Intelligent welding machine operation parameter control method and system

Abstract

The present application belongs to the technical field of welding machine control, and particularly relates to an intelligent welding machine operation parameter control method and system, which comprises the following steps: collecting a temperature field image of a welding area in real time, constructing and estimating a state vector representing the geometric shape and heat distribution of a heat affected zone based on extended Kalman filtering, constructing an ideal temperature field image based on optimal state estimation, performing spatial anomaly evaluation according to a residual temperature field between the actual temperature field image and the ideal temperature field image, obtaining a spatial anomaly index representing the severity of welding anomalies, fusing the optimal state estimation and the spatial anomaly index, dynamically adjusting welding parameters, and realizing control over the welding process. The present application improves the welding quality stability and process reliability.

Description

Technical Field

[0001] This invention relates to the field of welding machine control technology. More specifically, this invention relates to an intelligent welding machine operating parameter control method and system. Background Technology

[0002] In modern manufacturing, welding technology is a key process for joining metal materials and is widely used in aerospace, automotive manufacturing, power equipment and other fields. During the welding process, due to the movement of the welding torch, heat accumulates in the forward direction and cools rapidly in the backward direction. This dynamic heat conduction characteristic causes the heat-affected zone (HAZ) to typically appear as an elongated ellipse along the direction of the welding torch's movement, rather than a circular distribution as in the case of a stationary heat source. This elliptical heat-affected zone is a common feature of all moving welding processes.

[0003] Currently, the welding control methods commonly used in industry mainly rely on empirical parameter setting and macroscopic parameter monitoring. They are usually operated by fixing process parameters such as welding current, voltage and travel speed. Some systems are equipped with infrared thermal imagers to monitor temperature distribution, but this is limited to simple feedback control of macroscopic parameters such as peak temperature and molten pool size.

[0004] Existing methods focus only on macroscopic heat distribution while neglecting subtle changes in the geometry of the heat-affected zone (HAZ), lacking the ability to dynamically assess abnormal development trends and thus failing to intervene in time before defects form. For example, the presence of a tiny oxide layer or oil on the material surface can lead to abnormal local heat absorption. Although the overall peak temperature may still be within the normal range, the elliptical contour of the HAZ will be distorted or locally deformed. This subtle morphological change is a precursor to welding defects. However, traditional monitoring methods struggle to detect such anomalies, resulting in a persistently high incidence of defects such as cracks and porosity during welding. In actual production, this often necessitates additional post-weld inspection and rework to ensure product quality, significantly increasing production costs and timelines.

[0005] Therefore, there is an urgent need for a method and system for controlling the operating parameters of an intelligent welding machine. Summary of the Invention

[0006] To address the technical problem of high defect rates caused by the difficulty in monitoring changes in the geometric shape of the heat-affected zone during the welding process, the inability of existing technologies to effectively identify early signs of welding defects and intervene in advance, this invention provides solutions in the following aspects.

[0007] In a first aspect, the present invention provides a method for controlling the operating parameters of an intelligent welding machine, comprising:

[0008] The system acquires real-time temperature field images of the welding area; constructs and estimates a state vector characterizing the geometry and heat distribution of the heat-affected zone (HAZ) based on extended Kalman filtering. This state vector includes the coordinates of the geometric center of the HAZ, the semi-major axis, semi-minor axis, and rotation angle of the fitted ellipse describing the outer contour of the HAZ, the peak temperature of the welding area, and the average normal temperature gradient at the boundary of the HAZ. An ideal temperature field image is constructed based on the optimal state estimation. Spatial anomaly assessment is performed using the residual temperature field between the actual and ideal temperature field images to obtain spatial anomaly indicators characterizing the severity of welding anomalies. Finally, the optimal state estimation and spatial anomaly indicators are fused to dynamically adjust welding parameters, thereby achieving control of the welding process.

