Welding design closed-loop feedback verification optimization method and system based on digital twinning
Patent Information
- Application Number
- CN202511427426.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-30
- Publication Date
- 2026-02-06
Smart Images
Figure CN121480239A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of computer-aided design and machine learning technology, specifically to a closed-loop feedback verification and optimization method and system for welding design based on digital twins. Background Technology
[0002] With the advent of the era of intelligent manufacturing, the design quality of welding production lines has become the key to determining product quality and production efficiency. Traditional verification methods that rely on physical trial and error and offline simulation are no longer suitable for the rapidly iterating market demand due to their long cycle and high cost. Digital twin technology, as the core of digital transformation, is expected to achieve in-depth verification and optimization in the design stage.
[0003] Currently, the industry has attempted to apply digital twin technology to construct virtual models of welding production lines for preliminary simulation analysis. Related technological development focuses on the geometric visualization of the model and basic process simulation, and has achieved certain results in static resource allocation and rule scheduling. However, these existing technologies still have significant limitations in terms of model depth, system intelligence, and closed-loop application.
[0004] These existing technologies mainly suffer from the following prominent problems: First, there are discrepancies between the digital twin model and the physical entity, and there is a lack of effective online correction mechanisms, resulting in insufficient credibility of simulation results; second, the verification process lacks a quantitative comprehensive performance evaluation system, and the optimization direction is unclear; most importantly, the entire process fails to form an automatic closed loop of "evaluation-optimization-verification", the resource scheduling strategy is rigid, and it is impossible to make dynamic adjustments based on simulation feedback, resulting in low design optimization efficiency and difficulty in continuously compressing the design cycle and simultaneously improving quality.
[0005] In summary, existing welding design verification methods have significant shortcomings in terms of accuracy, intelligence, and closed-loop automation, and cannot meet the urgent needs of efficient and high-quality design. Summary of the Invention
[0006] To address the aforementioned issues, this invention proposes a closed-loop feedback verification and optimization method and system for welding design based on digital twins. By utilizing digital twin technology, online learning algorithms, and dynamic resource scheduling, the design cycle is significantly shortened, the defect rate of the designed welding production line is reduced, and the performance and stability of the welding production line are continuously improved.
[0007] According to some embodiments, the present invention adopts the following technical solution: A closed-loop feedback verification and optimization method for welding design based on digital twins includes: Obtain the design scheme to be optimized for the target welding production line, and build a digital twin model for the design scheme; Data from the simulation process of the digital twin model is collected and divided into several time segments. Through a multi-dimensional performance index calculation system, the response latency, error rate, defect rate and corresponding performance score of each time segment are calculated to obtain the performance time series. Based on performance time series, the optimal response latency, error rate, and defect rate are determined. Based on the optimal response delay, error rate, and defect rate, the digital twin model is dynamically optimized for resource scheduling to obtain the resource scheduling strategy of the current digital twin model, thus completing the closed-loop feedback verification and optimization of the welding design scheme.
[0008] According to some embodiments, the present invention adopts the following technical solution: A closed-loop feedback verification and optimization system for welding design based on digital twins includes: The model building module is configured to: acquire the design scheme to be optimized for the target welding production line, and build a digital twin model for the design scheme; The performance evaluation module is configured to: collect data during the simulation operation of the digital twin model, divide it into several time segments, and calculate the response latency, error rate, defect rate and corresponding performance score of each time segment through a multi-dimensional performance index calculation system to obtain the performance time series; The performance solution module is configured to: solve for the optimal response latency, error rate, and defect rate based on the performance time series. The resource optimization module is configured to: perform dynamic resource scheduling optimization on the digital twin model based on the optimal response latency, error rate, and defect rate, obtain the resource scheduling strategy of the current digital twin model, and complete the closed-loop feedback verification optimization of the welding design scheme.
[0009] According to some embodiments, the present invention adopts the following technical solution: A computer program product includes a computer program that, when executed by a processor, implements the aforementioned digital twin-based welding design closed-loop feedback verification optimization method.
[0010] According to some embodiments, the present invention adopts the following technical solution: A non-transitory computer-readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the aforementioned closed-loop feedback verification and optimization method for welding design based on digital twins.
