Welding seam quality monitoring method and system in sheet metal welding process

Abstract

The invention relates to the technical field of sheet metal welding, and provides a sheet metal welding process weld joint quality monitoring method and system.The method comprises the steps that according to a thermal expansion coefficient, a welding temperature field and the actual weld joint length, the weld joint deformation amount used for quantifying weld joint linear shrinkage caused by the thermal expansion effect is obtained; acquiring an environment correction factor according to the real-time environment temperature and the real-time environment humidity; according to the welding arc energy, the environment correction factor and the metal plate heat dissipating capacity, metal plate welding net energy used for quantifying effective deposition energy in the welding process is obtained; acquiring a gas flow interference index according to the gas flow rate; and the weld defect probability is obtained according to the welding temperature field, the weld deformation, the metal plate welding net energy and the airflow interference index, and the weld forming quality is monitored according to the weld defect probability. The welding seam defect probability is obtained by fusing metal plate characteristics, environmental attenuation and material nonlinear heat dissipation factors, and the welding seam forming quality monitoring accuracy can be improved.

Description

Technical Field

[0001] This invention relates to the field of sheet metal welding technology, and more specifically, to a method and system for monitoring weld quality during sheet metal welding. Background Technology

[0002] Sheet metal processing is a comprehensive cold working process for thin metal sheets (usually less than 6mm thick). It processes flat materials into parts of specific shapes through shearing, stamping, bending, welding, riveting and other processes. Its core feature is that it utilizes the ductility of metal to change the shape of the sheet material by mechanical force at room temperature, without the need for high-temperature melting or casting. It has advantages such as high processing accuracy, high material utilization and short production cycle.

[0003] In sheet metal welding, weld quality directly affects the product's strength, appearance, and service life. Weld defects (such as cracks, porosity, and over-welding) are common weld quality problems, which can seriously lead to safety hazards. Chinese invention patent application number CN202411809233.6 discloses an online weld quality monitoring method and platform for sheet metal welding. It determines the welding call sequence through welding sequence analysis, extracts the joint surface parameters of the reference sheet metal part and the first sheet metal part, performs material change synchronization analysis, and generates a welding parameter sequence. During the welding process using a battery-powered welding machine controlled by the welding parameters, it monitors the weld quality based on the weld image and outputs weld defect nodes. Finally, it controls the welding of the remaining sheet metal parts and identifies defects based on the welding call sequence, outputting a set of weld defect nodes. While it can monitor weld quality in real time during sheet metal welding, it neglects factors such as sheet metal characteristics, environmental degradation, and nonlinear heat dissipation of materials, leaving room for improvement in the accuracy of its weld quality monitoring. Summary of the Invention

[0004] Based on this, in order to improve the accuracy of weld quality monitoring in sheet metal welding, the present invention provides a method and system for monitoring weld quality during sheet metal welding, the specific technical solution of which is as follows: A method for monitoring weld quality during sheet metal welding includes the following steps: The coefficient of thermal expansion of the sheet metal material, the welding temperature field, and the actual weld length are obtained. Based on the coefficient of thermal expansion, the welding temperature field, and the actual weld length, the weld deformation is obtained to quantify the linear shrinkage of the weld caused by the thermal expansion effect. Get the real-time ambient temperature and humidity, and obtain the environmental correction factor based on the real-time ambient temperature and humidity. The welding arc energy and sheet metal heat dissipation are obtained, and the net sheet metal welding energy is obtained based on the welding arc energy, environmental correction factor and sheet metal heat dissipation to quantify the effective deposition energy during the welding process. Obtain the gas flow rate and then obtain the airflow disturbance index based on the gas flow rate. The probability of weld defects is obtained based on the welding temperature field, weld deformation, net energy of sheet metal welding, and airflow interference index, and the weld formation quality is monitored based on the probability of weld defects.

[0005] The described weld quality monitoring method acquires weld deformation, environmental correction factors, and net energy of sheet metal welding. It integrates sheet metal characteristics, environmental attenuation, and nonlinear heat dissipation factors of materials to obtain the probability of weld defects, which can improve the accuracy of weld defect probability monitoring and thus improve the accuracy of weld formation quality monitoring. In addition, traditional sheet metal welding weld monitoring rarely quantifies airflow interference, while gas flow rate directly affects arc stability and molten pool heat dissipation intensity. The described weld quality monitoring method acquires the airflow interference index and obtains the weld defect probability based on the airflow interference index, which can also improve the accuracy of weld formation quality monitoring to a certain extent.

[0006] Preferably, the specific method for obtaining the probability of weld defects includes the following steps: The curvature of the temperature field is obtained from the welding temperature field, and the net energy change rate is obtained from the net energy of sheet metal welding. The shrinkage thickness ratio is obtained by comparing the weld deformation with the thickness-weighted average. Real-time monitoring of sheet metal material thickness; obtaining the thickness-weighted average of the weld area based on the real-time thickness. A feature matrix is ​​constructed based on the net energy change rate, temperature field curvature, shrinkage thickness ratio, and airflow interference index. The feature matrix is ​​then transformed using a random forest to obtain decision tree feature values. The weld defect probability is then obtained based on the decision tree feature values.

[0007] Preferably, the specific method for monitoring the weld formation quality includes the following steps: Obtain the actual weld width, and then obtain a weld width score based on the actual weld width and the theoretical optimal width. Obtain the actual weld reinforcement height, and obtain the weld reinforcement height score based on the actual weld reinforcement height and the theoretical optimal reinforcement height; Weld defect scores are obtained based on weld defect probability, and sheet metal weld quality scores are obtained based on weld width scores, weld reinforcement scores, and weld defect scores.

[0008] Preferably, the weld quality monitoring method further includes the following steps: The partial derivative of the welding current with respect to the quality score is obtained based on the sheet metal weld quality score, and the current score gradient is obtained based on the partial derivative to characterize the amount of adjustment current required to represent the unit quality score change. Obtain the second-order partial derivative of the weld defect probability, and obtain the acceleration of the defect probability change based on the second-order partial derivative; The welding current adjustment amount is obtained based on the current scoring gradient and the acceleration of the defect probability change, and the welding machine current is adjusted according to the welding current adjustment amount.

[0009] Preferably, the specific method for obtaining welding arc energy and sheet metal heat dissipation includes the following steps: The arc voltage, welding current, and thermal efficiency coefficient are obtained, and the welding arc energy is obtained based on the arc voltage, welding current, and thermal efficiency coefficient. The real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material are obtained. The heat dissipation of the sheet metal is obtained based on the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material.

[0010] Preferably, the specific method for obtaining the weld deformation includes the following steps: The initial weld deformation is obtained based on the coefficient of thermal expansion, the welding temperature field, and the actual weld length to quantify the linear shrinkage of the weld caused by the thermal expansion effect. The initial weld deformation is corrected based on the environmental correction factor to obtain the final weld deformation.

