Welding damage repair precision control method and system based on digital twinning-edge calculation

Abstract

The invention belongs to the technical field of damage data processing, and particularly relates to a digital twinning-edge calculation welding damage repair precision control method and system.The control method comprises the steps that a multispectral scanning technology is adopted to obtain surface topography data and internal structure information of an aviation aluminum alloy component; obtaining damage degree grading data, establishing an energy input demand mapping relation according to the damage degree grading data, and obtaining energy input parameter combination data according to the energy input demand mapping relation; performing regional energy irradiation processing according to the energy input parameter combination data to obtain a microstructure reconstruction prediction result; adjusting an energy input strategy according to a microstructure reconstruction prediction result, and determining a fatigue damage repair effect evaluation index; and an energy input parameter database is optimized through feedback of fatigue damage repair effect evaluation indexes, a corresponding relation model of damage types and optimal energy input parameters is established, and a closed-loop control system of aviation aluminum alloy multi-scale fatigue damage collaborative repair is formed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of damage data processing, and particularly relates to a welding damage repair precision control method and system based on digital twinning and edge computing. BACKGROUND

[0002] The research on the fatigue performance of materials in the field of aerospace is crucial, and is directly related to the safety and service life of components. In particular, for high-strength and complex load environments, the fatigue damage repair technology of aluminum alloy materials has decisive significance for prolonging the service life of components and ensuring flight safety. Currently, traditional fatigue damage repair methods rely on mechanical processing or chemical treatment. Although these methods have certain effects on surface repair, they often fail to restore the microstructure of the material inside, and the stability of the performance after repair is insufficient. In particular, under multi-axial complex load, the repair effect is difficult to predict, which can easily lead to accidental failure of components in service. In addition, traditional methods lack systematic research on the relationship between energy input and damage healing effect when dealing with different damage levels, which limits the universality of the repair technology.

[0003] The core challenge of the research is how to accurately control the repair process of fatigue damage through energy input (such as electron beam or excimer laser irradiation), and establish a quantitative relationship between damage healing and process parameters. Under complex load, aerospace aluminum alloys will produce multi-scale damage, from macroscopic cracks to microscopic grain boundary defects. These damages have significantly different responses to energy input during the repair process. For example, high energy input can repair macroscopic cracks, but can cause local overheating, change the microstructure of the material, and thus affect the hardness or residual stress distribution. On the other hand, low energy input can avoid overheating, but may not effectively repair deep microscopic damage. This contradiction between energy input and multi-scale damage repair effect makes it difficult to optimize the repair process, and cannot meet the needs of both surface performance restoration and internal mechanical performance stability.

[0004] Therefore, how to accurately regulate energy input to achieve the coordinated repair of multi-scale fatigue damage of aerospace aluminum alloys, and reveal the quantitative relationship between energy input, microstructure evolution and macroscopic performance restoration, has become a key problem of this research. SUMMARY

[0005] To solve the above technical problems, the application provides a welding damage repair precision control method and system based on digital twinning and edge computing, which can achieve the coordinated repair of multi-scale fatigue damage of aerospace aluminum alloys.

[0006] To achieve the above purpose, the application provides a welding damage repair precision control method based on digital twinning and edge computing, which comprises:

[0007] The multispectral scanning technology is used to acquire surface topography data and internal structure information of the aviation aluminum alloy component;

[0008] According to the surface topography data and the internal structure information, damage degree grading data is acquired;

[0009] According to the damage degree grading data, an energy input requirement mapping relationship is established, and energy input parameter combination data is acquired according to the energy input requirement mapping relationship;

[0010] According to the energy input parameter combination data, a partition energy irradiation treatment is performed, and a microstructure reconstruction prediction result is obtained;

[0011] According to the microstructure reconstruction prediction result, an energy input strategy is adjusted, accurate irradiation repair is performed by using the adjusted energy input control instruction, and a fatigue damage repair effect evaluation index is determined;

[0012] Through the fatigue damage repair effect evaluation index, an energy input parameter database is optimized, a corresponding relationship model of damage type and optimal energy input parameter is established, and a closed-loop control system of aviation aluminum alloy multi-scale fatigue damage collaborative repair is formed.

[0013] Optionally, according to the surface topography data and the internal structure information, the damage degree grading data includes:

[0014] The surface topography data and the internal structure information are stored as three-dimensional point cloud data and stereoscopic microscopic data;

[0015] The three-dimensional point cloud data and the stereoscopic microscopic data are processed by using a preset multi-scale fatigue damage detection algorithm, macro crack length and depth distribution characteristics are extracted, and crack geometric parameters are obtained;

[0016] The stereoscopic microscopic data is analyzed by using the multi-scale fatigue damage detection algorithm, micro grain boundary defects are identified, defect density distribution is calculated, and micro defect parameters are obtained;

[0017] According to the crack geometric parameters and the micro defect parameters, the damage degree grading data is determined.

[0018] Optionally, according to the crack geometric parameters and the micro defect parameters, the damage degree grading data includes:

[0019] The crack severity in the crack geometric parameters is classified by using a support vector machine algorithm, and preliminary damage grading is obtained;

[0020] According to the defect density distribution in the micro defect parameters, a clustering algorithm is used to group the grain boundary defects, and defect density grading is obtained;

[0021] If the initial damage classification is inconsistent with the defect density classification, a weighted fusion algorithm is used to integrate the crack geometric parameters and the microscopic defect parameters to determine the final damage degree classification data.