[0011] According to some embodiments, the present invention adopts the following technical solution: An electronic device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the aforementioned closed-loop feedback verification and optimization method for welding design based on digital twins.
[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Significantly improve verification efficiency and quality: By constructing an automated closed loop of "simulation-evaluation-optimization", the traditional serial design mode that relies on physical trial and error and has a long cycle is transformed into a highly efficient digital parallel iterative mode, which greatly shortens the design cycle. At the same time, the accurate evaluation and feedback based on a multi-dimensional quantitative indicator system ensures that the quality of the design solution is optimized from the source, and achieves simultaneous improvement in efficiency and quality.
[0013] 2. The system achieves intelligent self-adaptation and continuous optimization: This invention innovatively integrates online learning and prediction models (such as ARIMA) into the optimization process, enabling the system to predict trends and dynamically adjust optimization objectives based on historical performance data, and transform error data into a driving force for model self-improvement. This gives the digital twin model the ability to continuously learn from data and evolve, significantly improving the foresight and accuracy of optimization.
[0014] 3. The optimization process shifts from experience-driven to data-driven: This method automatically transforms optimal parameters into specific simulation resource allocation schemes through dynamic resource scheduling strategies. All decisions are based on data and algorithms, avoiding the subjectivity and limitations of traditional manual decision-making. This achieves intelligent resource allocation and design verification processes, ensuring the scientific validity and reliability of the optimization results. Attached Figure Description
[0015] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.
[0016] Figure 1 This is a flowchart of the method in Example 1. Detailed Implementation The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0017] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0018] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0019] Example 1 One embodiment of the present invention provides a closed-loop feedback verification and optimization method for welding design based on digital twins, comprising: Step S1: Obtain the design scheme to be optimized for the target welding production line and build a digital twin model for the design scheme.
[0020] For subsequent optimization, a high-fidelity, interactive, and data-driven virtual model of the welding production line will be created using digital twins, specifically as follows: 1. Obtain the design scheme to be optimized, including CAD data, process planning data, control system logic, and factory layout diagram.
[0021] Specifically, CAD data of the three-dimensional geometric model of the production line (such as fixtures, robots, and conveying equipment) is obtained from the design department; process planning data such as work instructions, time quotas, welding parameters (current, voltage, speed), process sequence, and cycle time requirements are obtained from the process department; control system logic such as PLC ladder diagrams and robot control programs is obtained to simulate the control logic and behavior of the equipment; and factory layout diagrams such as the two-dimensional layout of the plant are obtained to accurately locate the equipment.
[0022] 2. Digital twin model construction (multi-dimensional modeling), including geometric model, physical model, behavioral model, and data model.
[0023] Specifically, CAD data is imported into a professional digital twin platform for lightweight model processing, geometric model construction, and smooth real-time rendering. Motion joints and constraints are defined for moving parts such as robots and cylinders to simulate physical effects such as gravity, collisions, and vibrations, thus constructing a physical model. Based on PLC programs, the control logic of the production line is established in the digital twin environment, including sensor triggering, robot motion sequences, and conveyor belt start / stop. A welding process model is integrated to simulate the welding process and preliminarily predict weld quality (e.g., simulating spatter based on current and voltage), thus constructing a behavioral model. All data points that need to be collected (data twin) are defined, and attributes and interfaces are created for each data point, such as robot status (running / stopping / alarm), cycle time, energy consumption counter, and virtual sensor readings, resulting in a data model.
[0024] The data points are time-series data (such as current, voltage, humidity, air pressure, etc.) continuously collected by virtual sensors set up at key locations (such as robot end effector, welding torch, fixture positioning point).
[0025] Step S2: Collect data during the simulation of the digital twin model and divide it into several time series segments. Calculate the response latency, error rate, defect rate, and corresponding performance score for each time series segment using a multi-dimensional performance index calculation system to obtain the performance time series.
[0026] Specifically, the process involves driving the digital twin model to run a simulation, dividing the simulated data stream into several time-series segments, and accurately calculating the response latency, error rate, defect rate, and overall performance score for each segment. The specific steps include: 1. Time-series data acquisition and partitioning The production process of the welding production line is simulated using a digital twin model, and continuous simulation data streams are collected. The collected multi-source data includes: Welding process data: Acquired from the control logic and physics engine of the twin model, including current, voltage, welding speed, predicted temperature field, etc. (simulated 1kHz sampling).