[0011] A sheet metal welding process weld quality monitoring system, used to implement the sheet metal welding process weld quality monitoring method, comprising: The deformation acquisition module is used to acquire the thermal expansion coefficient of the sheet metal material, the welding temperature field, and the actual weld length. Based on the thermal expansion coefficient, the welding temperature field, and the actual weld length, it acquires the weld deformation amount used to quantify the linear shrinkage of the weld caused by the thermal expansion effect. The correction factor acquisition module is used to acquire the real-time ambient temperature and humidity, and to acquire the environmental correction factor based on the real-time ambient temperature and humidity. The net energy acquisition module is used to acquire the welding arc energy and sheet metal heat dissipation. Based on the welding arc energy, environmental correction factor and sheet metal heat dissipation, it acquires the sheet metal welding net energy to quantify the effective deposition energy during the welding process. The interference index acquisition module is used to acquire the gas flow rate and obtain the airflow interference index based on the gas flow rate. The weld quality monitoring module is used to obtain the probability of weld defects based on the welding temperature field, weld deformation, net energy of sheet metal welding, and airflow interference index, and to monitor the weld formation quality based on the probability of weld defects.

[0012] Preferably, the weld quality monitoring module includes: The net energy change rate acquisition unit is used to obtain the temperature field curvature based on the welding temperature field and the net energy change rate based on the net energy of sheet metal welding. The thickness weighting value acquisition unit is used to monitor the real-time thickness of sheet metal materials and obtain the thickness weighting average value of the weld area based on the real-time thickness. The shrinkage thickness ratio acquisition unit is used to obtain the shrinkage thickness ratio based on the ratio between the weld deformation and the thickness-weighted average value. The feature matrix construction unit is used to construct a feature matrix based on the net energy change rate, temperature field curvature, shrinkage thickness ratio, and airflow disturbance index. The defect probability acquisition unit is used to perform feature transformation on the feature matrix based on random forest, obtain decision tree feature values, and obtain the weld defect probability based on the decision tree feature values.

[0013] Preferably, the weld quality monitoring module further includes: The weld width score acquisition unit is used to obtain the actual weld width and to obtain the weld width score based on the actual weld width and the theoretical optimal width. The weld reinforcement height score acquisition unit is used to obtain the actual weld reinforcement height and obtain the weld reinforcement height score based on the actual weld reinforcement height and the theoretical optimal reinforcement height. The weld quality scoring unit is used to obtain weld defect scores based on weld defect probability, and to obtain sheet metal weld quality scores based on weld width scores, weld reinforcement scores, and weld defect scores.

[0014] Preferably, the net energy change rate acquisition unit is based on the formula Obtain the net rate of change of energy; in, Indicates the net energy consumption of sheet metal welding. These represent the arc voltage, welding current, and thermal efficiency coefficient, respectively. Indicates the environmental correction factor. These represent the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material, respectively. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall process of a method for monitoring weld quality in a sheet metal welding process according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a specific method for obtaining welding arc energy and sheet metal heat dissipation in one embodiment of the present invention. Figure 3 This is a flowchart illustrating a specific method for obtaining the probability of weld defects in one embodiment of the present invention. Figure 4 This is a flowchart illustrating a specific method for monitoring weld formation quality in one embodiment of the present invention. Figure 5 This is a flowchart illustrating a method for monitoring weld quality during sheet metal welding, according to another embodiment of the present invention. Figure 6 This is a flowchart illustrating a specific method for obtaining weld deformation in one embodiment of the present invention. Figure 7 This is a schematic diagram of the overall structure of a sheet metal welding process weld quality monitoring system according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the functional module structure of the weld quality monitoring module in one embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to its embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of the invention.

[0017] Before describing the specific embodiments of the present invention, a brief introduction to the prior art will be given first.

[0018] Sheet metal processing can be broadly categorized into forming and joining. Forming processes include shearing (cutting sheet metal using a CNC shearing machine), stamping (using dies to perform punching, blanking, and stretching operations on a press), bending (bending sheet metal to a specified angle using a bending machine, commonly 30°-180°), and stretching (using dies to plastically deform sheet metal, forming hollow or concave / convex parts). Joining processes encompass welding (argon arc welding, spot welding, etc.), riveting (solid rivets or blind rivets), and bolting, suitable for scenarios with varying strength and sealing requirements.

[0019] Currently, quality monitoring during the welding process relies on manual inspection or basic welding control systems. However, these methods suffer from poor real-time performance, insufficient accuracy, and excessive human intervention, making them ineffective in handling complex welding tasks. Existing technologies still face the challenge of synchronously monitoring and optimizing parameter changes during the welding process. There is an urgent need for a precise, real-time, and efficient weld quality monitoring system to improve automated control during welding, reduce defects, lower production costs, and enhance overall efficiency. To address this, Chinese invention patent application number CN202411809233.6 discloses an online weld quality monitoring method and platform for sheet metal welding. This method determines the welding call sequence through welding sequence analysis, extracts the joint surface parameters of the reference sheet metal part and the first sheet metal part, performs material change synchronization analysis, generates a welding parameter sequence, and during the welding process using a battery-powered welding machine controlled by the welding parameters, performs quality monitoring based on weld images and outputs weld defect nodes. Finally, based on the welding call sequence, it performs welding control and defect identification on the remaining sheet metal parts, outputting a set of weld defect nodes.

[0020] However, factors such as thermodynamics, geometric deformation, and environmental interference during sheet metal welding can affect welding defects and even weld quality. While the aforementioned online weld quality monitoring methods and platforms for sheet metal welding can monitor weld quality in real time, they neglect factors such as sheet metal characteristics, environmental degradation, and nonlinear heat dissipation of materials, and overlook the coupling effects between different factors. Therefore, there is room for further optimization and improvement in the accuracy of weld quality monitoring.

[0021] In addition, traditional sheet metal welding weld quality monitoring rarely quantifies airflow interference. However, airflow velocity directly affects arc stability and molten pool heat dissipation intensity. There is also a certain coupling effect between gas density and the thermal sensitivity of sheet metal materials. Traditional sheet metal welding weld monitoring, as well as the aforementioned online weld quality monitoring methods and platforms used in the sheet metal welding process, have ignored the impact of gas interference on weld quality and need improvement.

[0022] To improve the accuracy of monitoring the forming quality of sheet metal welds, such as Figure 1 As shown, an embodiment of the present invention provides a method for monitoring weld quality during sheet metal welding, comprising the following steps: S1, obtain the thermal expansion coefficient of the sheet metal material, the welding temperature field, and the actual weld length. Based on the thermal expansion coefficient, the welding temperature field, and the actual weld length, obtain the weld deformation amount used to quantify the linear shrinkage of the weld caused by the thermal expansion effect.

