A Method and System for Real-Time Strain Monitoring and Closed-Loop Control of Digital Twin for Hot Gas Expansion Forming of General-Purpose Metal Pipes

Abstract

This invention relates to the field of metal pipe plastic forming, and discloses a general digital twin real-time strain monitoring and closed-loop control method and system for hot gas expansion forming of metal pipes. The method includes: real-time acquisition of process parameters and image data of the metal pipe during hot gas expansion forming, followed by time-series synchronization and normalization processing to generate a standardized real-time data source; inputting the standardized real-time data source into a pre-trained general AI real-time proxy model to predict the deformation state data of the pipe at the current and future times; using measured full-field strain data to perform online real-time calibration of the deformation state data predicted by the AI ​​real-time proxy model to obtain accurate twin data; based on the accurate twin data and a preset general forming limit criterion, real-time judgment of the pipe deformation safety state, and dynamic adjustment of the hot gas expansion forming process parameters according to the judgment result, thereby achieving closed-loop control of the forming process. This improves the process stability and yield of hot gas expansion forming of different metal pipes.

Description

Technical Field

[0001] This invention relates to the field of metal pipe plastic processing technology, and in particular to a digital twin real-time strain monitoring and closed-loop control method and system for hot gas expansion forming of general-purpose metal pipes. Background Technology

[0002] Thermoforming of metal tubes is a flexible plastic processing technology that uses internal gas pressure to drive the tube to be molded and formed. It has advantages such as high forming accuracy, good part integrity, and high forming efficiency, and is widely used in aerospace, automotive, rail transportation, and engineering machinery.

[0003] Existing hot gas expansion forming technology for pipes suffers from several common technical bottlenecks: First, the forming process takes place inside a closed mold, making it impossible to perceive the axial / circumferential strain, wall thickness reduction, and local deformation of the pipe in real time. Data can only be obtained through post-processing sectioning and sampling inspection, resulting in high inspection costs, significant data lag, and difficulty in accurately determining the forming limits. Second, process parameter control relies on offline simulation and human experience, leading to poor matching of parameters such as temperature, internal pressure, and axial material replenishment. This can easily cause defects such as wrinkling, cracking, excessive thinning, and insufficient forming, especially for different metal materials, resulting in long process debugging cycles and large fluctuations in yield. Third, traditional strain testing methods (such as etched meshes and strain gauges) cannot be adapted to multiple working conditions at high, medium, and normal temperatures. They are prone to failure at high temperatures and difficult to achieve real-time monitoring across the entire process at normal temperatures. Furthermore, existing technologies are mostly customized for single materials and single forming rates, resulting in extremely poor versatility. Fourth, there is a lack of virtual-real synchronous mapping and intelligent control methods for the forming process, making it impossible to achieve early warning of defects and real-time optimization of process parameters. Summary of the Invention

[0004] To address the aforementioned problems, the purpose of this invention is to provide a digital twin real-time strain monitoring and closed-loop control method and system for hot gas expansion forming of general-purpose metal pipes. This method overcomes the limitations of material type, forming temperature, and forming rate, and realizes non-contact real-time monitoring of the entire process of pipe deformation under all working conditions, virtual twin synchronous mapping, intelligent early warning of defects, and closed-loop control of process parameters, which greatly improves the process stability and yield of hot gas expansion forming of different metal pipes.

[0005] To achieve the above objectives, in a first aspect, the technical solution adopted by the present invention is as follows: a method for real-time strain monitoring and closed-loop control of digital twins in the hot gas expansion forming of general-purpose metal pipes, comprising: real-time acquisition of process parameters and image data of metal pipes during the hot gas expansion forming process; time-series synchronization and normalization processing of the acquired process parameters and image data to generate a standardized real-time data source; inputting the standardized real-time data source into a pre-trained general-purpose AI real-time agent model to predict the deformation state data of the pipe at the current moment and future moments; wherein, the general-purpose AI real-time agent model is pre-trained based on a multi-material general-purpose offline simulation database; online real-time calibration of the deformation state data predicted by the AI ​​real-time agent model is performed using measured full-field strain data to eliminate prediction bias and obtain accurate twin data; based on the accurate twin data and a preset general-purpose forming limit criterion, the deformation safety state of the pipe is judged in real time, and the process parameters of hot gas expansion forming are dynamically adjusted in real time according to the judgment result to achieve closed-loop control of the forming process.

[0006] In some embodiments, real-time acquisition of process parameters and image data of metal pipes during hot gas expansion forming is achieved through a multi-source data acquisition module installed on a general-purpose hot gas expansion forming equipment; wherein, the multi-source data acquisition module includes: The adjustable band DIC vision system, in conjunction with a detachable high-temperature resistant quartz observation window installed on a general hot gas expansion forming equipment, is used to acquire full-field image data of the pipe surface in real time. The DIC vision system is equipped with an adjustable filter and a water-cooled or air-cooled heat insulation device. It uses visible light band imaging under normal temperature conditions and ultraviolet band imaging under high temperature conditions. Temperature data of the forming mold, tube and cavity environment are collected in real time by multiple thermocouples arranged along the axial and circumferential directions of the forming mold in the general hot gas expansion forming equipment. The expansion pressure data inside the pipe is collected in real time by a gas pressure sensor; The axial feeding stroke data and feeding thrust data at both ends of the pipe are collected in real time using grating displacement sensors and pressure sensors.