[0027] Geometric measurement data: Acquired from the output of virtual sensors (such as laser scanners and cameras) of the twin model to obtain the position, orientation, key dimensions, etc. of the components (simulating 10Hz sampling).
[0028] Material property data: collected from integrated material models or external CAE analysis results, such as simulated hardness distribution, strength properties, and microstructure evolution.
[0029] Environmental parameter data: collected from the model's environmental settings, such as ambient temperature, humidity, and protective gas flow rate.
[0030] After data collection, in order to improve the accuracy of subsequent data, preprocessing is performed on the multi-source data, including timestamp-based data alignment, filtering-based denoising, and outlier removal.
[0031] After preprocessing, the continuous simulation data stream is divided into segments based on the pre-set data window length to obtain several time-series segments.
[0032] 2. Calculation of performance metrics for a single time segment After segmentation, the performance of each time series segment is evaluated from three dimensions based on a multi-dimensional performance index calculation system, specifically: Dimension 1: Response Latency Assessment Specifically, this is used to evaluate the overall efficiency and real-time performance of digital twin models from receiving input to generating feedback: First, the time difference between the end and start times of each stage is used to calculate the latency of each stage, including the design response latency T_design, the simulation calculation latency T_simulation, and the verification feedback latency T_validation. The design response latency is the design baseline latency, a theoretical upper limit roughly calculated based on algorithm complexity, CPU frequency, and memory bandwidth. The simulation calculation latency is the simulation measured latency, the time taken up by simulation software, which includes additional time for memory and CPU queuing compared to the baseline latency. The verification feedback latency is the on-site calibration latency, the true value of the production environment obtained by timing the device after it arrives on-site.
[0033] Then, the total delay T_e2e is calculated, expressed by the formula: T_e2e = T_design + T_simulation + T_validation Finally, the delay performance metric Delay_Performance is calculated and expressed by the formula:
[0034] in, T_target is the target latency, used to examine whether the latency meets the standard. Consistency_Score is the quantitative score of the latency fluctuation within the simulation time series. The closer the value of Delay_Performance is to 1, the better the performance.
[0035] Dimension Two: Statistical Analysis of Error Distribution Used to evaluate the prediction accuracy of digital twin models in geometry, physics, and manufacturing processes, specifically: First, calculate the various errors (Error_Distribution), including: E_geo (geometric error): E_geo =|P_predicted - P_actual| / P_actual, where P_predicted is the simulated position and P_actual is the reference position.
[0036] E_material (Material property error): ,in, To simulate performance, This is the baseline performance.
[0037] E_process (process parameter error): , in, For simulation parameters, These are the baseline parameters.
[0038] E_thermal (Temperature Field Error): E_thermal = RMSE(T_field_sim, T_field_measured), uses the root mean square error to compare the simulated temperature field with the reference temperature field.
[0039] Finally, calculate the total error assessment (Total_Error):
[0040] This metric integrates errors of different types and obtains an overall fidelity index by taking the square root of the weighted sum of squares. The smaller the value of Total_Error, the higher the model accuracy.
[0041] Dimension 3: Quantitative Assessment of Defect Rate Specifically, this is used to evaluate the accuracy of digital twin models in predicting production defects: First, a confusion matrix (Defect_Rate_Analysis) is constructed, which compares the defect results (pass / fail) predicted by the twin model with the baseline results, and counts the true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN). Here, the defect results represent the model's prediction hit rate for defects in the welding scenario.
[0042] Then, the prediction accuracy and recall are calculated, expressed by the following formulas:
[0043] Finally, the F1 score is calculated, expressed by the formula:
[0044] The F1 score is a comprehensive indicator that balances prediction accuracy and recall.
[0045] 3. Calculation of overall performance score The evaluation results from the three dimensions are combined into a single comprehensive performance score, which is used to intuitively compare the merits of different models. Specifically: First, normalization: normalize the three metrics Delay_Performance, (1 - Total_Error) (converting error to precision), and F1 score to the [0, 1] interval.