[0023] Specifically, the coefficient of thermal expansion of sheet metal materials refers to the rate of expansion of the sheet metal material's length per 1°C increase in temperature. It reflects the material's thermal sensitivity and can be obtained by referring to tables in material handbooks. For example, for 304 stainless steel, it is generally taken as 17.2 × 10⁻⁶. -6 / ℃. The welding temperature field can be acquired through infrared thermal imager monitoring. After acquiring the welding temperature field, the temperature difference value is calculated based on the welding temperature field and the ambient temperature. Specifically, the highest temperature in the weld area can be monitored in real time using an infrared thermal imager, and the surrounding temperature can be obtained using environmental sensors. Then, the temperature difference between the temperature of the weld center molten pool area and the ambient temperature is calculated to obtain the temperature difference value. The actual weld length is the geometric length of the actual welding path, which serves as the basis for the cumulative weld shrinkage and can be determined through the planned welding path.

[0024] For example, the weld deformation is calculated based on the product of the thermal expansion coefficient of the sheet metal material, the temperature difference, and the actual weld length, i.e., weld deformation = thermal expansion coefficient of sheet metal material × temperature difference × actual weld length.

[0025] After obtaining the weld deformation, the welding path of the welding robot can be compensated in real time. For example, a mapping function between different weld deformations and path offsets can be established first, and then the path offset can be obtained based on the mapping function and the calculated weld deformation to compensate the welding path in real time.

[0026] Here, by quantifying the linear contraction caused by thermal expansion, a real-time compensation basis can be provided for the welding process, solving the problem of uncontrolled deformation caused by the thin-walled characteristics of sheet metal parts.

[0027] S2, Obtain real-time ambient temperature In addition, real-time ambient temperature, and environmental correction factors are obtained based on real-time ambient temperature and real-time ambient humidity.

[0028] Specifically, you can first obtain the calibrated ambient reference temperature (usually 20℃), and then obtain the ambient temperature difference based on the real-time ambient temperature and the ambient reference temperature. An environmental correction factor function is constructed using real-time ambient humidity and ambient temperature difference as independent variables and an environmental correction factor as the dependent variable. The relevant references in the environmental correction factor function are set empirically or calibrated experimentally. Finally, the environmental correction factor is obtained based on the environmental correction factor function. The type of this environmental correction factor function includes, but is not limited to, a linear function in two variables, an exponential function, a polynomial function, and a piecewise function.

[0029] For example, environmental correction factors .in, These are the humidity attenuation coefficient and the temperature difference attenuation coefficient, respectively. The humidity attenuation coefficient characterizes the strength of humidity's suppression of system performance, typically between 0.005 and 0.02. The temperature difference attenuation coefficient characterizes the strength of interference from ambient temperature fluctuations on the system, typically between 0.001 and 0.008. Of course, both the humidity attenuation coefficient and the temperature difference attenuation coefficient can be adjusted empirically, and will not be elaborated further here.

[0030] These can be understood as humidity compensation and temperature difference compensation, respectively. Since high humidity dilutes the protective gas, the humidity compensation item allows for an appropriate increase in the flow rate of the protective device, such as argon gas, when humidity rises. Because temperature deviations from the calibration reference can easily cause thermal expansion / thermal noise, the temperature difference compensation item allows for dynamic adjustment of the sampling frequency and compensation of the welding current when the temperature difference increases.

[0031] This environmental correction factor, through an exponential decay mechanism, can transform linearly changing environmental disturbances into nonlinear decays, which better reflects actual physical processes.

[0032] S3, obtains the welding arc energy and sheet metal heat dissipation, and obtains the net sheet metal welding energy based on the welding arc energy, environmental correction factor, and sheet metal heat dissipation to quantify the effective deposition energy during the welding process.

[0033] Generally, according to the principle of dynamic energy conservation, after the arc input energy decays due to environmental factors, the net energy used for actual welding is obtained after deducting the heat dissipation characteristics of the sheet metal. Therefore, a dynamic energy balance model can be constructed to quantify the effective welding energy during the welding process. This addresses the real-time decay effect of environmental interference on arc energy and the nonlinear heat dissipation loss caused by the thin-walled characteristics of sheet metal, ultimately yielding the net welding energy of sheet metal that directly determines the stability of the weld pool and the weld penetration depth. Specifically, the net welding energy of sheet metal = welding arc energy × environmental correction factor - sheet metal heat dissipation.

[0034] As a preferred technical solution, to improve the accuracy of the net energy of sheet metal welding, radiation loss can be introduced. The final net energy of sheet metal welding is calculated as: welding arc energy × environmental correction factor - sheet metal heat dissipation - radiation loss. Radiation loss is calculated as: the product of the Stefan-Boltzmann constant, plasma emissivity, and the fourth power difference between plasma temperature and ambient temperature. Since obtaining radiation loss is a conventional technique in this field, it will not be elaborated upon here.

[0035] As a preferred technical solution, such as Figure 2 As shown, the specific methods for obtaining welding arc energy and sheet metal heat dissipation include the following steps: S31, obtain the arc voltage, welding current and thermal efficiency coefficient, and obtain the welding arc energy based on the arc voltage, welding current and thermal efficiency coefficient.

[0036] Arc voltage can be directly collected from the welding power source, reflecting the stability of the arc length; welding current can be measured by a Hall sensor, which determines the core parameter of penetration depth; the thermal efficiency coefficient is calibrated by calorimetry or set based on experience, and is generally related to the composition of the shielding gas. For example, the thermal efficiency of argon arc welding is usually between 60% and 85%, that is, the thermal efficiency coefficient of argon gas is between 0.6 and 0.85, and the thermal efficiency coefficient of carbon dioxide welding is usually between 0.8 and 0.95.

[0037] Here, welding arc energy = arc voltage × welding current × thermal efficiency coefficient.

[0038] S32, obtain the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material, and obtain the heat dissipation of the sheet metal based on the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material.

[0039] The real-time temperature of the weld pool can be obtained by scanning the weld pool area with an infrared thermal imager, while the initial temperature of the substrate before welding can be measured by a contact thermocouple. Thermal conductivity characterizes the heat dissipation capacity of a material; for example, the thermal conductivity of aluminum is typically in the range of 205-250 W / (m·K).

[0040] For example, heat dissipation of sheet metal = thermal conductivity of sheet metal material × real-time temperature of weld pool × ln(real-time temperature of weld pool / initial temperature of substrate before welding). Here, the logarithm is used to characterize nonlinear heat dissipation.