[0007] In some embodiments, the method for constructing a multi-material universal offline simulation database is as follows: For various metallic materials, plastic constitutive models are established for each material under normal temperature, medium temperature and high temperature conditions to adapt to different forming rates; among them, the plastic constitutive models include elastoplastic constitutive, superplastic constitutive and viscoplastic constitutive models; Parametric modeling methods were used to establish hot gas tension models of pipes of different sizes. The properties of the hot gas tension models of pipes were assigned through plastic constitutive models, the pipe surface mesh was divided, and full-condition simulation parameters were set to calculate the deformation state data of pipes under different materials, temperatures and forming rates in batches, forming a generalized simulation database. The deformation state data includes strain field, wall thickness distribution, stress field and critical thresholds for wrinkling and cracking.

[0008] In some embodiments, a standardized real-time data source is input into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current and future times, including: The general AI real-time agent model adopts a hybrid neural network model of CNN-LSTM-Transformer; The AI ​​agent model was pre-trained using offline simulation data from a multi-material generalized simulation database as the training set and field test data as the calibration set. Using standardized real-time data sources as input, a pre-trained AI agent model predicts the deformation status data of the pipe at the current and future moments within milliseconds.

[0009] In some embodiments, online real-time calibration is performed on the deformation state data predicted by the AI ​​real-time proxy model using measured full-field strain data, including: An adaptive weighted fusion algorithm is adopted, using real-time collected full-field strain data of the pipe surface as measured benchmark data, to correct the deformation state data predicted by the AI ​​real-time agent model in real time.

[0010] In some embodiments, based on precise twin data and a preset universal forming limit criterion, the deformation safety status of the pipe is determined in real time, including: The full-field strain data in the precise twin data obtained after online calibration is compared and calculated in real time with the pre-stored forming limit curve corresponding to the current metal material and the current temperature to calculate the current deformation safety margin. Based on the deformation safety margin, a three-level early warning mechanism is set up: Level 1 warning: When the deformation safety margin is within the safe range, it is determined that there is no defect risk, and the current process parameters are maintained to operate normally; Level 2 warning: When the deformation safety margin is within the warning zone, it is determined that there is a defect risk, a warning signal is issued and the automatic fine-tuning of process parameters is triggered; Level 3 warning: When the deformation safety margin is within the danger zone, it is determined that a risk is about to occur, and pressure reduction or shutdown protection actions are immediately executed.

[0011] In some embodiments, the process parameters for hot gas expansion forming are dynamically adjusted in real time based on the judgment result, including: The MPC algorithm is adopted with defect-free and uniform forming as the control objective. Based on the real-time strain information and defect warning signals in the precise twin data, multiple process parameters are dynamically adjusted in real time to achieve closed-loop collaborative control of the forming process.

[0012] Secondly, the technical solution adopted by this invention is as follows: a digital twin real-time strain monitoring and closed-loop control system for hot gas expansion forming of general-purpose metal pipes, comprising: a physical layer, which collects process parameters and image data of metal pipes in real time during hot gas expansion forming; an edge data fusion layer, which performs time-series synchronization and normalization processing on the collected process parameters and image data to generate a standardized real-time data source; a digital twin core layer, which inputs the standardized real-time data source into a pre-trained general-purpose AI real-time agent model to predict the deformation state data of the pipe at the current moment and future moments; wherein, the general-purpose AI real-time agent model is pre-trained based on a multi-material general-purpose offline simulation database; the deformation state data predicted by the AI ​​real-time agent model is calibrated online in real time using measured full-field strain data to eliminate prediction bias and obtain accurate twin data; and an intelligent closed-loop control layer, which judges the deformation safety state of the pipe in real time based on the accurate twin data and a preset general-purpose forming limit criterion, and dynamically adjusts the process parameters of hot gas expansion forming in real time according to the judgment result to achieve closed-loop control of the forming process.

[0013] Thirdly, the technical solution adopted by the present invention is: a computer-readable storage medium for storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by a computing device, the computing device performs any of the methods described above.

[0014] Fourthly, the technical solution adopted by the present invention is: a computing device comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described above.

[0015] The present invention has the following advantages due to the adoption of the above technical solutions: 1. This invention is applicable to all metal pipes and all hot gas expansion forming conditions, eliminating the need to redesign the system for different materials and temperatures, thus significantly reducing equipment and process development costs.

[0016] 2. This invention provides highly efficient and accurate monitoring, enabling real-time non-contact strain monitoring throughout the entire process without damaging the sample. It improves detection efficiency by over 70% and significantly enhances the accuracy of strain and wall thickness testing.