[0046] Then, weighted summation: assign weights to the three dimensions according to project requirements (e.g., β1, β2, β3, and β1+β2+β3=1).
[0047] Finally, the overall performance score is calculated using the following formula: .
[0048] 4. Output The calculation results of all time segments (response delay, error rate, defect prediction F1 value and overall performance score of each time segment) are arranged in chronological order to form a complete performance time series.
[0049] Step S3: Based on the performance time series, solve for the optimal response delay, error rate, and defect rate.
[0050] Analyzing the performance time series generated in step S2, and combining dynamic thresholding and online learning techniques, the system intelligently solves for the optimal combination of response latency, error rate, and defect rate parameters that enable the digital twin model to achieve globally optimal performance. Specifically: 1. Data Preparation and Optimization Problem Definition (1) Input data construction: Performance Time Series (Y): Receive the comprehensive performance score sequence for each time segment arranged in chronological order from step S2. .
[0051] System Parameter Recording (X): Synchronously records the system configuration parameters (such as robot movement speed, virtual sensor accuracy, etc.) corresponding to each time segment simulation run. These parameters directly affect the final response delay, error rate, and defect rate.
[0052] (2) Definition of optimization problem: Objective: To find a set of system parameter configurations (Config_opt) that maximizes the corresponding prediction performance score.
[0053] Constraints: Under this configuration, the response latency, error rate, and defect rate must meet the hard thresholds (T_hard, E_hard, DR_hard) set in the multi-level threshold system (Threshold_System), that is, they must not exceed the quality baseline.
[0054] 2. ARIMA Trend Forecasting and Dynamic Threshold Coordination Based on the above input data and the defined optimization problem, the future trend of performance is predicted, and a dynamic target, i.e. a soft threshold, is set for optimization.
[0055] (1) Performance trend prediction: An ARIMA time series model is built for the performance score sequence Y, and the model is used to predict the performance score Y_predicted for the next (or the next few) time series segments, which indicates the natural direction of performance evolution.
[0056] (2) Dynamic optimization of target setting (Threshold Adaptation): Analyze the performance scores of several recent time series segments and combine the ARIMA-predicted trend Y_predicted with the current soft threshold (T_soft, E_soft, DR_soft).
[0057] According to the formula The "expected target" of dynamic adjustment optimization, i.e. the soft threshold, is that if the predicted trend is positive, the optimization algorithm should pursue a more stringent (lower) T_soft, E_soft, DR_soft target.
[0058] 3. Agent Model Construction and Online Learning Optimization (1) Agent model construction and incremental learning: A complex nonlinear mapping model (such as a Transformer regression model) is established, whose input is the system parameter configuration X and output is the predicted performance score, i.e., Score = f(X). This surrogate model replaces the time-consuming full simulation, making rapid optimization possible.
[0059] This regression model is updated online to ensure its prediction accuracy. Whenever step S2 generates a number of new time-series data segments (new X, Y pairs), an update of the surrogate model is triggered. An incremental learning approach is adopted, using new data and some key historical data (selected in a 1:4 ratio through gradient-based sample importance scoring) for mixed training. To ensure stability, cosine annealing learning rate scheduling, L2 regularization, and elastic weight merging techniques are applied to ensure that the model does not suffer catastrophic forgetting while absorbing new knowledge, thus maintaining global stability.
[0060] (2) Constraint optimization solution: Under hard threshold constraints, global optimization algorithms such as Bayesian optimization or genetic algorithms are used to search within the space of system parameters.
[0061] The goal is to maximize the performance score predicted by the surrogate model f(X) and make it approximate the forward-looking objective jointly set by the ARIMA model and the dynamic soft threshold, ultimately outputting a set of optimal system parameter configurations Config_opt. The response latency, error rate, and defect rate corresponding to this configuration are the "optimal response latency, error rate, and defect rate" obtained in this step.
[0062] Step S4: Based on the optimal response delay, error rate, and defect rate, perform dynamic resource scheduling optimization on the digital twin model to obtain the resource scheduling strategy of the current digital twin model, and complete the closed-loop feedback verification and optimization of the welding design scheme.