[0041] Generally, in existing technologies, heat dissipation of sheet metal is calculated using a linear heat dissipation model: heat dissipation = thermal conductivity of sheet metal material × (real-time temperature of weld pool - initial temperature of substrate before welding). However, due to significant differences in thin-plate welding, the linear heat dissipation model may deviate from reality. This embodiment uses a logarithmic form: thermal conductivity of sheet metal material × real-time temperature of weld pool × ln(real-time temperature of weld pool / initial temperature of substrate before welding). It addresses the large specific surface area characteristics of thin sheet metal by replacing the conventional linear heat dissipation model with a nonlinear heat dissipation model, enabling lateral comparisons under different working conditions and reflecting the dramatic temperature gradient changes along the sheet metal thickness direction, thus being more realistic.

[0042] S4, Obtain the gas flow rate, and then... Obtain the airflow disturbance index.

[0043] Real-time monitoring of airflow disturbances, such as using a micro-differential pressure sensor to detect gas velocity, generates an airflow disturbance index. For example, the airflow disturbance index... .

[0044] Specifically, the unit of gas velocity, m / s, reflects the intensity of airflow in the welding workshop, which directly affects arc stability and the heat dissipation intensity of the molten pool; the material-environment coupling coefficient... The effects of gas density and material thermal sensitivity can be calibrated experimentally, such as using 0.11 for 304 stainless steel. Since dynamic pressure is proportional to the square of the flow velocity in aerodynamics, the gas flow velocity in the airflow interference exponential function adopts a square relationship, which can quantitatively describe the nonlinear destructive effect of airflow on the welding process, amplify the influence of strong airflow, and make it conform to actual physical phenomena.

[0045] Generally, the higher the airflow disturbance index, the stronger the intervention required. When it exceeds 0.4 or a higher preset value, an alarm can be triggered. The following is a graded control strategy based on the airflow disturbance index:

[0046] The airflow interference index has three core functions: 1. Quantitatively assessing the impact of airflow on the welding process; 2. Serving as an input parameter for heat loss compensation models; 3. Providing a basis for judging defect warning thresholds.

[0047] S4. The probability of weld defects is obtained based on the welding temperature field, weld deformation, net energy of sheet metal welding, and airflow interference index. The weld formation quality is then monitored based on the probability of weld defects.

[0048] As a preferred technical solution, such as Figure 3 As shown, the specific method for obtaining the probability of weld defects includes the following steps: S41, Obtain the curvature of the temperature field based on the welding temperature field. And based on the net energy of sheet metal welding Obtain the net energy change rate .

[0049] The net energy change rate can be understood as the fluctuation of effective heat input per unit time. It is a core parameter that determines the stability of the molten pool. A value greater than 0 indicates heat accumulation, which may lead to burn-through. A value less than 0 indicates that the heat dissipation is too fast, which may pose a risk of non-fusion.

[0050] Temperature field curvature can be understood as the two-dimensional temperature curvature of the molten pool and heat-affected zone, used to predict the phase transformation stress of microstructure. Positive curvature indicates heat concentration, which is prone to cracking, while negative curvature indicates heat dissipation imbalance, which is prone to porosity.

[0051] S42 monitors the real-time thickness of sheet metal materials and obtains the thickness-weighted average of the weld area based on the real-time thickness.

[0052] For example, thickness-weighted average .in, These represent the number of effective measurement points and the real-time thickness of the i-th measurement point at time t, respectively. The real-time thickness of the i-th measurement point at time t reflects microscopic fluctuations in material thickness, such as rolling deviations and localized corrosion, and can be acquired through a laser displacement sensor array. The number of effective measurement points can be determined by the weld width to optimize the sampling density. This thickness-weighted average is used to eliminate the influence of local anomalies.

[0053] Generally, uneven sheet metal thickness can lead to differences in heat capacity, causing distortion of the molten pool temperature field and resulting in defects such as incomplete fusion or burn-through. In this case, dynamic energy compensation based on thickness-weighted average can be used. For example, the welding current can be linearly adjusted according to a certain ratio based on the ratio between the thickness-weighted average and the design sheet thickness to stabilize the penetration depth.

[0054] S43, based on weld deformation The shrinkage thickness ratio is obtained by comparing the shrinkage thickness ratio with the thickness-weighted mean. .

[0055] Shrinkage ratio is a key indicator for controlling assembly precision. Generally speaking, when it is greater than a certain value of 0.1 or 0.15, it indicates that there is a risk of thin plate warping. When it is less than a certain value such as 0.05 or 0.03, it can be understood that the residual stress exceeds the standard.

[0056] S44. A feature matrix is ​​constructed based on the net energy change rate, temperature field curvature, shrinkage thickness ratio, and airflow interference index. The feature matrix is ​​then transformed using a random forest to obtain decision tree feature values. The weld defect probability is then obtained based on the decision tree feature values.

[0057] For example, the feature matrix F is denoted as Output weld defect probability .in, These represent the feature weights and the feature transformation of the decision tree, respectively. σ represents the Sigmoid function, which is used to transform the output of the random forest into the probability of welding defects.

[0058] Decision tree feature transformation is used to convert physical features into defect criteria. Typical branching rules include: if the temperature field curvature is >15°C / mm 2 If the net energy change rate is <-50 J / ms, then the decision tree feature transformation = 1; if the airflow disturbance index is >0.4 and the shrinkage thickness ratio is <0.05, then the decision tree feature transformation = 0.7.

[0059] If the probability of welding defects is greater than 0.85, the machine should be stopped immediately for inspection or process adjustment; if it is between 0.6 and 0.85, the speed should be reduced appropriately; if it is less than 0.3, the welding speed can be increased appropriately.

[0060] For example, suppose that at a certain time t, the sensor collects the following data: net energy change rate -45 J / s, temperature field curvature = 18°C / mm 2 Shrinkage thickness ratio = 0.06, airflow interference index = 0.38.

[0061] The random forest model consists of 5 decision trees, with weights of 0.22, 0.18, 0.25, 0.20, and 0.15 respectively. Each tree outputs a predicted value after a non-linear transformation of the feature matrix F: Decision Tree 1 (focusing on heat input): If the net energy change rate is <-30 J / s and the airflow disturbance index is >0.3, then ,otherwise ; Decision Tree 2 (Focusing on Temperature Field): If the curvature of the temperature field is >15°C / mm 2 And if the shrinkage thickness ratio is <0.08, then ,otherwise ; Decision Tree 3 (Comprehensive Criterion): If the airflow disturbance index > 0.35 and the net energy change rate < -40, then ,otherwise ; Decision Tree 4 (Geometric Stability): If the shrinkage thickness ratio > 0.05 and the temperature field curvature < 20°C / mm 2,but ,otherwise ; Decision Tree 5 (Airflow Sensitive): If the airflow disturbance index is >0.4, then ,otherwise .