[0009] Its effects are as follows: This invention, by acquiring real-time temperature field images of the welding area, can accurately capture the asymmetric temperature distribution characteristics caused by the movement of the welding torch during the welding process; Based on extended Kalman filtering, this invention constructs and estimates a state vector containing the elliptical geometry and heat distribution characteristics of the heat-affected zone, overcoming the strong nonlinear dynamic characteristics of the welding process and providing stable and reliable state estimation, laying the foundation for characterizing the morphology of the heat-affected zone; Based on optimal state estimation, this invention reconstructs an ideal temperature field image and performs residual analysis with the actual temperature field image, effectively identifying local heat absorption anomalies caused by material surface contamination or oxide layers, achieving early warning of welding defects; Through comprehensive evaluation of the threat weight of the residual region location and the dynamic change rate, this invention can reflect the severity and development trend of anomalies; This invention uses the state vector as the input of the main control loop to adjust the welding current to control the macroscopic heat distribution, and simultaneously dynamically adjusts the walking speed according to spatial anomaly indicators to suppress local thermal disturbances, achieving dual precise control of the welding process, improving welding quality, reducing the generation of defects such as cracks and porosity, and ensuring the reliability and stability of the welding process.

[0010] Preferably, the method for obtaining the residual temperature field is as follows: subtract the actual temperature field image at each moment from the ideal temperature field image at that moment to obtain the residual temperature field at that moment.

[0011] Preferably, the step of assessing spatial anomalies based on the residual temperature field between the actual temperature field image and the ideal temperature field image to obtain spatial anomaly indicators characterizing the severity of welding anomalies includes: obtaining residual regions in the residual temperature field; for each residual region, determining the positional threat weight of the residual region based on the cosine of the angle between the centroid vector of the residual region and the welding torch travel speed vector; determining the dynamic rate of change of the residual region based on the area change and residual temperature change of the residual region at adjacent time points; and obtaining spatial anomaly indicators based on the positional threat weight, area, residual temperature, and dynamic rate of change of all residual regions in the residual temperature field.

[0012] Its effects are as follows: By analyzing the residual region in the residual temperature field, this invention can identify local thermal anomalies during the welding process. By calculating the relative positional relationship between the centroid vector of the residual region and the direction of the welding torch, it can assess the threat level of residual regions at different locations to weld quality, thus giving higher attention to anomaly regions located in front of the molten pool. By monitoring the area and temperature change trends of the residual region over time, this invention can capture the dynamic development of anomalies, distinguishing whether the anomaly is expanding or gradually weakening, thereby giving higher attention to rapidly developing anomalies. This invention constructs a spatial anomaly index by integrating positional threat weight, residual region area, residual temperature, and dynamic change rate. This index not only reflects the severity of the current anomaly but also embodies the development trend and spatial distribution characteristics of the anomaly, enabling the system to identify potential risks that may lead to defects such as cracks and porosity in advance, and to intervene in time before defects form. This improves the quality stability and process reliability of welding, and realizes comprehensive and multi-dimensional quality monitoring of the welding process.

[0013] Preferably, obtaining the residual region in the residual temperature field includes: performing Otsu threshold segmentation on the residual temperature field to obtain a binary residual image, performing connected component analysis on the binary residual image, and taking each obtained connected component as a residual region.

[0014] Preferably, the direction of the centroid vector of the residual region is from the center of the molten pool to the centroid of the residual region, and the modulus is the Euclidean distance between the centroid of the residual region and the center of the molten pool; the direction of the welding torch travel speed vector is the welding forward direction, and the modulus is the travel speed magnitude.

[0015] Preferably, the location threat weights satisfy the expression: In the formula, for The first moment Location threat weights for each residual region; for The first moment The cosine of the angle between the centroid vector of the residual region and the welding torch travel speed vector; It is a natural exponential function.

[0016] The effect is as follows: During the welding process, the abnormal area in front of the molten pool has a significant impact on the weld quality because it is in an unsolidified state, and is very likely to cause crack formation. On the other hand, the abnormal area behind the molten pool has a relatively small impact on the weld quality because it has been cooled. This invention utilizes the characteristics of the exponential function to give the residual area located in the direction of the welding torch forward a greater positional threat weight, while the residual area located behind the welding torch is given a smaller positional threat weight. This reduces the impact of the residual area behind the welding torch on the overall evaluation, so that the subsequent spatial anomaly index can more accurately reflect the abnormal area that truly threatens the welding quality.