[0022] Optionally, an energy input requirement mapping relationship is established according to the damage degree classification data, and the energy input parameter combination data is obtained according to the energy input requirement mapping relationship, which includes:

[0023] The macroscopic crack depth and the microscopic defect density are obtained from the damage degree classification data, and the damage feature distribution of each region is determined;

[0024] If the macroscopic crack depth exceeds a preset threshold, a high energy density parameter is used to obtain a high energy input requirement;

[0025] If the microscopic defect density is lower than a critical value, a low energy density parameter is used to obtain a low energy input requirement;

[0026] According to the damage feature distribution and the energy input requirement, an energy input mapping relationship of each region is established to determine a regional energy distribution scheme;

[0027] The energy input mapping relationship is optimized by a genetic algorithm to calculate the laser power density and irradiation time combination of each region to obtain a preliminary parameter set;

[0028] The laser power density and irradiation time combination are obtained from the preliminary parameter set to obtain the energy input parameter combination data.

[0029] Optionally, the partitioned energy irradiation processing is performed according to the energy input parameter combination data to obtain a microstructure reconstruction prediction result, which includes:

[0030] After obtaining the energy input parameter combination, the partitioned energy irradiation processing is started, and the temperature change curve and the molten pool formation state of each region are monitored;

[0031] The microstructure prediction algorithm is driven by the temperature change curve and the molten pool state data to calculate the grain size change trend, the phase transition temperature interval and the cooling rate distribution, to judge the local overheating risk degree and the organization evolution path, and to obtain the microstructure reconstruction prediction result.

[0032] Optionally, after obtaining the energy input parameter combination, the partitioned energy irradiation processing is started, and the temperature change curve and the molten pool formation state of each region are monitored, which includes:

[0033] Based on the energy input parameter combination, a preset classification algorithm is used to determine the distribution of the macroscopic crack region and the microscopic defect region to obtain an initial configuration of the partitioned energy irradiation;

[0034] According to the initial configuration of the partition energy irradiation, high energy input is applied to the macroscopic crack area by using a pulsed laser mode, and low energy input is applied to the microscopic defect area by using a continuous laser mode, to obtain a laser treatment state of each area;

[0035] The laser treatment state of each area is collected by the real-time monitoring system to obtain temperature change curves and molten pool formation state data, and real-time feedback information is obtained;

[0036] The updated partition energy irradiation state is collected through real-time feedback information to obtain the temperature change curves and molten pool formation state.

[0037] Optionally, obtaining the microstructure reconstruction prediction result comprises:

[0038] From the temperature change curves and molten pool formation state, time sequence features are obtained, and a fast Fourier transform method is used to extract periodic change patterns to obtain temperature fluctuation features;

[0039] According to the temperature fluctuation features, a finite element analysis method is used to simulate the heat flow distribution inside the molten pool, and the grain size change trend is calculated;

[0040] From the grain size change trend, the temperature gradient of the key time node is extracted, and a preset phase change temperature database is combined to determine the phase change temperature interval;

[0041] Through the phase change temperature interval, a Monte Carlo method is used to simulate the random heat conduction behavior in the cooling process to obtain a cooling rate distribution, and if the cooling rate distribution exceeds a preset threshold, a convolutional neural network is used to analyze the local overheating risk to determine the overheating area range;

[0042] According to the overheating area range and the cooling rate distribution, a preset organization evolution model is combined to calculate the organization evolution path;

[0043] Through the organization evolution path, a stereology method is used to reconstruct a three-dimensional structure of the microstructure to obtain the microstructure reconstruction prediction result.

[0044] Optionally, according to the microstructure reconstruction prediction result, the energy input strategy is adjusted, and precise irradiation repair is performed by using the adjusted energy input control instruction to determine the fatigue damage repair effect evaluation index, which comprises:

[0045] According to the microstructure reconstruction prediction result, the subsequent energy input strategy is adjusted, if a local overheating risk is predicted, the laser power density is reduced and the irradiation time is prolonged, if the organization evolution is insufficient, the energy density is increased and the irradiation interval is shortened, and a corrected energy input control instruction is obtained;

[0046] The precise irradiation repair is performed by using the modified energy input control instruction, the surface roughness change, internal stress distribution and grain boundary healing degree data of the repair area are collected in real time, the fatigue damage repair effect evaluation index is determined by comparing the change amount of the damage characteristic parameters before and after the repair.

[0047] Optionally, the fatigue damage repair effect evaluation index comprises:

[0048] The energy input control instruction parameters are calculated by using the finite element analysis algorithm, and the energy input sequence of the irradiation repair is obtained;

[0049] The irradiation repair is performed according to the energy input sequence, the target area is processed by using a laser irradiation device, and real-time processing data of the repair area are obtained;

[0050] The surface roughness data of the repair area are collected by using an optical microscope, the roughness parameters are extracted by using an image processing algorithm, and the surface roughness change amount is obtained;

[0051] The internal stress distribution data of the repair area are collected by using an X-ray diffractometer, the stress distribution characteristics are analyzed by using a data fitting algorithm, and the internal stress change amount is obtained;

[0052] The grain boundary healing data are collected by using an electron backscatter diffraction technology, the grain boundary healing degree is quantified by using a statistical analysis method, and the grain boundary healing change amount is obtained;

[0053] If the surface roughness change amount, the internal stress change amount and the grain boundary healing change amount all satisfy the preset threshold value, the change amount data are fused by using a weighted average algorithm, and the fatigue damage repair effect evaluation index is determined;

[0054] If any one of the surface roughness change amount, the internal stress change amount and the grain boundary healing change amount does not satisfy the preset threshold value, the parameters of the energy input model are adjusted, the control instruction is regenerated, the irradiation repair and data collection are repeatedly performed, and a new fatigue damage repair effect evaluation index is obtained.