[0062] Substituting the current features yields Then, after linear weighted summation, The final weld defect probability, converted by the Sigmoid function, is 65.5%.

[0063] In summary, the weld quality monitoring method described above obtains weld deformation, environmental correction factors, and net energy of sheet metal welding. It integrates sheet metal characteristics, environmental attenuation, and nonlinear heat dissipation factors of materials to obtain the probability of weld defects, which can improve the accuracy of weld defect probability monitoring and thus improve the accuracy of weld formation quality monitoring. In addition, traditional sheet metal welding weld monitoring rarely quantifies airflow interference, while gas flow rate directly affects arc stability and molten pool heat dissipation intensity. The weld quality monitoring method described above obtains the airflow interference index and obtains the weld defect probability based on the airflow interference index, which can also improve the accuracy of weld formation quality monitoring to a certain extent.

[0064] In one embodiment, such as Figure 4 As shown, the specific method for monitoring weld formation quality includes the following steps: S45, obtain the actual weld width, and obtain the weld width score based on the actual weld width and the theoretical optimal width.

[0065] The actual weld width is the surface width of the solidified molten metal, which can be obtained using a laser profile scanner and reflects whether the heat input is sufficient. The theoretical optimal width is generally set based on the plate thickness or experience.

[0066] S46, obtain the actual weld reinforcement height, and obtain the weld reinforcement height score based on the actual weld reinforcement height and the theoretical optimal reinforcement height.

[0067] The actual weld reinforcement height is the height by which the weld protrudes above the base material surface. If it is too high, it indicates stress concentration; if it is too low, it indicates insufficient strength. The theoretically optimal reinforcement height is generally set based on the plate thickness or experience.

[0068] S47. Obtain a weld defect score based on the weld defect probability, and obtain a sheet metal weld quality score based on the weld width score, weld reinforcement score, and weld defect score.

[0069] Sheet metal weld quality rating ;in, These represent the actual weld width, the theoretical optimal width, the actual weld reinforcement height, and the theoretical optimal weld reinforcement height, respectively. These are all weighting coefficients, which can be set based on experience; The tolerance variance for excess height is the allowable range of excess height fluctuation. Generally, it is 0.04 for thin plates and 0.12 for thick plates. These represent the weld width score, weld reinforcement score, and weld defect score, respectively.

[0070] The weld width score uses a reciprocal domain constraint, maximizing the score when the actual weld width equals the theoretical optimal width. The greater the deviation of the actual weld width from the theoretical optimal width, the lower the score. The weld reinforcement score uses a Gaussian distribution fitting, assuming the reinforcement deviation follows a normal distribution, with the standard deviation... When the absolute value of the difference between the actual weld reinforcement height and the theoretical optimal reinforcement height is 0.1 mm, 88% of the score is retained; when the absolute value is 0.2 mm, 61% of the score remains; and when the absolute value is 0.3 mm, only 32% of the score remains. Therefore, using a Gaussian distribution fitting function to calculate the weld reinforcement height score is more consistent with the fatigue strength decay law than linear penalty. The weld defect score is calculated by direct deduction.

[0071] Based on the sheet metal weld quality score, a mapping table between the sheet metal weld quality score and the weld formation quality level can be constructed, and corresponding control measures can be formulated, as shown in the table below:

[0072] In summary, this sheet metal weld quality scoring function model unifies and quantifies macroscopic dimensional control and microscopic defect prediction. It can not only serve as an evaluation index for weld formation quality but also as a basis for real-time adjustment of welding parameters.

[0073] In one embodiment, such as Figure 5 As shown, the weld quality monitoring method further includes the following steps: S5. Obtain the partial derivative of welding current with respect to quality score based on sheet metal weld quality score, and obtain the current score gradient based on the partial derivative to characterize the amount of adjustment current required to represent a unit change in quality score. .

[0074] The current rating gradient can be obtained by fitting a mapping relationship to historical data. For example, for every 1 point decrease in the score, the reference current needs to be increased by 0.8%, and the gradient value is updated after each weld segment (about 100mm) is completed.

[0075] S6, Obtain the second partial derivative of the weld defect probability, and obtain the acceleration of the defect probability change based on the second partial derivative. .

[0076] The acceleration of the change in defect probability reflects the degree of sudden escalation of risk, such as sudden airflow interference or material contamination. It can be calculated using the central difference method based on the time series data of real-time weld defect probability. It is used to provide early warning of sudden defects, such as the instantaneous surge in porosity risk caused by shielding gas failure.

[0077] S7, Obtain the welding current adjustment amount based on the current scoring gradient and the acceleration of defect probability change. The welding machine current is adjusted according to the welding current adjustment amount.

[0078] The welding current adjustment is directly input to the welding machine power supply to adjust the heat input and optimize the welding process. A positive value indicates that the current needs to be increased, and a negative value indicates that the current needs to be decreased.

[0079] in, These represent the proportional gain coefficient and the derivative gain coefficient, respectively. The proportional gain coefficient controls the sensitivity to scoring deviations; the larger its value, the stronger the system's response to score changes. It can be determined by process experiments or set empirically. The derivative gain coefficient controls the sensitivity to defect probability acceleration; the larger its value, the faster the system responds to sudden risks. It can be set based on the material's thermal sensitivity or empirically.

[0080] For example, the proportional gain coefficient and the differential gain coefficient can be set with reference to the following table:

[0081] These can be understood as proportional terms and differential terms, respectively. The proportional term addresses gradual quality degradation, such as welding torch aging and gas purity decay, by slowly compensating for the current through a linear gradient relationship. The differential term addresses sudden risks, such as airflow impact and slag clogging, by capturing the second-order rate of change of the defect probability and achieving millisecond-level advance adjustment.

[0082] Overall, the welding current adjustment is proportional to the change trend of the quality score to maintain stable quality, thus having a score feedback control function; it provides advance compensation for sudden changes in the defect probability to avoid sudden defects, thus having a risk warning control function; in addition, it can balance the heat input by precisely controlling the thermodynamic state of the molten pool through current adjustment.

[0083] In one embodiment, such as Figure 6 As shown, the specific method for obtaining weld deformation includes the following steps: S11, based on the coefficient of thermal expansion, welding temperature field and actual weld length, obtain the initial weld deformation amount used to quantify the linear shrinkage of the weld caused by the thermal expansion effect.

[0084] For example, the initial weld deformation = thermal expansion coefficient × welding temperature field × actual weld length, which reflects the intrinsic deformation trend of the material.

[0085] S12, correct the initial weld deformation amount according to the environmental correction factor to obtain the final weld deformation amount.

[0086] Specifically, since water vapor evaporation absorbs heat in high-temperature environments, it reduces the effective temperature difference, and temperature fluctuations in the workshop cause changes in the heat dissipation rate. Therefore, an environmental correction factor can be superimposed to correct the initial weld deformation.