[0063] Using the optimal response delay, error rate, and defect rate obtained in step S3 as inputs, an intelligent algorithm dynamically schedules and configures the simulation resources within the digital twin model to generate an optimal resource scheduling strategy and complete closed-loop verification. Specifically: 1. Target conversion and real-time performance monitoring (1) Optimize target transformation: The optimal response latency, error rate, and defect rate are transformed into quantifiable multi-dimensional performance metrics (PerformanceMetrics) within the digital twin model, including: Quality indicator: Quality = (1 - Error_rate) × (1 - Defect_rate), Resource utilization rate: Utilization = Resource_used / Resource_allocated Throughput: Throughput = Jobs_completed / Hour Where Quality is the overall quality score, Error_rate is the normalized value of the overall error of the current simulation batch, Defect_rate is the proportion of missed defects in the current batch; Utilization is the resource utilization rate, Resource_used is the actual computing power units consumed within the period, Resource_allocated is the computing power units reserved by the scheduler for this simulation task within the period; Throughput is the task throughput, Jobs_completed is the number of solder joint-level simulation task packages successfully completed within 1 hour, and Hour is a fixed time window that can be slidable.
[0064] Therefore, the optimization objective is clear: to maximize the overall performance of Quality, Utilization, and Throughput while meeting computing resource constraints (e.g., CPU < 80%).
[0065] (2) Real-time performance curve analysis (TrendPrediction): Real-time monitoring of multi-dimensional performance metrics for digital twin simulation tasks.
[0066] LSTM networks are used for short-term forecasting to predict load and performance trends in the next few minutes. Meanwhile, Isolation Forest is used to detect performance anomalies (such as a simulation task suddenly freezing), providing forward-looking information for dynamic scheduling.
[0067] 2. Execution of intelligent resource scheduling algorithm Using multi-objective optimization scheduling (ResourceScheduling), under the upper limit of hardware resources (CPU≤80%, memory≤90%, etc.) and the hard threshold in step S1, we solve for the comprehensive maximization of Quality, Utilization and Throughput. The objective function is defined as Maximize ( w1*Throughput + w2*Utilization + w3*Quality ), and the weights (w1, w2, w3) are adjusted according to the focus of the project stage (such as efficiency in the early stage and quality in the later stage).
[0068] A hybrid strategy of genetic algorithm (GA) and particle swarm optimization (PSO) is used to solve the problem. GA is responsible for global search, while PSO is used for local fine-tuning, which quickly solves the optimal resource allocation scheme, including computational resource allocation, network bandwidth allocation and memory pre-allocation.
[0069] Computing resources are allocated to high-priority simulation tasks (such as bottleneck workstations) with more CPU cores / higher computing frequencies. Network bandwidth allocation ensures that large-scale data (such as high-precision point clouds) will not become a bottleneck during synchronization. Memory pre-allocation reserves memory in advance for large simulation tasks that are about to begin, avoiding interruptions.
[0070] 3. Closed-loop feedback verification optimization The generated dynamic resource scheduling strategy is applied to the digital twin model, driving the model to re-run the simulation under the new resource configuration. Performance data from the new simulation is collected, and the process returns to step S2 to calculate new performance metrics and scores. The new scores are then compared with those before optimization. If the overall performance score significantly improves and reaches or exceeds the predicted trend of the ARIMA model, the dynamic resource scheduling optimization is considered successful. If the target is not achieved, or a new bottleneck is detected, a new optimization cycle (S2→S3→S4) is triggered until the design meets all requirements.
[0071] Example 2 One embodiment of the present invention provides a welding design closed-loop feedback verification and optimization system based on digital twins, comprising: The model building module is configured to: acquire the design scheme to be optimized for the target welding production line, and build a digital twin model for the design scheme; The performance evaluation module is configured to: collect data during the simulation operation of the digital twin model, divide it into several time segments, and calculate the response latency, error rate, defect rate and corresponding performance score of each time segment through a multi-dimensional performance index calculation system to obtain the performance time series; The performance solution module is configured to: solve for the optimal response latency, error rate, and defect rate based on the performance time series. The resource optimization module is configured to: perform dynamic resource scheduling optimization on the digital twin model based on the optimal response latency, error rate, and defect rate, obtain the resource scheduling strategy of the current digital twin model, and complete the closed-loop feedback verification optimization of the welding design scheme.