[0055] The application further provides a welding damage repair precision control system based on digital twinning and edge computing, comprising a data collection module, a data processing module, an index determination module and a control module.

[0056] The data collection module is used for acquiring surface topography data and internal structure information of an aviation aluminum alloy component by using a multispectral scanning technology.

[0057] The data processing module is configured to acquire damage degree grading data according to the surface topography data and internal structure information, establish an energy input demand mapping relationship according to the damage degree grading data, acquire energy input parameter combination data according to the energy input demand mapping relationship, and perform partition energy irradiation processing according to the energy input parameter combination data to obtain a microstructure reconstruction prediction result.

[0058] The index determination module is configured to adjust an energy input strategy according to the microstructure reconstruction prediction result, perform accurate irradiation repair by using the adjusted energy input control instruction, and determine a fatigue damage repair effect evaluation index.

[0059] The control module is configured to optimize an energy input parameter database through the fatigue damage repair effect evaluation index, establish a corresponding relationship model of damage types and optimal energy input parameters, and form a closed-loop control system for aviation aluminum alloy multi-scale fatigue damage collaborative repair.

[0060] Compared with the prior art, the present application has the following advantages and technical effects:

[0061] The present application solves the problem that the traditional repair method is difficult to simultaneously consider the accurate repair of macroscopic cracks and microscopic defects. By acquiring the surface topography and internal structure data of the component through multi-spectral scanning, the crack and defect characteristics are analyzed by using a multi-scale fatigue damage detection algorithm, and a mapping relationship between the damage degree and the energy input demand is established. According to the damage characteristics, the present application adaptively adjusts the laser power density and irradiation time, uses pulsed laser to repair macroscopic cracks and continuous laser to repair microscopic defects, simultaneously drives the microstructure prediction through temperature and molten pool state monitoring, dynamically adjusts the energy input strategy to avoid overheating or insufficient repair. The present application optimizes the energy input parameter database by real-time collection of roughness, stress and grain boundary healing data in the repair area, establishes a corresponding model of damage types and optimal parameters, and forms a closed-loop control system. The present application significantly improves the precision and efficiency of aviation aluminum alloy component fatigue damage repair, prolongs the service life of the component, and provides an efficient and reliable repair technology for aviation manufacturing. BRIEF DESCRIPTION OF DRAWINGS

[0062] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The embodiments of the present application illustrated in the drawings, and their description, are used to explain the present application and are not intended to limit the present application. In the drawings:

[0063] Figure 1 is a digital twin-edge computing welding damage repair precision control method flowchart of an embodiment of the present application. DETAILED DESCRIPTION

[0064] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0065] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in an order different from that shown herein.

[0066] Embodiment one:

[0067] The present embodiment proposes a digital twin-edge computing welding damage repair precision control method, as shown in Figure 1 The specific steps include the following steps:

[0068] The multi-spectral scanning technology is used to obtain the surface topography data and internal structure information of the aluminum alloy component;

[0069] According to the surface topography data and internal structure information, the damage degree classification data is obtained;

[0070] According to the damage degree classification data, the energy input requirement mapping relationship is established, and according to the energy input requirement mapping relationship, the energy input parameter combination data is obtained;

[0071] According to the energy input parameter combination data, the partition energy irradiation treatment is carried out, and the microstructure reconstruction prediction result is obtained;

[0072] According to the microstructure reconstruction prediction result, the energy input strategy is adjusted, the accurate irradiation repair is carried out by using the adjusted energy input control instruction, and the fatigue damage repair effect evaluation index is determined;

[0073] Through the fatigue damage repair effect evaluation index, the energy input parameter database is optimized, the corresponding relationship model of damage type and optimal energy input parameter is established, and the closed-loop control system of aviation aluminum alloy multi-scale fatigue damage collaborative repair is formed.

[0074] Further, according to the surface topography data and internal structure information, the damage degree classification data includes:

[0075] The surface topography data and internal structure information are stored as three-dimensional point cloud data and stereoscopic microscopic data;

[0076] The preset multi-scale fatigue damage detection algorithm is used to process the three-dimensional point cloud data and stereoscopic microscopic data, the macroscopic crack length and depth distribution characteristics are extracted, and the crack geometric parameters are obtained;

[0077] The stereomicroscopy data is analyzed by a multi-scale fatigue damage detection algorithm to identify micro-crystal boundary defects, calculate defect density distribution, and obtain micro-defect parameters;

[0078] According to the crack geometric parameters and the micro-defect parameters, damage degree classification data is determined.

[0079] Further, determining the damage degree classification data according to the crack geometric parameters and the micro-defect parameters comprises:

[0080] The crack severity in the crack geometric parameters is classified by using a support vector machine algorithm to obtain preliminary damage classification;

[0081] According to the defect density distribution in the micro-defect parameters, a clustering algorithm is used to group the crystal boundary defects to obtain defect density classification;

[0082] If the preliminary damage classification and the defect density classification are inconsistent, the crack geometric parameters and the micro-defect parameters are integrated by using a weighted fusion algorithm to determine final damage degree classification data.

