%, and ≤20% are mapped to 1.0, 0.5, and 0.2, respectively; and the first-level, second-level, and third-level priorities are mapped to 1.0, 0.7, and 0.4, respectively.

[0059] The normalized values ​​are weighted and summed according to the weight percentages determined in the aforementioned steps (e.g., 40%, 35%, 25%). The formula is: Comprehensive score = (Type normalized value × 0.4) + (Amplitude normalized value × 0.35) + (Priority normalized value × 0.25).

[0060] Establish a negative correlation mapping rule between the overall score and the window length, such as a score of 1.0 corresponding to 5 seconds, 0.8 corresponding to 10 seconds, 0.6 corresponding to 20 seconds, 0.4 corresponding to 30 seconds, and 0.2 corresponding to 60 seconds. Based on this negative correlation mapping rule, the overall score is converted into a specific initial update window duration.

[0061] Next, dynamic fine-tuning and boundary constraints are used to ensure the practical applicability of the update window length, specifically including: The real-time load rate (average of CPU utilization and memory usage) is collected through the monitoring node interface of the digital twin model. When the load rate is >80%, the initial window length is extended by 10%-20% (e.g., a 10-second window is extended to 11-12 seconds); when the load rate is <30%, it is shortened by 10%-20% (e.g., a 10-second window is shortened to 8-9 seconds); the 30%-80% range remains unchanged to avoid system overload under high load or response lag under low load.

[0062] Additionally, the system retrieves historical response time data for similar adjustments over the past three months from the device's characteristic parameter library, extracting the minimum response threshold (e.g., core parameters not exceeding 8 seconds) and maximum latency threshold (e.g., auxiliary parameters not exceeding 120 seconds) within a 95% confidence interval. The fine-tuned update window length is compared to these thresholds; if it exceeds the range, a forced correction is applied (e.g., if the fine-tuned length is 15 seconds but the minimum response threshold is 8 seconds, it is corrected to 8 seconds). Finally, the system determines the final update window length to satisfy both real-time requirements and historical operational patterns of the device.

[0063] like Figure 3 As shown, this embodiment of the invention also provides an adjustment and optimization system for a digital twin model used for monitoring an automotive production line, applied to an edge computing node. The system includes: The working condition feature extraction unit 1001 collects real-time operating data of each physical equipment in the automobile production line and extracts working condition fluctuation features from the real-time operating data, including performance degradation features caused by equipment aging, attribute differences caused by material batch changes, and drift features of process parameters affected by environment and load. The comprehensive deviation analysis unit 1002 inputs the operating condition fluctuation characteristics into the digital twin model and calculates the comprehensive deviation value between the simulation results of the digital twin model and the real-time operating data through a pre-built deviation evaluation model. The parameter adjustment strategy generation unit 1003 calls the equipment characteristic parameter library when the comprehensive deviation value exceeds the preset threshold. Based on the matching result between the operating condition fluctuation characteristics and the equipment characteristic parameter library, it generates a parameter adjustment strategy for the corresponding physical equipment. The equipment characteristic parameter library contains operating principle parameters, loss law parameters and sensitive parameter ranges for different types of physical equipment. The parameter adjustment strategy includes parameter adjustment items, adjustment range and adjustment priority. The virtual parameter update unit 1004 dynamically updates the virtual parameters of the digital twin model according to the parameter adjustment strategy within the time period corresponding to the update window length, completes the mapping calibration between the digital twin model and the corresponding physical device, and feeds back the adjustment and optimization record to the edge computing node to optimize the extraction strategy of the operating condition fluctuation characteristics; wherein, the update window length is determined based on the parameter adjustment strategy.

[0064] As an example, the deviation assessment model includes a parameter weight allocation module, a multi-dimensional deviation calculation module, and a comprehensive fusion module; then the comprehensive deviation analysis unit 1002 specifically: The parameter weight allocation module assigns weight values ​​to core process parameters, equipment status parameters, and environmental auxiliary parameters according to the importance of the process. The multi-dimensional deviation calculation module calculates numerical parameter deviation, curvilinear parameter deviation, and spatial parameter deviation based on the simulation results of the digital twin model and the real-time running data. The integrated fusion module standardizes the numerical parameter deviation, curve parameter deviation, and spatial parameter deviation, and then weights and sums them with the corresponding weight values ​​to obtain the integrated deviation value.