[0087] Environmental Correction Factors Used to quantify environmental disturbances, its exponential decay form ensures the coefficient remains positive, and the shrinkage decreases as the environment deteriorates. Here, the environmental correction factor is essentially a heat exchange efficiency correction term.

[0088] For example, the final weld deformation = initial weld deformation × environmental correction factor. Thus, the final weld deformation incorporates the triple coupling effect of material properties, process parameters, and environmental factors, and can be used to predict the actual shrinkage deformation.

[0089] Because thin plates have low stiffness and fast heat dissipation, they deform more than thick plates under the same heat input. After welding, thin plates are prone to amplified deformation due to insufficient stiffness, which directly affects the assembly sealing performance. Therefore, for thin plates with a certain thickness, such as 0.15mm-3.5mm or less than 3.0mm, the final weld deformation can be further corrected based on the plate thickness.

[0090] Specifically, first, a linear amplification factor negatively correlated with plate thickness is obtained. Then, the final weld deformation is corrected based on the linear amplification factor, resulting in the final weld deformation = initial weld deformation × environmental correction factor × linear amplification factor. For example, the linear amplification factor = 1.2 - adjustment factor × plate thickness, where the adjustment factor is typically 0.05. Parameter 1.2 corresponds to the maximum deformation amplification rate of the ultra-thin plate, and the adjustment factor × plate thickness term is used to quantify the effect of increased plate thickness on deformation suppression.

[0091] When the plate thickness is 0.5mm (ultra-thin), the linear amplification factor is 1.2−0.05×0.5=1.175, and the deformation is amplified by 17.5%; when the plate thickness is 3.0mm (critical value), the linear amplification factor is 1.2−0.05×3=1.05, and the deformation is amplified by 5%.

[0092] Specifically, increased welding speed leads to a decrease in heat input. Simultaneously, high-speed welding narrows the heat-affected zone and accelerates cooling, thus reducing thermal deformation. Therefore, it is necessary to increase the adjustment factor to reduce the linear amplification coefficient. In other words, the faster the welding torch moves during welding, the more concentrated the heat input and the faster the cooling rate, significantly impacting the deformation behavior of thin plates. Therefore, the adjustment factor can be correlated with the welding speed, increasing the adjustment factor to compensate for the cooling rate as the welding speed increases.

[0093] For example, the adjustment factor = 0.05 + 0.001 × welding speed (mm / s). Assuming a welding speed of 2 mm / s under low-speed welding conditions, the adjustment factor = 0.05 + 0.001² = 0.05². If the plate thickness is 1 mm, the linear amplification factor = 1.2 - 0.05² = 1.148. Assuming a welding speed of 10 mm / s under high-speed welding conditions, the adjustment factor = 0.05 + 0.01 = 0.06. For the same plate thickness of 1 mm, the corresponding linear amplification factor = 1.2 - 0.06 = 1.14. It can be seen that for the same plate thickness, the linear amplification factor of high-speed welding is smaller, meaning the predicted actual deformation is smaller than that of low-speed welding, which conforms to the principle that high-speed welding involves less heat input and less deformation.

[0094] The adjustment factor function is applicable to the prediction of welding deformation of thin plates less than 3mm. Through this function, the amount of weld deformation can be predicted more accurately in the welding process design stage, so that compensation measures can be taken in advance, such as anti-deformation, optimization of welding sequence, and adjustment of parameters.

[0095] It should be noted that the linear amplification factor and adjustment factor function are empirical formulas derived from experimental data of specific materials (such as aluminum alloys, stainless steel, etc.) and welding methods (such as MIG, laser welding, etc.). For different materials and welding methods, parameters 1.2, 0.05, and 0.001 need to be adjusted appropriately.

[0096] In summary, the combination of linear amplification factor and adjustment factor function, which takes into account the influence of plate thickness and welding speed on deformation, makes the prediction of deformation of thin plate welds more accurate and is conducive to optimizing sheet metal welding process.

[0097] like Figure 7 As shown, an embodiment of the present invention also provides a weld quality monitoring system for sheet metal welding process, used to implement the weld quality monitoring method for sheet metal welding process, which includes a deformation acquisition module, a correction factor acquisition module, a net energy acquisition module, an interference index acquisition module, and a weld quality monitoring module.

[0098] The deformation acquisition module is used to obtain the thermal expansion coefficient of the sheet metal material, the welding temperature field, and the actual weld length. Based on the thermal expansion coefficient, the welding temperature field, and the actual weld length, it obtains the weld deformation amount used to quantify the linear shrinkage of the weld caused by the thermal expansion effect.

[0099] The correction factor acquisition module is used to acquire the real-time ambient temperature and humidity, and to acquire the environmental correction factor based on the real-time ambient temperature and humidity. The net energy acquisition module is used to acquire the welding arc energy and sheet metal heat dissipation, and to acquire the sheet metal welding net energy, which is used to quantify the effective deposition energy during the welding process, based on the welding arc energy, environmental correction factor, and sheet metal heat dissipation.

[0100] The interference index acquisition module is used to acquire the gas flow rate and obtain the airflow interference index based on the gas flow rate; the weld quality monitoring module is used to acquire the weld defect probability based on the welding temperature field, weld deformation, sheet metal welding net energy and airflow interference index, and monitor the weld forming quality based on the weld defect probability.

[0101] Specifically, such as Figure 8 As shown, the weld quality monitoring module includes a net energy change rate acquisition unit, a thickness weighted value acquisition unit, a shrinkage thickness ratio acquisition unit, a feature matrix construction unit, and a defect probability acquisition unit.

[0102] The net energy change rate acquisition unit is used to obtain the temperature field curvature based on the welding temperature field and the net energy change rate based on the net energy of sheet metal welding; the thickness weighting value acquisition unit is used to monitor the real-time thickness of the sheet metal material and obtain the thickness weighting average of the weld area based on the real-time thickness; the shrinkage thickness ratio acquisition unit is used to obtain the shrinkage thickness ratio based on the ratio between the weld deformation and the thickness weighting average.

[0103] The feature matrix construction unit is used to construct a feature matrix based on the net energy change rate, temperature field curvature, shrinkage thickness ratio, and airflow interference index; the defect probability acquisition unit is used to perform feature transformation on the feature matrix based on random forest, obtain decision tree feature values, and obtain the weld defect probability based on the decision tree feature values.

[0104] For example, the net energy change rate acquisition unit obtains the net energy change rate according to the formula Obtain the net rate of change of energy; where, Indicates the net energy consumption of sheet metal welding. These represent the arc voltage, welding current, and thermal efficiency coefficient, respectively. Indicates the environmental correction factor. These represent the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material, respectively.