[0083] Specifically, in the laser processing parameter optimization of an aviation aluminum alloy component, the acquisition of damage degree classification data is a core step. The determination of macro-crack depth and micro-defect density needs to be based on three-dimensional point cloud data and stereomicroscopy data generated by multi-spectral scanning. The crack depth distribution is analyzed through the point cloud data. It is assumed that the crack depth in a certain area reaches 2.5 millimeters, which exceeds the preset threshold of 1.8 millimeters, indicating that the damage in this area is serious. The micro-defect density is calculated through the stereomicroscopy data. It is assumed that the defect density in a certain area is 0.03 per cubic millimeter, which is lower than the critical value of 0.05 per cubic millimeter, indicating that the micro-damage is relatively light. Such classification data provides a basis for subsequent energy allocation. Specifically, when establishing the energy input mapping relationship, high energy density parameters are suitable for areas with excessive crack depth.

[0084] Further, according to the damage degree classification data, an energy input requirement mapping relationship is established, and according to the energy input requirement mapping relationship, energy input parameter combination data is acquired.

[0085] The macro-crack depth and the micro-defect density are obtained from the damage degree classification data to determine the damage feature distribution of each area;

[0086] If the macro-crack depth exceeds the preset threshold, high energy density parameters are used to obtain high energy input requirements;

[0087] If the micro-defect density is lower than the critical value, low energy density parameters are used to obtain low energy input requirements;

[0088] According to the damage feature distribution and the energy input requirements, an energy input mapping relationship for each area is established to determine a regional energy allocation scheme.

[0089] The energy input mapping relationship is optimized by a genetic algorithm to calculate the laser power density and irradiation time combination of each region, and a preliminary parameter set is obtained.

[0090] The laser power density and irradiation time combination is obtained from the preliminary parameter set, and the energy input parameter combination data is obtained.

[0091] Specifically, in the laser repair process, when obtaining the energy input parameter combination, the initial parameters can be determined through experimental data and historical database. For example, for the macro crack region, the laser power density can be set to 500 W / cm 2 , and the irradiation time is 0.1s; the micro defect region is set to 200 W / cm 2 , and the irradiation time is 0.3s. These parameters are selected based on material type (such as aluminum alloy) and damage degree (such as crack depth 0.5mm) to ensure that the energy input is adapted to the damage characteristics.

[0092] It should be noted that the parameter combination needs to consider the thermal conductivity and melting point of the material to avoid overheating or deficiency. In the initial configuration of partition energy irradiation, the K-means clustering algorithm can be used to classify the damage region. The region with crack depth greater than 0.3mm is classified as macro crack region, and the rest is classified as micro defect region. The initial configuration can set the macro crack region to use pulsed laser with frequency of 100Hz and duty cycle of 50% to concentrate energy to repair deep damage; the micro defect region uses continuous laser with constant power to ensure uniform repair of shallow defects. Specifically, the real-time monitoring system can use an infrared thermometer and a high-speed camera to collect temperature changes and molten pool state.

[0093] Further, according to the energy input parameter combination data, the partition energy irradiation processing is started, and the microstructure reconstruction prediction result is obtained, including:

[0094] After obtaining the energy input parameter combination, start the partition energy irradiation processing, and monitor the temperature change curve and molten pool formation state of each region;

[0095] The microstructure prediction algorithm is driven by the temperature change curve and molten pool state data to calculate the grain size change trend, phase transition temperature interval and cooling rate distribution, judge the local overheating risk degree and organization evolution path, and obtain the microstructure reconstruction prediction result.

[0096] Further, after obtaining the energy input parameter combination, start the partition energy irradiation processing, and monitor the temperature change curve and molten pool formation state of each region, including:

[0097] Based on the energy input parameter combination, the distribution of macro crack region and micro defect region is determined by a preset classification algorithm, and the initial configuration of partition energy irradiation is obtained;

[0098] According to the initial configuration of the partition energy irradiation, high energy input is applied to the macroscopic crack area by using the pulse laser mode, and low energy input is applied to the microscopic defect area by using the continuous laser mode, to obtain the laser treatment state of each area;

[0099] The laser treatment state of each area is collected by the real-time monitoring system to obtain the temperature change curve and the data of the molten pool formation state, and real-time feedback information is obtained;

[0100] The updated partition energy irradiation state is collected through the real-time feedback information to obtain the temperature change curve and the molten pool formation state.

[0101] Further, the microstructure reconstruction prediction result includes:

[0102] The time sequence feature is obtained from the temperature change curve and the molten pool formation state, the periodic change mode is extracted by using the fast Fourier transform method, and the temperature fluctuation feature is obtained;

[0103] According to the temperature fluctuation feature, the heat flow distribution inside the molten pool is simulated by using the finite element analysis method, and the grain size change trend is calculated;

[0104] The temperature gradient of the key time node is extracted from the grain size change trend, and the phase change temperature interval is determined by combining the preset phase change temperature database;

[0105] Through the phase change temperature interval, the random heat conduction behavior in the cooling process is simulated by using the Monte Carlo method to obtain the cooling rate distribution, if the cooling rate distribution exceeds the preset threshold value, then the local overheating risk is analyzed by using the convolutional neural network to determine the overheating area range;

[0106] According to the overheating area range and the cooling rate distribution, the microstructure evolution path is calculated by combining the preset microstructure evolution model;

[0107] Through the microstructure evolution path, the three-dimensional structure of the microstructure is reconstructed by using the stereology method to obtain the microstructure reconstruction prediction result.