[0065] As an example, the parameter adjustment strategy generation unit 1003 specifically: The operating condition fluctuation characteristics are matched with typical feature templates pre-stored in the equipment characteristic parameter library using vector similarity, and templates that meet the preset matching degree conditions are selected as matching results. Based on the matching results, the adjustment rule engine of the corresponding type of entity equipment is activated. The adjustment rule engine determines the parameter adjustment items according to the degree of deviation of the operating condition fluctuation characteristics. Specifically: for the performance degradation characteristics caused by equipment aging, physical attribute adjustment items are determined according to the loss law parameters; for the attribute difference characteristics caused by material batch changes, process adaptation adjustment items are determined according to the sensitive parameter range; for the drift characteristics of process parameters affected by the environment and load, control logic adjustment items are determined according to the operating principle parameters. Based on the degree of deviation of the operating condition fluctuation characteristics and the safety threshold in the equipment characteristic parameter library, the parameter adjustment range is generated; based on the degree of influence of the parameters on the mapping accuracy and production quality, the parameter adjustment items are prioritized to form a parameter adjustment strategy that includes parameter adjustment items, adjustment range and adjustment priority.

[0066] As an example, the virtual parameter update unit 1004 specifically: The update window length is determined by considering the type, magnitude, and priority of the parameter adjustment items in the parameter adjustment strategy. Within the time period corresponding to the update window length, update operations are performed sequentially on each virtual parameter in the digital twin model, including: S1, for the adjustment items of the basic physical layer, update the kinematic and dynamic parameters of the virtual device; S2, for the adjustment items of the process simulation layer, correct the algorithm parameters of the process simulation model; S3, for the adjustment items of the control logic layer, update the logic parameters of the virtual controller. After each update window period ends, the digital twin model is run for overall simulation verification to confirm whether the updated virtual parameters make the corresponding dimension deviation meet the preset requirements. If not, the window length is recalibrated and the update is performed.

[0067] As an example, the virtual parameter update unit 1004 specifically: The Analytic Hierarchy Process (AHP) is used to perform a weighted analysis on the type, adjustment range, and adjustment priority of each parameter adjustment item to obtain the relative importance of each parameter adjustment item. Based on the relative importance, the weight ratio of each parameter adjustment item is assigned. Based on the weight ratio, the normalized values ​​corresponding to the type, adjustment range and adjustment priority of each parameter adjustment item are weighted and calculated to obtain a comprehensive score, which is then mapped to the initial update window length. The initial update window length is fine-tuned based on the real-time load rate of the digital twin model, and the boundary verification of the fine-tuned update window length is performed in combination with the historical adjustment response time in the device characteristic parameter library to ensure that the final update window length is not less than the minimum response threshold and not greater than the maximum delay threshold.

[0068] This invention also provides an electronic device for use in an edge computing node. The electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor. When the computer program is executed by the processor, it implements the method as described in any of the foregoing embodiments.

[0069] This invention also provides a computer storage medium for use in an edge computing node, the computer storage medium storing a computer program that can be executed by a processor to implement the methods described in any of the foregoing claims.

[0070] This invention also provides a computer program product applied to an edge computing node, the computer program product comprising a computer program executable by a processor to implement the methods described in any of the foregoing claims.

[0071] Although the invention has been specifically shown and described with reference to preferred embodiments, those skilled in the art will understand that various modifications in form and detail may be made without departing from the spirit and scope of the invention. Accordingly, the disclosed invention should be considered merely illustrative and limited only by the scope specified in the appended claims.