[0072] Example 3 One embodiment of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned digital twin-based welding design closed-loop feedback verification optimization method.
[0073] Example 4 In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided for storing computer instructions. When the computer instructions are executed by a processor, the described closed-loop feedback verification and optimization method for welding design based on digital twins is implemented.
[0074] Example 5 One embodiment of the present invention provides an electronic device, including: a processor, a memory, and a computer program; wherein, the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the aforementioned closed-loop feedback verification and optimization method for welding design based on digital twins.
[0075] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0076] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0077] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.
Claims
1. A closed-loop feedback verification and optimization method for welding design based on digital twins, characterized in that, include: Obtain the design scheme to be optimized for the target welding production line, and build a digital twin model for the design scheme; Data from the simulation process of the digital twin model is collected and divided into several time segments. Through a multi-dimensional performance index calculation system, the response latency, error rate, defect rate and corresponding performance score of each time segment are calculated to obtain the performance time series. Based on performance time series, the optimal response latency, error rate, and defect rate are determined. Based on the optimal response delay, error rate, and defect rate, the digital twin model is dynamically optimized for resource scheduling to obtain the resource scheduling strategy of the current digital twin model, thus completing the closed-loop feedback verification and optimization of the welding design scheme.
2. The welding design closed-loop feedback verification and optimization method based on digital twin as described in claim 1, characterized in that, The simulation data stream collected during the operation of the digital twin model includes welding process data, geometric measurement data, material property data, and environmental parameter data.
3. The welding design closed-loop feedback verification and optimization method based on digital twin as described in claim 1, characterized in that, The system utilizes a multi-dimensional performance index calculation framework to calculate the response latency, error rate, defect rate, and corresponding performance score for each time segment, thereby obtaining a performance time series. Specifically: For each time segment, scores are calculated for response delay performance, comprehensive error, and defect rate. The scores of the three indicators are then combined to obtain the performance score for each time segment.
4. The welding design closed-loop feedback verification and optimization method based on digital twin as described in claim 1, characterized in that, The optimal response delay, error rate, and defect rate are obtained by using the ARIMA model to dynamically and adaptively adjust the soft threshold, and then using the adjusted soft threshold as the optimization target to find the optimal response delay, error rate, and defect rate.
5. The welding design closed-loop feedback verification and optimization method based on digital twin as described in claim 1, characterized in that, The dynamic resource scheduling optimization of the digital twin model specifically includes: The optimal response latency, error rate, and defect rate are transformed into quantifiable multidimensional performance indicators within the digital twin model. We employ multi-objective optimization scheduling to maximize the comprehensive performance of quantifiable multi-dimensional indicators under the constraints of hardware resource limits and hard thresholds.
6. The welding design closed-loop feedback verification and optimization method based on digital twin as described in claim 5, characterized in that, The solution is obtained by a hybrid strategy of genetic algorithm and particle swarm optimization. The genetic algorithm is responsible for global search, while the particle swarm optimization is used for local fine-tuning to find the optimal resource allocation scheme, including computational resource allocation, network bandwidth allocation, and memory pre-allocation.
7. A welding design closed-loop feedback verification and optimization system based on digital twins, characterized in that, include: The model building module is configured to: acquire the design scheme to be optimized for the target welding production line, and build a digital twin model for the design scheme; The performance evaluation module is configured to: collect data during the simulation operation of the digital twin model, divide it into several time segments, and calculate the response latency, error rate, defect rate and corresponding performance score of each time segment through a multi-dimensional performance index calculation system to obtain the performance time series; The performance solution module is configured to: solve for the optimal response latency, error rate, and defect rate based on the performance time series. The resource optimization module is configured to: perform dynamic resource scheduling optimization on the digital twin model based on the optimal response latency, error rate, and defect rate, obtain the resource scheduling strategy of the current digital twin model, and complete the closed-loop feedback verification optimization of the welding design scheme.
8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the closed-loop feedback verification and optimization method for welding design based on digital twins as described in any one of claims 1-6.
9. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement the welding design closed-loop feedback verification optimization method based on digital twins as described in any one of claims 1-6.
10. An electronic device, characterized in that, include: The device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the closed-loop feedback verification optimization method for welding design based on digital twins as described in any one of claims 1-6.