[0108] Specifically, when extracting time series features from the temperature change curve and the molten pool state data, periodic changes can be identified by the fast Fourier transform method. When processing the metal surface with a laser, the temperature curve may exhibit periodic fluctuations, which are related to the laser pulse frequency. Assuming that the laser pulse frequency is 100 Hz, the dominant frequency can be extracted by fast Fourier transform, resulting in a feature of a temperature fluctuation period of 0.01 seconds. This periodic feature helps to judge the stability of the laser energy input and provides a basis for subsequent heat flow simulation. When performing finite element analysis based on the temperature fluctuation feature, the heat flow distribution inside the molten pool can be simulated. Assuming that the center temperature of the molten pool is 1500°C and the boundary temperature is 800°C, the finite element analysis can divide 1000 grid elements to simulate the heat flow transmission trend from the center to the edge. The results show that the heat flow density is higher near the center of the molten pool, which may lead to rapid grain size increase. By analyzing the grain size change trend, it can be found that the grain size increases from the initial 10 microns to 50 microns, reflecting the influence of heat input on the microstructure. In one possible implementation, when extracting the temperature gradient of the key time node from the grain size change trend, the phase change temperature database can be combined. For a certain alloy material, the database shows that its austenite transformation temperature is 850°C. By analyzing the temperature gradient, it is found that the temperature gradient of the molten pool center reaches 200°C / mm in the early cooling stage, which may trigger rapid phase change. The phase change temperature interval can be set to 800-900°C to guide the subsequent cooling process simulation. When simulating the random heat conduction behavior in the cooling process using the Monte Carlo method, 10000 random heat conduction paths can be assumed to calculate the cooling rate distribution. The results show that the cooling rate at the edge of the molten pool reaches 50°C / s, and the center area is 20°C / s. If the threshold is set to 40°C / s, there may be a risk of excessive cooling at the edge area. When analyzing the local overheating risk by the convolutional neural network, the temperature distribution image can be input to identify that there is an overheating area within 5 mm of the edge of the molten pool, accounting for about 20%. In one possible implementation, according to the overheating area range and the cooling rate distribution, the microstructure evolution model is combined to calculate the microstructure evolution path.

[0109] Further, adjusting the energy input strategy according to the microstructure reconstruction prediction result, performing precise irradiation repair by using the adjusted energy input control instruction, and determining the fatigue damage repair effect evaluation index include:

[0110] Adjusting the subsequent energy input strategy according to the microstructure reconstruction prediction result, if it is predicted that there is a local overheating risk, reducing the laser power density and prolonging the irradiation time, if it is predicted that the microstructure evolution is insufficient, increasing the energy density and shortening the irradiation interval, to obtain a corrected energy input control instruction;

[0111] The precise irradiation repair is performed by using the modified energy input control instruction, the surface roughness change, the internal stress distribution and the grain boundary healing degree data of the repair area are collected in real time, the fatigue damage repair effect evaluation index is determined by comparing the change amount of the damage characteristic parameters before and after the repair.

[0112] Further, the fatigue damage repair effect evaluation index includes:

[0113] The energy input control instruction parameters are calculated by using the finite element analysis algorithm, and the energy input sequence of the irradiation repair is obtained;

[0114] The irradiation repair is performed according to the energy input sequence, the target area is processed by using the laser irradiation equipment, and the real-time processing data of the repair area is obtained;

[0115] The surface roughness data of the repair area is collected by using an optical microscope, the roughness parameters are extracted by using an image processing algorithm, and the surface roughness change amount is obtained;

[0116] The internal stress distribution data of the repair area is collected by using an X-ray diffractometer, the stress distribution characteristics are analyzed by using a data fitting algorithm, and the internal stress change amount is obtained;

[0117] The grain boundary healing data is collected by using an electron backscatter diffraction technique, the grain boundary healing degree is quantified by using a statistical analysis method, and the grain boundary healing change amount is obtained;

[0118] If the surface roughness change amount, the internal stress change amount and the grain boundary healing change amount all meet the preset threshold value, the change amount data is fused by using a weighted average algorithm, and the fatigue damage repair effect evaluation index is determined;

[0119] If any of the surface roughness change amount, the internal stress change amount and the grain boundary healing change amount does not meet the preset threshold value, the parameters of the energy input model are adjusted, the control instruction is regenerated, the irradiation repair and data collection are repeatedly performed, and a new fatigue damage repair effect evaluation index is obtained.