[0017] Preferably, the dynamic rate of change of the residual region satisfies the expression: In the formula, for The first moment The dynamic rate of change of each residual region; for The first moment The area of ​​each residual region; for The first moment The average residual temperature of each residual region; for The first moment Each residual region in The area of ​​the residual region in the residual temperature field at time t; for The first moment Each residual region in The average residual temperature of the corresponding residual region in the residual temperature field at time t; For time step.

[0018] Its effect is that during the welding process, when there is a tiny oxide layer or contamination on the material surface, it will cause local heat absorption abnormalities. These abnormalities usually do not disappear instantly, but gradually expand or weaken over time. This invention tracks the evolution of the same residual region in the time dimension and uses the dynamic rate of change to reflect the development rate of the abnormality, enabling the control system to take corresponding response strategies according to the development trend of the abnormality, thereby improving the early warning capability and control accuracy of potential defects in the welding process.

[0019] Preferably, the spatial anomaly index satisfies the expression: In the formula, for Spatial anomaly indicators at any given time; for The first moment Location threat weights for each residual region; for The first moment The area of ​​each residual region; for The first moment The average residual temperature of each residual region; for The first moment The dynamic rate of change of each residual region; It is the hyperbolic tangent function; express The number of residual regions at each time step; This represents the maximum area of ​​all residual regions across all historical welding tasks involving the same material. This represents the maximum average residual temperature across all residual regions in all historical welding tasks involving the same material.

[0020] Its effects are as follows: This invention achieves multi-dimensional evaluation of local thermal disturbances during the welding process through spatial anomaly indicators, enabling the control system to identify potential risks that may lead to defects such as cracks and porosity in advance, improving the early warning capability and response accuracy of microscopic anomalies during the welding process, and providing a basis for subsequent adjustment of welding parameters.

[0021] Preferably, the step of integrating the optimal state estimate and spatial anomaly index to dynamically adjust welding parameters includes: using the optimal state estimate as the input to the main control loop, and calculating the base current adjustment amount through a PID controller. Spatial anomaly indicators Input to a threshold comparator: in response to Then a proportional to walking speed increment ,otherwise ;Will and The outputs are respectively sent to the welding power supply and the motor driver to achieve synchronous adjustment of the welding current and the travel speed. This is the abnormal threshold.

[0022] The advantages are as follows: The main control loop of this invention, based on the peak temperature and heat-affected zone size estimated in the optimal state, adjusts the welding current through a PID controller to ensure that the weld penetration and width meet the process requirements, thereby achieving stable control of the macroscopic heat distribution. Simultaneously, the auxiliary control loop intelligently generates corresponding travel speed increments based on the comparison results of spatial anomaly indicators and preset thresholds. When a local thermal anomaly is detected, the travel speed is moderately increased to shorten the dwell time of the welding torch in the abnormal area, thereby suppressing local overheating. The control method of this invention can maintain the macroscopic stability of the welding process while responding promptly to microscopic anomalies, avoiding the limitations of traditional welding control methods that only focus on macroscopic parameters and ignore local anomalies. It can effectively reduce the generation of defects such as cracks and porosity, ensuring the safety and reliability of the welding process and the stability of product quality.

[0023] Secondly, the present invention provides an intelligent welding machine operating parameter control system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned intelligent welding machine operating parameter control method is implemented.