[0017] 3. The process of this invention is intelligent and controllable, freeing it from reliance on manual experience, realizing automatic optimization of process parameters and early warning of defects, increasing the yield of various metal pipes by more than 15%, and shortening the process debugging cycle by 60%.

[0018] 4. The data in this invention is complete and traceable, recording forming data throughout the process, building a process knowledge base, realizing process optimization and iteration, and providing core support for the intelligent and digital transformation of pipe hot gas expansion forming. Attached Figure Description

[0019] Figure 1 This is an overall flowchart of the digital twin real-time strain monitoring and closed-loop control method for hot gas expansion forming of general metal pipes in this invention embodiment; Figure 2 This is a detailed flowchart of the digital twin real-time strain monitoring and closed-loop control method for hot gas expansion forming of general-purpose metal pipes in this invention. Detailed Implementation

[0020] 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0021] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0022] In one embodiment of the present invention, a method and system for real-time strain monitoring and closed-loop control of digital twins in the hot gas expansion forming of general-purpose metal pipes are provided. This method is applicable to various metal pipes, covering real-time strain monitoring, digital twin simulation, and process closed-loop control in hot gas expansion forming under all forming conditions (room temperature / medium temperature / high temperature / superplastic). It can be widely used in the irregular bulging processing of various pipes such as carbon steel, stainless steel, aluminum alloy, copper alloy, titanium alloy, and high-temperature alloy. In this embodiment, as... Figure 1 , Figure 2 As shown, the method includes the following steps: 1) Real-time acquisition of process parameters and image data of metal pipes during hot gas expansion forming process.

[0023] 2) Perform time-series synchronization and normalization processing on the collected process parameters and image data to generate a standardized real-time data source.

[0024] 3) Input standardized real-time data sources into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current time and future time; wherein, the general AI real-time agent model is pre-trained based on a multi-material general offline simulation database.

[0025] 4) The deformation state data predicted by the AI ​​real-time agent model is calibrated online in real time using the measured full-field strain data to eliminate prediction bias and obtain accurate twin data.

[0026] 5) Based on precise twin data and preset universal forming limit criteria, the deformation safety status of the pipe is judged in real time, and the process parameters of hot gas expansion forming are dynamically adjusted in real time according to the judgment results to achieve closed-loop control of the forming process.

[0027] In one possible implementation, step 1) above, which involves real-time acquisition of process parameters and image data of the metal tube during hot gas expansion forming, is achieved through a multi-source data acquisition module installed on a general-purpose hot gas expansion forming equipment. The process parameters include temperature, expansion pressure, axial feeding stroke data, and feeding thrust.

[0028] The multi-source data acquisition module includes a DIC (binocular digital image correlation) vision system, thermocouples, a gas pressure sensor, and a grating displacement sensor and pressure sensor. The real-time acquisition process is as follows: 1.1) An adjustable band DIC vision system is installed on the front and side of the pipe under test in the general hot gas forming equipment, in conjunction with a detachable high-temperature resistant quartz observation window installed on the general hot gas forming equipment, to acquire full-field image data of the pipe surface in real time; wherein, the DIC vision system is equipped with an adjustable filter and a water-cooled or air-cooled heat insulation device, and uses visible light band imaging under normal temperature conditions and ultraviolet band imaging under high temperature conditions to suppress thermal radiation interference.

[0029] In this embodiment, room temperature refers to room temperature, i.e., below 40°C; high temperature refers to above 400°C.

[0030] 1.2) Temperature data of the forming mold, pipe, and cavity environment are collected in real time using multiple thermocouples arranged along the axial and circumferential directions of the forming mold in the general hot gas expansion forming equipment. In this embodiment, the signal acquisition frequency is ≥50Hz.

[0031] 1.3) Real-time acquisition of expansion pressure data inside the pipe using a gas pressure sensor. In this embodiment, the gas pressure sensor is a high-precision gas pressure sensor with a range of 0~20MPa and a response time ≤10ms.

[0032] 1.4) A grating displacement sensor and a pressure sensor are used to collect axial feeding stroke data and feeding thrust data at both ends of the pipe in real time. In this embodiment, the acquisition frequency is ≥50Hz.

[0033] In this embodiment, the general hot gas expansion forming equipment includes a closed forming mold, a gas pressurization system, a temperature control system, and an axial feeding drive system; the forming mold has a pre-reserved mounting position for a detachable high-temperature resistant quartz observation window, which is suitable for imaging requirements at room temperature, medium temperature, and high temperature. The temperature control range covers room temperature to 1200℃, meeting the full-condition requirements of low-temperature plasticity, medium-temperature forming, and high-temperature superplasticity.

[0034] In one possible implementation, step 2) above involves time-series synchronization and normalization of the acquired process parameters and image data to generate a standardized real-time data source, specifically including the following steps: 2.1) Normalize temperature, pressure, displacement, and image data to remove abnormal data caused by environmental interference and signal drift.

[0035] 2.2) Establish a unified timestamp mechanism to ensure that all sensor data are fully aligned in time sequence and that the data transmission delay is ≤50ms.