[0120] Specifically, in the scenario of laser irradiation repair of metal fatigue damage, the preset energy input model can calculate the instruction parameters based on heat conduction and material response characteristics through a finite element analysis algorithm. The finite element analysis divides the target area into small units, simulates the distribution of laser energy in the material, and generates an energy input sequence containing laser power, irradiation time, and scanning speed. Assuming that the surface of an aviation aluminum alloy is repaired, the model may output a sequence of power 500 W, irradiation time 0.2 s, and scanning speed 10 mm / s to ensure uniform energy distribution. Specifically, when performing irradiation repair, the laser irradiation equipment processes the target area according to the sequence. The device focuses the laser beam on the damage area through a high-precision optical system, and the real-time monitoring system records temperature and material response data. For example, an infrared thermometer can detect whether the temperature of the repaired area exceeds 600°C to avoid overheating. After repair, the surface roughness data is collected by an optical microscope. Image processing algorithms analyze the microscopic images to extract the roughness parameter Ra, assuming that Ra is 3.5 μm before repair and decreases to 1.2 μm after repair, indicating that the surface smoothness has been significantly improved. In one embodiment, an X-ray diffractometer is used to collect internal stress distribution data. By fitting the diffraction peak displacement, the residual stress change is quantified. For example, the tensile stress before repair is 200 MPa, and it decreases to 50 MPa after repair, indicating that the stress release effect is obvious. Electron backscatter diffraction technology further collects grain boundary healing data, and statistical analysis shows that the grain boundary integrity improves from 60% to 85%, indicating that the microstructure is optimized. For example, if the roughness change, stress change, and grain boundary healing degree all meet the threshold (such as Ra reduction > 50%, stress reduction > 60%, and grain boundary healing > 80%), then the data is fused by a weighted average algorithm to calculate the repair effect index. Assuming that the weights are 0.4, 0.3, and 0.3 respectively, the comprehensive index is 0.82, indicating that the repair effect is good. If one of them does not meet the standard, such as grain boundary healing is only 70%, then the model parameters are adjusted, such as increasing the power to 550 W or extending the irradiation time to 0.25 s, and the instructions are regenerated and the repair process is repeated. Specifically, when adjusting the parameters, the material characteristics and device limitations need to be considered. For example, aluminum alloy is sensitive to laser, and high power may cause ablation, so the adjustment range needs to be controlled within 10%. The real-time monitoring and feedback mechanism ensures the accuracy of the instruction sequence, and the comparison of dynamic data and predicted results can further optimize the process. This method ensures efficient and stable repair process through multi-dimensional data analysis and parameter iteration, and is suitable for high-precision repair needs in the fields of aerospace, mechanical manufacturing, etc.

[0121] Further, the fatigue damage repair effect evaluation index feedback optimizes the energy input parameter database, establishes a corresponding relationship model between damage types and optimal energy input parameters, and forms a closed-loop control system for multi-scale fatigue damage collaborative repair of aviation aluminum alloys, including:

[0122] Multi-scale feature data of fatigue damage of aerospace aluminum alloy is acquired, and damage type classification is obtained through image processing and sensor acquisition.

[0123] If the damage type classification result contains cracks, wear or corrosion, the corresponding damage type feature vector is determined according to the preset multi-scale damage feature library.

[0124] Key parameters are extracted from the damage type feature vector, and the optimal energy input parameter set is obtained through regression analysis algorithm.

[0125] The optimal energy input parameter set is used to update the energy input parameter database, and the mapping model of damage type and parameter is generated through database query.

[0126] According to the mapping model, the energy input parameters of the repair equipment are adjusted, and the repair effect index data is obtained through real-time monitoring of the repair process.

[0127] If the repair effect index data is lower than the preset threshold, the energy input parameters are optimized through the gradient descent algorithm to obtain the adjusted parameter set.

[0128] The adjusted parameter set drives the closed-loop control mechanism to generate control instructions for the collaborative repair system.

[0129] Specifically, in acquiring the multi-scale feature data of fatigue damage of aerospace aluminum alloy, a variety of sensors and image processing techniques can be combined to comprehensively capture the damage features. An optical microscope can be used to collect the surface micro-topography and generate high-resolution images. Combined with edge detection algorithms, parameters such as crack width and length can be extracted. Assuming the crack width is 5 microns and the length is 200 microns. An ultrasonic sensor can detect internal defects and generate subsurface crack data with a depth of 1 millimeter. It should be noted that these multi-scale data need to be integrated through data fusion algorithms, such as weighted average method, to unify the surface and internal features into a multi-dimensional feature vector, thereby providing a basis for subsequent classification. In the damage type classification, machine learning algorithms such as support vector machines can be used to classify the feature vectors. For example, the crack feature vector may include width, length, depth, etc., while the corrosion feature vector may include corrosion pit diameter and density. Assuming that the classification result shows that the damage is a combination of cracks and corrosion, the corresponding feature vectors are extracted from the pre-set multi-scale damage feature library, such as crack vector [5, 200, 1] and corrosion vector [50, 0.02], representing crack width, length, depth, and corrosion pit diameter and density, respectively. Specifically, when extracting key parameters from the feature vector, principal component analysis can be used to select the parameters that have the greatest impact on the repair effect, such as crack depth and corrosion density. Based on these parameters, a regression analysis algorithm can fit the optimal energy input parameter set. For example, for a crack depth of 1 millimeter, the recommended laser repair energy density is 10 joules per square centimeter, and the pulse frequency is 50 hertz. It should be noted that these parameters will be updated to the energy input parameter database, forming a mapping model of damage type and parameters. For example, the crack type is mapped to high energy density parameters, and the corrosion type is mapped to low frequency pulse parameters. When adjusting the energy input parameters of the repair equipment, a closed-loop control mechanism can be used to monitor the repair process in real time. For example, the laser repair equipment adjusts to 10 joules per square centimeter according to the mapping model, and real-time temperature field data is collected. Assuming that the repair area temperature stabilizes at 500 degrees Celsius, the repair effect is ensured. If the repair effect indicators, such as crack filling rate, are lower than the pre-set threshold of 80%, the parameters are optimized through the gradient descent algorithm. The energy density is adjusted to 12 joules per square centimeter, and the repair is re-executed. When generating control instructions for the collaborative repair system, multiple devices can be controlled collaboratively. The laser device and the mechanical arm work together, with the laser responsible for melting repair and the mechanical arm precisely positioning the repair area. Assuming that the positioning accuracy of the mechanical arm is 0.1 millimeters, the repair area coverage rate can be ensured to be more than 95%. It should be noted that this collaborative mechanism optimizes the instructions through real-time feedback to ensure repair consistency and significantly improve repair efficiency and accuracy.