Claims

1. A method for adjusting and optimizing a digital twin model for monitoring an automotive production line, applied to edge computing nodes, characterized in that: The methods and steps include the following: Real-time operating data of various physical equipment in the automobile production line is collected, and operating condition fluctuation characteristics are extracted from the real-time operating data, including performance degradation characteristics caused by equipment aging, attribute differences caused by material batch changes, and drift characteristics of process parameters affected by environment and load. The operating condition fluctuation characteristics are input into the digital twin model, and the comprehensive deviation value between the simulation results of the digital twin model and the real-time operating data is calculated through a pre-built deviation evaluation model. When the overall deviation value exceeds the preset threshold, the equipment characteristic parameter library is invoked. Based on the matching result between the operating condition fluctuation characteristics and the equipment characteristic parameter library, a parameter adjustment strategy for the corresponding physical equipment is generated. The equipment characteristic parameter library contains operating principle parameters, loss law parameters and sensitive parameter ranges for different types of physical equipment. The parameter adjustment strategy includes parameter adjustment items, adjustment range and adjustment priority. Within the time period corresponding to the update window length, the virtual parameters of the digital twin model are dynamically updated according to the parameter adjustment strategy to complete the mapping calibration between the digital twin model and the corresponding physical device, and the adjustment and optimization record is fed back to the edge computing node to optimize the extraction strategy of the operating condition fluctuation characteristics; wherein, the update window length is determined based on the parameter adjustment strategy.

2. The method for adjusting and optimizing a digital twin model for monitoring an automobile production line according to claim 1, characterized in that: The deviation evaluation model includes a parameter weight allocation module, a multi-dimensional deviation calculation module, and a comprehensive fusion module; the calculation of the comprehensive deviation value between the simulation results of the digital twin model and the real-time running data through the pre-constructed deviation evaluation model includes: The parameter weight allocation module assigns weight values ​​to core process parameters, equipment status parameters, and environmental auxiliary parameters according to the importance of the process. The multi-dimensional deviation calculation module calculates numerical parameter deviation, curvilinear parameter deviation, and spatial parameter deviation based on the simulation results of the digital twin model and the real-time running data. The integrated fusion module standardizes the numerical parameter deviation, curve parameter deviation, and spatial parameter deviation, and then weights and sums them with the corresponding weight values ​​to obtain the integrated deviation value.

3. The method for adjusting and optimizing a digital twin model for monitoring an automobile production line according to claim 1, characterized in that: Based on the matching results between the operating condition fluctuation characteristics and the equipment characteristic parameter library, a parameter adjustment strategy for the corresponding physical equipment is generated, including: The operating condition fluctuation characteristics are matched with typical feature templates pre-stored in the equipment characteristic parameter library using vector similarity, and templates that meet the preset matching degree conditions are selected as matching results. Based on the matching results, the adjustment rule engine of the corresponding type of entity equipment is activated. The adjustment rule engine determines the parameter adjustment items according to the degree of deviation of the operating condition fluctuation characteristics. Specifically: for the performance degradation characteristics caused by equipment aging, physical attribute adjustment items are determined according to the loss law parameters; for the attribute difference characteristics caused by material batch changes, process adaptation adjustment items are determined according to the sensitive parameter range; for the drift characteristics of process parameters affected by the environment and load, control logic adjustment items are determined according to the operating principle parameters. Based on the degree of deviation of the operating condition fluctuation characteristics and the safety threshold in the equipment characteristic parameter library, the parameter adjustment range is generated; based on the degree of influence of the parameters on the mapping accuracy and production quality, the parameter adjustment items are prioritized to form a parameter adjustment strategy that includes parameter adjustment items, adjustment range and adjustment priority.

4. The method for adjusting and optimizing a digital twin model for monitoring an automobile production line according to claim 3, characterized in that: Within the time period corresponding to the update window length, the virtual parameters of the digital twin model are dynamically updated according to the parameter adjustment strategy, including: The update window length is determined by considering the type, magnitude, and priority of the parameter adjustment items in the parameter adjustment strategy. Within the time period corresponding to the update window length, update operations are performed sequentially on each virtual parameter in the digital twin model, including: S1, for the adjustment items of the basic physical layer, update the kinematic and dynamic parameters of the virtual device; S2, for the adjustment items of the process simulation layer, correct the algorithm parameters of the process simulation model; S3, for the adjustment items of the control logic layer, update the logic parameters of the virtual controller. After each update window period ends, the digital twin model is run for overall simulation verification to confirm whether the updated virtual parameters make the corresponding dimension deviation meet the preset requirements. If not, the window length is recalibrated and the update is performed.