[0036] 2.3) Perform data format conversion, caching, and high-speed transmission to provide a stable and accurate real-time data source for the core layer of the digital twin.

[0037] In one possible implementation, the method for constructing the multi-material universal offline simulation database in step 3) above includes the following steps: 3.1.1) For various metal materials, plastic constitutive models are established for each material under normal temperature, medium temperature and high temperature conditions to adapt to different forming rates; among them, the plastic constitutive models include elastoplastic constitutive, superplastic constitutive and viscoplastic constitutive models to be applicable to different forming rates (quasi-static, low speed, superplastic low speed).

[0038] Among them, various metal materials include carbon steel, stainless steel, aluminum alloy, copper alloy, titanium alloy, high-temperature alloy and other types of metals.

[0039] 3.1.2) Using parametric modeling methods, hot gas tension models of pipes with different dimensions (e.g., different diameters, lengths, and wall thicknesses) are established. Properties are assigned to these models using a plastic constitutive model, pipe surface meshes are generated, and full-condition simulation parameters are set to batch calculate pipe deformation state data under different materials, temperatures, and forming rates, forming a generalized simulation database. The deformation state data includes strain field, wall thickness distribution, stress field, and critical thresholds for wrinkling and cracking. The generalized simulation database supports rapid retrieval and replacement of material and process parameters without repetitive modeling, adapting to rapid process design for different metal pipes.

[0040] Optional full-condition simulation parameters include internal pressure, temperature, axial feeding, friction coefficient, etc.

[0041] In this embodiment, a standardized real-time data source is input into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current moment and future moments, including the following steps: 3.2.1) The general AI real-time agent model adopts a CNN-LSTM-Transformer hybrid neural network model, which integrates spatial feature extraction and temporal evolution prediction capabilities. Among them, the CNN layer is used to extract the spatial features of the input data, the LSTM layer is used to capture the dynamic evolution law of the time series, and the Transformer layer is used to model long-distance dependencies.

[0042] 3.2.2) Using offline simulation data from a multi-material generalized simulation database as the training set and field test data as the calibration set, the AI ​​agent model is pre-trained to achieve millisecond-level prediction of pipe deformation under any material and any working condition.

[0043] 3.2.3) Standardized real-time data sources are used as input data. The standardized real-time data stream includes real-time temperature, bulging pressure, axial feeding stroke, material type and forming rate. Through the pre-trained AI agent model, the deformation state data of the output pipe at the current moment and future moment are predicted in milliseconds.

[0044] The CNN-LSTM-Transformer hybrid neural network model used in this embodiment has a strain prediction error of ≤5%, a wall thickness prediction error of ≤3%, and a single inference time of ≤50ms, which can meet the requirements of real-time monitoring and control.

[0045] In this embodiment, an AI real-time agent model is used as a digital twin model, which breaks through the limitations of materials and forming rate, and realizes generalized modeling and real-time prediction.

[0046] In one possible implementation, step 4) above, which uses measured full-field strain data to perform online real-time calibration of the deformation state data predicted by the AI ​​real-time agent model, includes: using an adaptive weighted fusion algorithm to use the real-time collected full-field strain data of the pipe surface as measured benchmark data to perform real-time correction of the deformation state data predicted by the AI ​​real-time agent model.

[0047] In this embodiment, the adaptive weighted fusion algorithm includes the following steps: 4.1) Map the measured strain field on the pipe surface acquired in real time by the DIC vision system and the strain field predicted by the AI ​​real-time proxy model to a unified grid node or a unified axial-circumferential coordinate system; 4.2) Calculate the error between the measured strain and the predicted strain for each node or local area; 4.3) Determine adaptive fusion weights based on error, DIC measurement confidence, and historical prediction errors of the AI ​​model; 4.4) The AI-predicted strain field and the DIC-measured strain field are weighted and fused according to the adaptive fusion weight to obtain the corrected strain field, and the wall thickness field, thinning rate field and defect risk field are corrected according to the corrected strain field.

[0048] For the i-th node, the AI ​​model predicts the strain as follows: The measured strain of DIC is The error between the two is The fusion weights are calculated based on the error and measurement confidence level. And obtain the corrected strain:

[0049] in, The results change dynamically with the prediction error and the confidence level of the DIC measurement. When the DIC data is reliable and the prediction error is large, the correction results use more of the measured DIC data; when the DIC data is noisy, partially occluded, or missing, the correction results use more of the AI ​​prediction data.

[0050] In this embodiment, based on the full-field strain data measured by the DIC vision system, the prediction results of the AI ​​proxy model are corrected in real time to eliminate prediction deviations caused by material performance dispersion, temperature fluctuations, and changes in working conditions, ensuring accurate synchronization between the digital twin model and the physical pipe deformation, and long-term operation without drift.

[0051] In one possible implementation, step 5) above, which uses precise twin data and a preset universal forming limit criterion to determine the pipe deformation safety status in real time, includes the following steps: 5.1) The full-field strain data in the accurate twin data obtained after online calibration is compared and calculated in real time with the pre-stored forming limit curve (FLC) corresponding to the current metal material and the current temperature to calculate the current deformation safety margin.