[0130] Embodiment Two

[0131] The embodiment also provides a welding damage repair precision control system based on digital twinning and edge computing, comprising a data acquisition module, a data processing module, an index determination module and a control module.

[0132] The data acquisition module is configured to acquire surface topography data and internal structure information of the aluminum alloy component by using a multispectral scanning technology.

[0133] The data processing module is configured to acquire damage degree grading data according to the surface topography data and the internal structure information, establish an energy input demand mapping relationship according to the damage degree grading data, acquire energy input parameter combination data according to the energy input demand mapping relationship, perform partition energy irradiation processing according to the energy input parameter combination data, and obtain a microstructure reconstruction prediction result.

[0134] The index determination module is configured to adjust an energy input strategy according to the microstructure reconstruction prediction result, perform accurate irradiation repair by using an adjusted energy input control instruction, and determine a fatigue damage repair effect evaluation index.

[0135] The control module is configured to optimize an energy input parameter database by using the fatigue damage repair effect evaluation index, establish a corresponding relationship model between damage types and optimal energy input parameters, and form a closed-loop control system for aviation aluminum alloy multi-scale fatigue damage collaborative repair.

[0136] The above is only a preferred specific embodiment of the application, but the protection scope of the application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the application, which should be covered in the protection scope of the application. Therefore, the protection scope of the application should be subject to the protection scope of the claims.

Claims

1. A method for welding damage repair precision control of digital twin-edge computing, characterized in that, The method comprises the following steps: obtaining surface topography data and internal structure information of the aviation aluminum alloy component by using a multi-spectral scanning technology; obtaining damage degree classification data according to the surface topography data and the internal structure information; establishing an energy input requirement mapping relationship according to the damage degree classification data, and obtaining energy input parameter combination data according to the energy input requirement mapping relationship; performing regional energy irradiation treatment according to the energy input parameter combination data to obtain microstructure reconstruction prediction results; adjusting the energy input strategy according to the microstructure reconstruction prediction results, performing accurate irradiation repair by using the adjusted energy input control instruction, and determining a fatigue damage repair effect evaluation index; feedback optimizing the energy input parameter database through the fatigue damage repair effect evaluation index, establishing a corresponding relationship model of damage type and optimal energy input parameter, and forming a closed-loop control system of aviation aluminum alloy multi-scale fatigue damage collaborative repair.

2. The digital twin-edge computing-based welding damage repair precision control method of claim 1, wherein The method comprises the following steps: obtaining damage degree classification data according to the surface topography data and the internal structure information comprises the following steps: storing the surface topography data and the internal structure information as three-dimensional point cloud data and stereomicroscopic data; processing the three-dimensional point cloud data and the stereomicroscopic data by using a preset multi-scale fatigue damage detection algorithm, extracting macro crack length and depth distribution characteristics, and obtaining crack geometric parameters; analyzing the stereomicroscopic data by using the multi-scale fatigue damage detection algorithm, identifying micro grain boundary defects, calculating defect density distribution, and obtaining micro defect parameters; determining the damage degree classification data according to the crack geometric parameters and the micro defect parameters.

3. The digital twin-edge computing-based welding damage repair precision control method of claim 2, wherein The method comprises the following steps: determining the damage degree classification data according to the crack geometric parameters and the micro defect parameters comprises the following steps: classifying crack severity in the crack geometric parameters by using a support vector machine algorithm to obtain preliminary damage classification; grouping grain boundary defects according to defect density distribution in the micro defect parameters by using a clustering algorithm to obtain defect density classification; if the preliminary damage classification and the defect density classification are inconsistent, then integrating the crack geometric parameters and the micro defect parameters by using a weighted fusion algorithm to determine final damage degree classification data.

4. The digital twin-edge computing-based welding damage repair precision control method of claim 1, wherein The method comprises the following steps: establishing an energy input requirement mapping relationship according to the damage degree classification data, and obtaining energy input parameter combination data according to the energy input requirement mapping relationship comprises the following steps: obtaining macro crack depth and micro defect density from the damage degree classification data to determine damage feature distribution of each region; if the macro crack depth exceeds a preset threshold, then a high energy density parameter is used to obtain high energy input requirement; if the micro defect density is lower than a critical value, then a low energy density parameter is used to obtain low energy input requirement; establishing an energy input mapping relationship of each region according to the damage feature distribution and the energy input requirement to determine a regional energy distribution scheme; optimizing the energy input mapping relationship by using a genetic algorithm to calculate laser power density and irradiation time combination of each region to obtain a preliminary parameter set; obtaining laser power density and irradiation time combination from the preliminary parameter set to obtain the energy input parameter combination data.

5. The digital twin-edge computing-based welding damage repair precision control method of claim 1, wherein The method comprises the following steps: The microstructure reconstruction prediction result is obtained by performing the zoned energy irradiation processing according to the energy input parameter combination data, and the microstructure reconstruction prediction result comprises: After obtaining the energy input parameter combination, start the zoned energy irradiation processing, and monitor the temperature change curve and the molten pool formation state of each region; Drive the microstructure prediction algorithm by the temperature change curve and the molten pool state data, calculate the grain size change trend, the phase change temperature interval and the cooling rate distribution, judge the local overheating risk degree and the organization evolution path, and obtain the microstructure reconstruction prediction result.