5. The method for adjusting and optimizing a digital twin model for monitoring an automobile production line according to claim 4, characterized in that: Based on the type, magnitude, and priority of the parameter adjustment items in the aforementioned parameter adjustment strategy, the length of the update window is determined, including: The Analytic Hierarchy Process (AHP) is used to perform a weighted analysis on the type, adjustment range, and adjustment priority of each parameter adjustment item to obtain the relative importance of each parameter adjustment item. Based on the relative importance, the weight ratio of each parameter adjustment item is assigned. Based on the weight ratio, the normalized values ​​corresponding to the type, adjustment range and adjustment priority of each parameter adjustment item are weighted and calculated to obtain a comprehensive score, which is then mapped to the initial update window length. The initial update window length is fine-tuned based on the real-time load rate of the digital twin model, and the boundary verification of the fine-tuned update window length is performed in combination with the historical adjustment response time in the device characteristic parameter library to ensure that the final update window length is not less than the minimum response threshold and not greater than the maximum delay threshold.

6. An adjustment and optimization system for a digital twin model used for monitoring an automobile production line, applied to edge computing nodes, characterized in that, The system includes: The working condition feature extraction unit collects real-time operating data of various physical equipment in the automobile production line and extracts working condition fluctuation features from the real-time operating data, including performance degradation features caused by equipment aging, attribute differences caused by material batch changes, and drift features of process parameters affected by environment and load. The comprehensive deviation analysis unit inputs the operating condition fluctuation characteristics into the digital twin model and calculates the comprehensive deviation value between the simulation results of the digital twin model and the real-time operating data through a pre-built deviation evaluation model. The parameter adjustment strategy generation unit calls the equipment characteristic parameter library when the comprehensive deviation value exceeds a preset threshold. Based on the matching result between the operating condition fluctuation characteristics and the equipment characteristic parameter library, it generates a parameter adjustment strategy for the corresponding physical equipment. The equipment characteristic parameter library contains operating principle parameters, loss law parameters and sensitive parameter ranges for different types of physical equipment. The parameter adjustment strategy includes parameter adjustment items, adjustment range and adjustment priority. The virtual parameter update unit dynamically updates the virtual parameters of the digital twin model according to the parameter adjustment strategy within the time period corresponding to the update window length, completes the mapping calibration between the digital twin model and the corresponding physical device, and feeds back the adjustment and optimization record to the edge computing node to optimize the extraction strategy of the operating condition fluctuation characteristics; wherein, the update window length is determined based on the parameter adjustment strategy.

7. The adjustment and optimization system for a digital twin model for monitoring an automobile production line according to claim 6, characterized in that: The deviation assessment model includes a parameter weight allocation module, a multi-dimensional deviation calculation module, and a comprehensive fusion module; the comprehensive deviation analysis unit specifically includes: The parameter weight allocation module assigns weight values ​​to core process parameters, equipment status parameters, and environmental auxiliary parameters according to the importance of the process. The multi-dimensional deviation calculation module calculates numerical parameter deviation, curvilinear parameter deviation, and spatial parameter deviation based on the simulation results of the digital twin model and the real-time running data. The integrated fusion module standardizes the numerical parameter deviation, curve parameter deviation, and spatial parameter deviation, and then weights and sums them with the corresponding weight values ​​to obtain the integrated deviation value.

8. An electronic device applied to an edge computing node, the electronic device comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, characterized in that: when the computer program is executed by the processor, it implements the method as described in any one of claims 1-5.

9. A computer storage medium applied to an edge computing node, characterized in that: The computer storage medium stores a computer program that can be executed by a processor to implement the method as described in any one of claims 1-5.

10. A computer program product applied to edge computing nodes, characterized in that: The computer program product includes a computer program that can be executed by a processor to implement the method as described in any one of claims 1-5.

Images