[0105] For example, the feature matrix F is denoted as Output weld defect probability .in, These represent the feature weights and the feature transformation of the decision tree, respectively. σ represents the Sigmoid function, which is used to transform the output of the random forest into the probability of welding defects.

[0106] Feature weights consist of a static proportion of feature importance and a dynamic time adjustment factor. For example, the real-time weight of the k-th feature in the decision-making process... , This represents the static importance score of the k-th feature. This represents the sum of the importance of all features, used as the normalized denominator. t is the time sensitivity adjustment factor, which can be understood as a hyperparameter that determines the rate of change of weights, and t is the duration of continuous operation of the decision system.

[0107] The core objective of the feature weighting function is to address the problem that static feature weights cannot adapt to the evolution of system state. In the early stages, it can quickly establish decision-making benchmarks based on highly important features, and in the later stages, it can suppress interference from outdated features and highlight the value of real-time data.

[0108] In summary, the essence of this feature weight function is to make the weights of all features decay by the same proportion over time, while maintaining the relative importance order among features, thereby achieving time-sensitive decay of feature importance.

[0109] Decision tree feature transformation is used to convert physical features into defect criteria. Typical branching rules include: if the temperature field curvature is >15°C / mm2 and the net energy change rate is <-50J / ms, then the decision tree feature transformation = 1; if the airflow disturbance index is >0.4 and the shrinkage thickness ratio is <0.05, then the decision tree feature transformation = 0.7.

[0110] If the probability of welding defects is greater than 0.85, the machine should be stopped immediately for inspection or process adjustment; if it is between 0.6 and 0.85, the speed should be reduced appropriately; if it is less than 0.3, the welding speed can be increased appropriately.

[0111] In summary, the weld quality monitoring system, by acquiring weld deformation, environmental correction factors, and net energy of sheet metal welding, integrates sheet metal characteristics, environmental attenuation, and nonlinear heat dissipation factors of materials to obtain the probability of weld defects. This improves the accuracy of weld defect probability monitoring, thereby enhancing the accuracy of weld formation quality monitoring. Furthermore, traditional sheet metal welding weld monitoring rarely quantifies airflow interference, while gas velocity directly affects arc stability and molten pool heat dissipation intensity. The weld quality monitoring method, by acquiring the airflow interference index and using it to obtain the probability of weld defects, can also improve the accuracy of weld formation quality monitoring to some extent.

[0112] In one embodiment, the weld quality monitoring module further includes a weld width score acquisition unit, a weld excess height score acquisition unit, and a weld quality score acquisition unit.

[0113] The weld width score acquisition unit is used to obtain the actual weld width and obtain a weld width score based on the actual weld width and the theoretical optimal width; the weld reinforcement height score acquisition unit is used to obtain the actual weld reinforcement height and obtain a weld reinforcement height score based on the actual weld reinforcement height and the theoretical optimal weld reinforcement height; the weld quality score acquisition unit is used to obtain a weld defect score based on the weld defect probability and obtain a sheet metal weld quality score based on the weld width score, weld reinforcement height score, and weld defect score.

[0114] Sheet metal weld quality rating ;in, These represent the actual weld width, the theoretical optimal width, the actual weld reinforcement height, and the theoretical optimal weld reinforcement height, respectively. These are all weighting coefficients, which can be set based on experience; The tolerance variance for excess height is the allowable range of excess height fluctuation. Generally, it is 0.04 for thin plates and 0.12 for thick plates. These represent the weld width score, weld reinforcement score, and weld defect score, respectively.

[0115] In summary, this sheet metal weld quality scoring function model unifies and quantifies macroscopic dimensional control and microscopic defect prediction. It can not only serve as an evaluation index for weld formation quality but also as a basis for real-time adjustment of welding parameters.

[0116] In one embodiment, after obtaining the sheet metal weld quality score, the welding parameters are optimized based on the back gradient of the quality score. Here, the back gradient refers to the partial derivative of the loss function with respect to the parameters. For example, the inverse gradient based on quality scores is expressed as follows: It optimizes weld quality by calculating the gradient of the sheet metal weld quality scoring function with respect to various welding parameters, and then adjusting the parameters along the gradient's ascending direction. Here, "Score" represents the weld quality score; a higher value indicates better quality. This represents the partial derivative of the sheet metal weld quality scoring function with respect to welding parameters.

[0117] Parameter correction amount Here, param represents adjustable process parameters, such as current I, voltage U, welding speed v, gas flow rate Q, etc. The parameter adjustment step size factor is used to prevent overshoot and is generally between 0.01 and 0.1. In other words, the parameter correction amount equals the product of the rate of change in quality score caused by a unit change in the parameter and the parameter adjustment step size factor.

[0118] When the reverse gradient is greater than 0, increase the parameters to improve the quality score; when the reverse gradient is less than 0, decrease the parameters to avoid degradation. Generally, the gradient direction of the welding current is positive, and a step-by-step adjustment strategy is implemented, with a parameter adjustment step size coefficient of 0.05; the gradient direction of the welding voltage is negative, and a smooth decrease adjustment strategy is implemented, with a parameter adjustment step size coefficient of 0.03.

[0119] Traditional single-parameter gradient optimization is prone to failure. For example, in a certain scenario, we might want to increase the desired penetration depth by increasing the welding current. However, increasing the current can easily lead to excessively rapid heat accumulation. To reduce heat accumulation and prevent burn-through, we would then have to increase the current, potentially resulting in a decrease in penetration depth instead of an increase. In other words, single-parameter gradient optimization suffers from coupling conflicts.

[0120] To overcome the limitations of traditional single-parameter optimization and achieve coordinated dynamic correction of multiple parameters such as current I, welding speed v, and gas flow rate Q, thereby resolving parameter coupling conflicts and eliminating multi-parameter interactive interference, multi-parameter coupled gradient optimization can also be performed. For example, Where n represents the number of parameters involved in the coupling. Param i This represents the i-th parameter. This represents the gradient of the quality score with respect to a single parameter.

[0121] If only the three parameters of current I, welding speed v, and gas flow rate Q are dynamically corrected in a coordinated manner, then .in, All are single-parameter gradient vectors. The parameter covariance matrix is ​​a quantification matrix of the coupling relationship between parameters. The diagonal lines correspond to the autovariance, and the off-diagonal lines correspond to the cross-covariance. The inverse covariance matrix can be understood as an operator that eliminates interference from the correlation between parameters. This represents the coupled gradient vector, i.e., the sensitivity of the true parameters after decorrelation, used to guide the coordinated adjustment of multiple parameters.