6. The digital twin-edge computing-based welding damage repair precision control method of claim 5, wherein Comprise: After obtaining the energy input parameter combination, start the zoned energy irradiation processing, and monitor the temperature change curve and the molten pool formation state of each region; Based on the energy input parameter combination, determine the distribution of macroscopic crack region and microscopic defect region by a preset classification algorithm, obtain the initial configuration of zoned energy irradiation; According to the initial configuration of zoned energy irradiation, high energy input is applied to the macroscopic crack region in the form of pulse laser, and low energy input is applied to the microscopic defect region in the form of continuous laser, to obtain the laser processing state of each region; Collect the laser processing state of each region through real-time monitoring system, obtain the data of temperature change curve and molten pool formation state, and obtain real-time feedback information; Collect the updated zoned energy irradiation state through real-time feedback information, and obtain the temperature change curve and molten pool formation state.

7. The digital twin-edge computing-based welding damage repair precision control method of claim 5, wherein Comprise: The microstructure reconstruction prediction result comprises: Obtain the temperature fluctuation characteristics by extracting the time sequence characteristics from the temperature change curve and the molten pool formation state, and using the fast Fourier transform method to extract the periodic change mode; According to the temperature fluctuation characteristics, simulate the heat flow distribution inside the molten pool by using the finite element analysis method, and calculate the grain size change trend; Extract the temperature gradient of the key time node from the grain size change trend, determine the phase change temperature interval by combining the preset phase change temperature database; Through the phase change temperature interval, simulate the random heat conduction behavior in the cooling process by using the Monte Carlo method, obtain the cooling rate distribution, and if the cooling rate distribution exceeds the preset threshold, analyze the local overheating risk by using convolutional neural network, and judge the overheating region range; According to the overheating region range and the cooling rate distribution, combine the preset organization evolution model to calculate the organization evolution path; Through the organization evolution path, reconstruct the three-dimensional structure of the microstructure by using the stereology method, and obtain the microstructure reconstruction prediction result.

8. The digital twin-edge computing-based welding damage repair precision control method of claim 1, wherein Comprise: According to the microstructure reconstruction prediction result, adjust the energy input strategy, execute accurate irradiation repair by using the adjusted energy input control instruction, and determine the fatigue damage repair effect evaluation index, which comprises: According to the microstructure reconstruction prediction result, adjust the subsequent energy input strategy, if it is predicted that there is a local overheating risk, reduce the laser power density and prolong the irradiation time, if it is predicted that the organization evolution is insufficient, increase the energy density and shorten the irradiation interval, and obtain the corrected energy input control instruction; The precise irradiation repair is performed by using the modified energy input control instruction, surface roughness change, internal stress distribution and grain boundary healing degree data of the repair area are collected in real time, the fatigue damage repair effect evaluation index is determined by comparing the change amount of the damage characteristic parameters before and after the repair.

9. The digital twin-edge computing-based welding damage repair precision control method of claim 8, wherein, The method comprises the following steps: The fatigue damage repair effect evaluation index is determined by the following steps: The energy input sequence of the irradiation repair is obtained by calculating the energy input control instruction parameters by using the finite element analysis algorithm; The irradiation repair is performed according to the energy input sequence, the target area is processed by using the laser irradiation equipment, and real-time processing data of the repair area are obtained; The surface roughness data of the repair area are collected by using an optical microscope, the roughness parameters are extracted by using an image processing algorithm, and the surface roughness change amount is obtained; The internal stress distribution data of the repair area are collected by using an X-ray diffractometer, the stress distribution characteristics are analyzed by using a data fitting algorithm, and the internal stress change amount is obtained; The grain boundary healing data are collected by using an electron backscatter diffraction technology, the grain boundary healing degree is quantified by using a statistical analysis method, and the grain boundary healing change amount is obtained; If the surface roughness change amount, the internal stress change amount and the grain boundary healing change amount all meet the preset threshold value, the change amount data are fused by using a weighted average algorithm, and the fatigue damage repair effect evaluation index is determined; If any of the surface roughness change amount, the internal stress change amount and the grain boundary healing change amount does not meet the preset threshold value, the parameters of the energy input model are adjusted, the control instruction is regenerated, the irradiation repair and data collection are repeatedly performed, and a new fatigue damage repair effect evaluation index is obtained.

10. A digital twin-edge computing system for weld damage repair precision control, characterized in that, The method comprises the following steps: The method comprises the following steps: The data acquisition module is used to obtain the surface morphology data and internal structure information of the aviation aluminum alloy component by using a multi-spectral scanning technology; The data processing module is used to obtain the damage degree grading data according to the surface morphology data and internal structure information, establish an energy input demand mapping relationship according to the damage degree grading data, obtain energy input parameter combination data according to the energy input demand mapping relationship, perform partition energy irradiation processing according to the energy input parameter combination data, and obtain a microstructure reconstruction prediction result; The index determination module is used to adjust the energy input strategy according to the microstructure reconstruction prediction result, perform precise irradiation repair by using the adjusted energy input control instruction, and determine the fatigue damage repair effect evaluation index; The control module is used to feedback and optimize the energy input parameter database through the fatigue damage repair effect evaluation index, establish a corresponding relationship model of damage type and optimal energy input parameter, and form a closed-loop control system of aviation aluminum alloy multi-scale fatigue damage collaborative repair.

Images