[0052] In this embodiment, the specific form of the FLC failure criterion is as follows: ; in, A function representing the current deformation state of the model; It is the principal strain in the FLC criterion; It is the principal strain of the pipe; This is the secondary strain of the pipe. This condition gives the damage initiation criterion for FLC as follows: When damage begins to occur, the corresponding equivalent plastic strain is calculated and denoted as . Assuming the equivalent plastic strain corresponding to complete failure, then Assume that the equivalent effect at any time after the damage occurs becomes Then the safety margin factor can be calculated. : .

[0053] 5.2) Based on the deformation safety margin, a three-level early warning mechanism is set up to provide a universal forming limit criterion applicable to different metals and temperatures: Level 1 warning: When the deformation safety margin is within the safe zone, it is determined that there is no defect risk and the current process parameters are maintained to operate normally; where the safe zone range is (0-0.8).

[0054] Level 2 warning: When the deformation safety margin is within the warning zone, a defect risk is determined, a warning signal is issued, and automatic fine-tuning of process parameters is triggered; the warning zone range is (0.8-0.9).

[0055] Level 3 warning: When the deformation safety margin is within the danger zone (0.9-1.0), it is determined that a risk is about to occur, and pressure reduction or shutdown protection actions are immediately executed; the danger zone range is (0.9-1.0).

[0056] In this embodiment, a 1:1 digital twin model is constructed to synchronize the deformation process of the physical pipe in real time. The pipe strain and wall thickness distribution are displayed intuitively through cloud maps, and high-risk areas are highlighted.

[0057] In this embodiment, the process parameters of hot gas expansion forming are dynamically adjusted in real time based on the judgment results. This includes: employing an MPC (Model Predictive Control) algorithm, with defect-free and uniform forming as the control objective, and dynamically adjusting multiple process parameters in real time based on real-time strain information and defect warning signals from precise twin data, to achieve closed-loop collaborative control of the forming process and avoid wrinkling, cracking, and excessive thinning defects from the source. These multiple process parameters include expansion pressure, pressure increase rate, axial feeding speed, and temperature.

[0058] Specifically, the MPC dynamic adjustment method works as follows: using the strain field, wall thickness reduction rate, and defect risk predicted in real time by the digital twin model as state inputs, an objective function is established that includes minimizing the thinning rate, uniform strain distribution, and minimizing defect risk; within each control cycle, the forming state for several future time steps is predicted based on the current state, and the optimal control quantity is solved under the conditions of satisfying equipment limits, material forming limits, and process constraints, using pressure, pressurization rate, axial feeding speed, and temperature as control variables; then the optimal control quantity is sent to the gas pressurization system, axial feeding system, and temperature control system for execution, and the prediction model and control quantity are updated again at the next moment based on real-time monitoring data, thus forming a rolling optimization closed-loop control.

[0059] In one possible implementation, step 6) is included after step 5): process data management for automatically storing process parameters, strain data, early warning records and forming results throughout the forming process, building a process knowledge base, and providing parameter references for subsequent forming of similar pipes.

[0060] In one possible implementation, the strain calculation of the pipe in this embodiment adopts a general strain calculation method, namely, calculating the true circumferential strain and the true axial strain.

[0061] This embodiment establishes a universal strain calculation method applicable to the calculation of true strain in the thermal expansion of all metal pipes. Specifically, the true strain of the pipe during thermal expansion is calculated using plane strain, which is applicable to all metal materials. Zhou Xiang's true response: ; Axial true strain: .

[0062] In the formula, The initial speckle reference length; The length of the circumferential speckle pattern after deformation; The axial speckle length after deformation is given. The initial speckle reference length is the speckle spacing before deformation. The circumferential speckle length and the axial speckle length after deformation are measured by the DIC vision system.

[0063] In summary, the specific implementation process of this invention includes the following steps: (1) Preliminary preparation: Based on the characteristics of the metal material of the pipe, select the corresponding constitutive model, improve the offline generalized simulation database, and complete the pre-training of the AI ​​agent model; prepare speckle patterns on the surface of the pipe to suit the working conditions, such as: room temperature: matte paint speckle; medium and high temperature: ceramic-based high temperature speckle, to ensure imaging contrast.

[0064] (2) Sample clamping and equipment initialization: clamp the pipe to the forming mold, seal the mold, and initialize the sensing system, DIC vision system and digital twin model.

[0065] (3) Working conditions setting and pretreatment: Set the target forming temperature, initial expansion pressure, and axial feeding parameters, turn on the temperature control system, and maintain the temperature uniformly after reaching the set temperature.

[0066] (4) Real-time forming and monitoring: Start the gas pressurization and axial feeding device to start hot gas expansion forming; the edge data layer collects sensor data in real time, the digital twin core layer predicts and calibrates the deformation state of the pipe in real time, and the twin is visualized synchronously.