[0122] The technical features of the embodiments described can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0123] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for monitoring weld quality during sheet metal welding, characterized in that, The weld quality monitoring method includes the following steps: The coefficient of thermal expansion of the sheet metal material, the welding temperature field, and the actual weld length are obtained. Based on the coefficient of thermal expansion, the welding temperature field, and the actual weld length, the weld deformation is obtained to quantify the linear shrinkage of the weld caused by the thermal expansion effect. Get the real-time ambient temperature and humidity, and obtain the environmental correction factor based on the real-time ambient temperature and humidity. The welding arc energy and sheet metal heat dissipation are obtained, and the net sheet metal welding energy is obtained based on the welding arc energy, environmental correction factor and sheet metal heat dissipation to quantify the effective deposition energy during the welding process. Obtain the gas flow rate and then obtain the airflow disturbance index based on the gas flow rate. The probability of weld defects is obtained based on the welding temperature field, weld deformation, net energy of sheet metal welding, and airflow interference index, and the weld formation quality is monitored based on the probability of weld defects.

2. The method for monitoring weld quality in sheet metal welding process as described in claim 1, characterized in that, The specific method for obtaining the probability of weld defects includes the following steps: The curvature of the temperature field is obtained from the welding temperature field, and the net energy change rate is obtained from the net energy of sheet metal welding. Real-time monitoring of sheet metal material thickness; obtaining the thickness-weighted average of the weld area based on the real-time thickness. The shrinkage thickness ratio is obtained by comparing the weld deformation with the thickness-weighted average. A feature matrix is ​​constructed based on the net energy change rate, temperature field curvature, shrinkage thickness ratio, and airflow interference index. The feature matrix is ​​then transformed using a random forest to obtain decision tree feature values. The weld defect probability is then obtained based on the decision tree feature values.

3. The method for monitoring weld quality in sheet metal welding process as described in claim 2, characterized in that, The specific methods for monitoring weld formation quality include the following steps: Obtain the actual weld width, and then obtain a weld width score based on the actual weld width and the theoretical optimal width. Obtain the actual weld reinforcement height, and obtain the weld reinforcement height score based on the actual weld reinforcement height and the theoretical optimal reinforcement height; Weld defect scores are obtained based on weld defect probability, and sheet metal weld quality scores are obtained based on weld width scores, weld reinforcement scores, and weld defect scores.

4. The method for monitoring weld quality in sheet metal welding process as described in claim 3, characterized in that, The weld quality monitoring method also includes the following steps: The partial derivative of the welding current with respect to the quality score is obtained based on the sheet metal weld quality score, and the current score gradient is obtained based on the partial derivative to characterize the amount of adjustment current required to represent the unit quality score change. Obtain the second-order partial derivative of the weld defect probability, and obtain the acceleration of the defect probability change based on the second-order partial derivative; The welding current adjustment amount is obtained based on the current scoring gradient and the acceleration of the defect probability change, and the welding machine current is adjusted according to the welding current adjustment amount.

5. The method for monitoring weld quality in sheet metal welding process as described in claim 4, characterized in that, The specific methods for obtaining welding arc energy and sheet metal heat dissipation include the following steps: The arc voltage, welding current, and thermal efficiency coefficient are obtained, and the welding arc energy is obtained based on the arc voltage, welding current, and thermal efficiency coefficient. The real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material are obtained. The heat dissipation of the sheet metal is obtained based on the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material.

6. The method for monitoring weld quality in sheet metal welding process as described in claim 5, characterized in that, The specific method for obtaining weld deformation includes the following steps: The initial weld deformation is obtained based on the coefficient of thermal expansion, the welding temperature field, and the actual weld length to quantify the linear shrinkage of the weld caused by the thermal expansion effect. The initial weld deformation is corrected based on the environmental correction factor to obtain the final weld deformation.

7. A sheet metal welding process weld quality monitoring system, used to implement the sheet metal welding process weld quality monitoring method as described in any one of claims 1-6, characterized in that, The weld quality monitoring system includes: The deformation acquisition module is used to acquire the thermal expansion coefficient of the sheet metal material, the welding temperature field, and the actual weld length. Based on the thermal expansion coefficient, the welding temperature field, and the actual weld length, it acquires the weld deformation amount used to quantify the linear shrinkage of the weld caused by the thermal expansion effect. The correction factor acquisition module is used to acquire the real-time ambient temperature and humidity, and to acquire the environmental correction factor based on the real-time ambient temperature and humidity. The net energy acquisition module is used to acquire the welding arc energy and sheet metal heat dissipation. Based on the welding arc energy, environmental correction factor and sheet metal heat dissipation, it acquires the sheet metal welding net energy to quantify the effective deposition energy during the welding process. The interference index acquisition module is used to acquire the gas flow rate and obtain the airflow interference index based on the gas flow rate. The weld quality monitoring module is used to obtain the probability of weld defects based on the welding temperature field, weld deformation, net energy of sheet metal welding, and airflow interference index, and to monitor the weld formation quality based on the probability of weld defects.

8. The sheet metal welding process weld quality monitoring system as described in claim 7, characterized in that, The weld quality monitoring module includes: The net energy change rate acquisition unit is used to obtain the temperature field curvature based on the welding temperature field and the net energy change rate based on the net energy of sheet metal welding. The thickness weighting value acquisition unit is used to monitor the real-time thickness of sheet metal materials and obtain the thickness weighting average value of the weld area based on the real-time thickness. The shrinkage thickness ratio acquisition unit is used to obtain the shrinkage thickness ratio based on the ratio between the weld deformation and the thickness-weighted average value. The feature matrix construction unit is used to construct a feature matrix based on the net energy change rate, temperature field curvature, shrinkage thickness ratio, and airflow disturbance index. The defect probability acquisition unit is used to perform feature transformation on the feature matrix based on random forest, obtain decision tree feature values, and obtain the weld defect probability based on the decision tree feature values.

9. The sheet metal welding process weld quality monitoring system as described in claim 8, characterized in that, The weld quality monitoring module also includes: The weld width score acquisition unit is used to obtain the actual weld width and to obtain the weld width score based on the actual weld width and the theoretical optimal width. The weld reinforcement height score acquisition unit is used to obtain the actual weld reinforcement height and obtain the weld reinforcement height score based on the actual weld reinforcement height and the theoretical optimal reinforcement height. The weld quality scoring unit is used to obtain weld defect scores based on weld defect probability, and to obtain sheet metal weld quality scores based on weld width scores, weld reinforcement scores, and weld defect scores.

10. The sheet metal welding process weld quality monitoring system as described in claim 9, characterized in that, The net energy change rate acquisition unit obtains the formula based on... Obtain the net rate of change of energy; in, Indicates the net energy consumption of sheet metal welding. These represent the arc voltage, welding current, and thermal efficiency coefficient, respectively. Indicates the environmental correction factor. These represent the real-time temperature of the weld pool, the initial temperature of the substrate before welding, and the thermal conductivity of the sheet metal material, respectively.

Images