[0067] (5) Defect warning and closed-loop control: Real-time determination of deformation safety margin, automatic optimization of process parameters when entering the warning zone, and immediate execution of protection actions when entering the danger zone until forming is completed.

[0068] (6) Post-processing: After forming, the tube is cooled in the furnace / naturally, the whole process data is exported, and the process review and model optimization are completed.

[0069] In one embodiment of the present invention, a digital twin real-time strain monitoring and closed-loop control system for hot gas expansion forming of general-purpose metal tubing is provided, comprising: The physical layer collects process parameters and image data of metal pipes in real time during the hot gas expansion forming process; The edge data fusion layer performs time-series synchronization and normalization processing on the collected process parameters and image data to generate standardized real-time data sources; The core layer of the digital twin inputs standardized real-time data sources into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current and future times. The general AI real-time agent model is pre-trained based on a multi-material general offline simulation database. The deformation state data predicted by the AI ​​real-time agent model is calibrated online in real time using measured full-field strain data to eliminate prediction bias and obtain accurate twin data. The intelligent closed-loop control layer, based on precise twin data and preset universal forming limit criteria, judges the safety status of pipe deformation in real time, and dynamically adjusts the process parameters of hot gas expansion forming in real time according to the judgment results, so as to realize closed-loop control of the forming process.

[0070] In one possible implementation, real-time acquisition of process parameters and image data of metal tubing during hot gas forming is achieved through a multi-source data acquisition module installed on a general-purpose hot gas forming equipment; wherein the multi-source data acquisition module includes: The adjustable band DIC vision system, in conjunction with a detachable high-temperature resistant quartz observation window installed on a general hot gas expansion forming equipment, is used to acquire full-field image data of the pipe surface in real time. The DIC vision system is equipped with an adjustable filter and a water-cooled or air-cooled heat insulation device. It uses visible light band imaging under normal temperature conditions and ultraviolet band imaging under high temperature conditions. Temperature data of the forming mold, tube and cavity environment are collected in real time by multiple thermocouples arranged along the axial and circumferential directions of the forming mold in the general hot gas expansion forming equipment. The expansion pressure data inside the pipe is collected in real time by a gas pressure sensor; The axial feeding stroke data and feeding thrust data at both ends of the pipe are collected in real time using grating displacement sensors and pressure sensors.

[0071] In one possible implementation, the method for constructing a multi-material universal offline simulation database is as follows: For various metallic materials, plastic constitutive models are established for each material under normal temperature, medium temperature and high temperature conditions to adapt to different forming rates; among them, the plastic constitutive models include elastoplastic constitutive, superplastic constitutive and viscoplastic constitutive models; Parametric modeling methods were used to establish hot gas tension models of pipes of different sizes. The properties of the hot gas tension models of pipes were assigned through plastic constitutive models, the pipe surface mesh was divided, and full-condition simulation parameters were set to calculate the deformation state data of pipes under different materials, temperatures and forming rates in batches, forming a generalized simulation database. The deformation state data includes strain field, wall thickness distribution, stress field and critical thresholds for wrinkling and cracking.

[0072] In one possible implementation, a standardized real-time data source is input into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current and future moments, including: The general AI real-time agent model adopts a hybrid neural network model of CNN-LSTM-Transformer; The AI ​​agent model was pre-trained using offline simulation data from a multi-material generalized simulation database as the training set and field test data as the calibration set. Using standardized real-time data sources as input, a pre-trained AI agent model predicts the deformation status data of the pipe at the current and future moments within milliseconds.

[0073] In one possible implementation, the deformation state data predicted by the AI ​​real-time proxy model is calibrated online in real time using measured full-field strain data, including: An adaptive weighted fusion algorithm is adopted, using real-time collected full-field strain data of the pipe surface as measured benchmark data, to correct the deformation state data predicted by the AI ​​real-time agent model in real time.

[0074] In one possible implementation, based on precise twin data and a preset universal forming limit criterion, the deformation safety status of the pipe is determined in real time, including: The full-field strain data in the precise twin data obtained after online calibration is compared and calculated in real time with the pre-stored forming limit curve corresponding to the current metal material and the current temperature to calculate the current deformation safety margin. Based on the deformation safety margin, a three-level early warning mechanism is set up: Level 1 warning: When the deformation safety margin is within the safe range, it is determined that there is no defect risk, and the current process parameters are maintained to operate normally; Level 2 warning: When the deformation safety margin is within the warning zone, it is determined that there is a defect risk, a warning signal is issued and the automatic fine-tuning of process parameters is triggered; Level 3 warning: When the deformation safety margin is within the danger zone, it is determined that a risk is about to occur, and pressure reduction or shutdown protection actions are immediately executed.

[0075] In one possible implementation, the process parameters for hot gas expansion forming are dynamically adjusted in real time based on the judgment result, including: The MPC algorithm is adopted with defect-free and uniform forming as the control objective. Based on the real-time strain information and defect warning signals in the precise twin data, multiple process parameters are dynamically adjusted in real time to achieve closed-loop collaborative control of the forming process.

[0076] The system provided in this embodiment is used to execute the above-described method embodiments. For specific processes and details, please refer to the above embodiments, which will not be repeated here.

[0077] In summary, this invention possesses universality across all materials and operating conditions, breaking through the limitations of single materials and single forming rates. It covers all metal pipes, including carbon steel, aluminum alloys, and titanium alloys, and is adaptable to the entire temperature range (room temperature, medium temperature, high temperature, and superplastic) and any forming rate. A single solution is suitable for all hot gas expansion forming conditions. Furthermore, this invention enables real-time strain monitoring across the entire domain. It employs an adjustable-band DIC vision system with a detachable quartz observation window, resolving compatibility issues between room temperature and high temperature strain testing. This allows for non-contact, real-time, and continuous monitoring of the pipe's strain within a closed mold, replacing traditional post-construction damage detection. Furthermore, this invention utilizes universal digital twin modeling to construct a multi-material constitutive universal simulation database and a hybrid neural network AI proxy model. This eliminates the need for remodeling for each material, enabling rapid adaptation to different pipe forming processes and achieving millisecond-level deformation prediction. Based on real-time strain data and universal forming limit criteria, it achieves collaborative closed-loop control of multiple parameters, including temperature, pressure, and axial feeding, proactively preventing various forming defects, significantly improving process stability and yield, and realizing intelligent closed-loop collaborative control. Furthermore, this invention enables online self-calibration and optimization, using DIC measured data to correct the twin model in real time, adapting to material and operating condition fluctuations, ensuring long-term monitoring and control accuracy, and eliminating the need for frequent manual model calibration.

[0078] In one embodiment of the present invention, a computing device is also provided. This computing device can be a terminal, and may include: a processor, a communication interface, memory, a display screen, and an input device. The processor, communication interface, and memory communicate with each other via a communication bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. When the computer programs are executed by the processor, they implement the methods described in the above embodiments. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, a network management system, NFC (Near Field Communication), or other technologies. The display screen can be a liquid crystal display or an e-ink display. The input device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad mounted on the casing of the computing device, or an external keyboard, touchpad, or mouse. The processor can call logical instructions in the memory to execute the methods described in the above embodiments.

[0079] In one embodiment of the present invention, a computer program product including instructions is also provided. This computer program product may be a software or program product including instructions and capable of running on a computing device or stored on any available medium. When the computer program product is run on at least one computing device, it causes the at least one computing device to perform the methods and functions of any of the foregoing embodiments.

[0080] In one embodiment of the invention, a computer-readable storage medium is also provided. The computer-readable storage medium can be any available medium accessible to a computing device, or a data storage device (e.g., a data center) that includes one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, or magnetic tapes), optical media (e.g., DVDs), semiconductor media (e.g., solid-state drives), etc. The computer-readable storage medium includes instructions, and the instructions instruct the computing device to perform the methods and functions in any of the foregoing embodiments.

[0081] Computer-readable media can be any tangible medium that includes or stores a program for executing instructions on or associated with a system, apparatus, or device, or a data storage device, such as a data center, including one or more available media. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable media can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, any suitable combination thereof. More detailed examples of computer-readable storage media include electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0082] Furthermore, although the operations of the methods disclosed in this invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations need to be performed in a specific order, or that all the operations shown need to be performed to achieve the desired result. Rather, the order of execution of the operations shown in the flowcharts can be changed. Additionally or alternatively, some operations may be omitted, multiple operations may be combined into one operation for execution, and / or one operation may be decomposed into multiple operations for execution. It should also be noted that the features and functions of two or more devices disclosed in this invention may be specific in one device. Conversely, the features and functions of the aforementioned one device may be further specific in multiple devices.

[0083] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for real-time strain monitoring and closed-loop control of general-purpose metal pipe hot gas expansion forming using digital twins, characterized in that, include: Real-time acquisition of process parameters and image data of metal pipes during hot gas expansion forming; The collected process parameters and image data are synchronized and normalized in time to generate a standardized real-time data source; Standardized real-time data sources are input into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current and future times; the general AI real-time agent model is pre-trained based on a multi-material general offline simulation database. The deformation state data predicted by the AI ​​real-time agent model is calibrated online in real time using measured full-field strain data to eliminate prediction bias and obtain accurate twin data. Based on precise twin data and preset universal forming limit criteria, the system can determine the safety status of pipe deformation in real time, and dynamically adjust the process parameters of hot gas expansion forming in real time according to the judgment results, so as to achieve closed-loop control of the forming process.

2. The method for real-time strain monitoring and closed-loop control of digital twin in hot gas expansion forming of general-purpose metal pipes as described in claim 1, characterized in that, Real-time acquisition of process parameters and image data of metal pipes during hot gas expansion forming is achieved through a multi-source data acquisition module installed on a general-purpose hot gas expansion forming equipment. This multi-source data acquisition module includes: The adjustable band DIC vision system, in conjunction with a detachable high-temperature resistant quartz observation window installed on a general hot gas expansion forming equipment, is used to acquire full-field image data of the pipe surface in real time. The DIC vision system is equipped with an adjustable filter and a water-cooled or air-cooled heat insulation device. It uses visible light band imaging under normal temperature conditions and ultraviolet band imaging under high temperature conditions. Temperature data of the forming mold, tube and cavity environment are collected in real time by multiple thermocouples arranged along the axial and circumferential directions of the forming mold in the general hot gas expansion forming equipment. The expansion pressure data inside the pipe is collected in real time by a gas pressure sensor; The axial feeding stroke data and feeding thrust data at both ends of the pipe are collected in real time using grating displacement sensors and pressure sensors.

3. The method for real-time strain monitoring and closed-loop control of digital twin in hot gas expansion forming of general-purpose metal pipes as described in claim 1, characterized in that, The method for constructing a multi-material universal offline simulation database is as follows: For various metallic materials, plastic constitutive models are established for each material under normal temperature, medium temperature and high temperature conditions to adapt to different forming rates; among them, the plastic constitutive models include elastoplastic constitutive, superplastic constitutive and viscoplastic constitutive models; Parametric modeling methods were used to establish hot gas tension models of pipes of different sizes. The properties of the hot gas tension models of pipes were assigned through plastic constitutive models, the pipe surface mesh was divided, and full-condition simulation parameters were set to calculate the deformation state data of pipes under different materials, temperatures and forming rates in batches, forming a generalized simulation database. The deformation state data includes strain field, wall thickness distribution, stress field and critical thresholds for wrinkling and cracking.

4. The method for real-time strain monitoring and closed-loop control of digital twin in hot gas expansion forming of general-purpose metal pipes as described in claim 1, characterized in that, Standardized real-time data sources are input into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current and future times, including: The general AI real-time agent model adopts a hybrid neural network model of CNN-LSTM-Transformer; The AI ​​agent model was pre-trained using offline simulation data from a multi-material generalized simulation database as the training set and field test data as the calibration set. Using standardized real-time data sources as input, a pre-trained AI agent model predicts the deformation status data of the pipe at the current and future moments within milliseconds.

5. The method for real-time strain monitoring and closed-loop control of digital twins for hot gas expansion forming of general-purpose metal pipes as described in claim 1, characterized in that, Online real-time calibration of the deformation state data predicted by the AI ​​real-time proxy model is performed using measured full-field strain data, including: An adaptive weighted fusion algorithm is adopted, using real-time collected full-field strain data of the pipe surface as measured benchmark data, to correct the deformation state data predicted by the AI ​​real-time agent model in real time.

6. The method for real-time strain monitoring and closed-loop control of digital twin in hot gas expansion forming of general-purpose metal pipes as described in claim 1, characterized in that, Based on precise twin data and preset universal forming limit criteria, the deformation safety status of the pipe is determined in real time, including: The full-field strain data in the precise twin data obtained after online calibration is compared and calculated in real time with the pre-stored forming limit curve corresponding to the current metal material and the current temperature to calculate the current deformation safety margin. Based on the deformation safety margin, a three-level early warning mechanism is set up: Level 1 warning: When the deformation safety margin is within the safe range, it is determined that there is no defect risk, and the current process parameters are maintained to operate normally; Level 2 warning: When the deformation safety margin is within the warning zone, it is determined that there is a defect risk, a warning signal is issued and the automatic fine-tuning of process parameters is triggered; Level 3 warning: When the deformation safety margin is within the danger zone, it is determined that a risk is about to occur, and pressure reduction or shutdown protection actions are immediately executed.

7. The method for real-time strain monitoring and closed-loop control of digital twin in hot gas expansion forming of general-purpose metal pipes as described in claim 1, characterized in that, The process parameters for hot gas expansion forming are dynamically adjusted in real time based on the judgment results, including: The MPC algorithm is adopted with defect-free and uniform forming as the control objective. Based on the real-time strain information and defect warning signals in the precise twin data, multiple process parameters are dynamically adjusted in real time to achieve closed-loop collaborative control of the forming process.

8. A digital twin real-time strain monitoring and closed-loop control system for hot gas expansion forming of general-purpose metal pipes, characterized in that, include: The physical layer collects process parameters and image data of metal pipes in real time during the hot gas expansion forming process; The edge data fusion layer performs time-series synchronization and normalization processing on the collected process parameters and image data to generate standardized real-time data sources; The core layer of the digital twin inputs standardized real-time data sources into a pre-trained general AI real-time agent model to predict the deformation state data of the pipe at the current and future times. The general AI real-time agent model is pre-trained based on a multi-material general offline simulation database. The deformation state data predicted by the AI ​​real-time agent model is calibrated online in real time using measured full-field strain data to eliminate prediction bias and obtain accurate twin data. The intelligent closed-loop control layer, based on precise twin data and preset universal forming limit criteria, judges the safety status of pipe deformation in real time, and dynamically adjusts the process parameters of hot gas expansion forming in real time according to the judgment results, so as to realize closed-loop control of the forming process.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 7.

Images