[0024] By adopting the above technical solution, a computer program is generated from the above-mentioned intelligent welding machine operation parameter control method and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0025] The beneficial effects of this invention are as follows: By acquiring real-time temperature field images of the welding area, this invention can accurately capture the asymmetric temperature distribution characteristics caused by the movement of the welding torch during the welding process; based on extended Kalman filtering, this invention constructs and estimates a state vector containing the elliptical geometry and heat distribution characteristics of the heat-affected zone, overcoming the strong nonlinear dynamic characteristics of the welding process and providing stable and reliable state estimation, laying the foundation for characterizing the morphology of the heat-affected zone; based on optimal state estimation, this invention reconstructs an ideal temperature field image and performs residual analysis with the actual temperature field image, effectively identifying local heat absorption anomalies caused by material surface contamination or oxide layers, achieving early warning of welding defects; through comprehensive evaluation of the threat weight of the residual region location and the dynamic change rate, this invention can reflect the severity and development trend of anomalies; this invention uses the state vector as the input of the main control loop to adjust the welding current to control the macroscopic heat distribution, and simultaneously dynamically adjusts the walking speed according to spatial anomaly indicators to suppress local thermal disturbances, achieving dual control of the welding process, improving welding quality, reducing the generation of defects such as cracks and porosity, and ensuring the reliability and stability of the joint process. Attached Figure Description

[0026] Figure 1This is a flowchart illustrating a method for controlling the operating parameters of an intelligent welding machine according to the present invention;

[0027] Figure 2 This is a schematic diagram of the temperature field.

[0028] Figure 3 This is a schematic diagram of a binary image of the temperature field.

[0029] Figure 4 This is a schematic diagram of the fitted ellipse. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0032] This invention discloses a method for controlling the operating parameters of an intelligent welding machine, referring to... Figure 1 This includes steps S1-S4:

[0033] S1. Real-time acquisition of temperature field images of the welding area.

[0034] It should be noted that during the welding process, due to the movement of the welding torch, heat accumulates in the forward direction and cools rapidly in the backward direction, forming an uneven temperature distribution. This unevenness changes dynamically with the welding process. Therefore, this invention acquires temperature field images of the welding area in real time, thereby capturing the instantaneous temperature distribution changes at the interface caused by the movement of the welding torch.

[0035] Specifically, an infrared thermal imager with a frame rate of no less than 50Hz and a temperature measurement range covering the melting point of the material is selected. Its optical axis is kept at a fixed relative position and angle with the welding torch to acquire temperature field images in real time. For example, Figure 2 This is a schematic diagram of the temperature field.

[0036] Optionally, for each acquired temperature field image frame, a Gaussian filtering operation is performed using a 5×5 Gaussian filter kernel convolution. It should be noted that Gaussian filtering can suppress random noise in the temperature field image while effectively preserving edge features, providing a foundation for subsequent heat-affected zone contour recognition.

[0037] S2. Construct and estimate a state vector characterizing the geometry and heat distribution of the heat-affected zone based on extended Kalman filtering. The state vector includes the coordinates of the geometric center of the heat-affected zone, the major semi-axis, minor semi-axis and rotation angle of the fitting ellipse of the isotherm describing the outer contour of the heat-affected zone, the peak temperature of the welding area, and the average normal temperature gradient at the boundary of the heat-affected zone.

[0038] It should be noted that the heat-affected zone (HAZ) refers to the region where the material is heated but not melted during welding. This region undergoes a thermal cycle from room temperature to near the melting point and then cooling, resulting in significant changes in the material's microstructure and mechanical properties. The morphology and temperature distribution of the HAZ directly determine the quality and performance of the weld joint. During welding, the shape and temperature distribution of the HAZ are influenced by various factors, including welding current, speed, and the material's thermal conductivity. These factors are coupled and exhibit strong nonlinearity. Therefore, this invention employs the Extended Kalman Filter (EKF) algorithm. By linearizing the nonlinear model near the current estimate, it achieves optimal state estimation of the nonlinear system. While maintaining computational efficiency, it effectively handles the nonlinear dynamic characteristics during welding, providing a foundation for subsequent residual analysis. To comprehensively describe the macroscopic geometric and thermal characteristics of the HAZ while maintaining a low-dimensional state vector to meet real-time requirements, this invention does not use discrete key temperature points as states. Instead, it uses a set of geometric and physical parameters that describe the core morphology of the HAZ as the state vector.

[0039] Specifically, define The state vector at time t is:

[0040]

[0041] in, for The state vector at time t, The coordinates of the geometric center of the heat-affected zone; These are the major and minor axes of the fitted ellipse of the isotherm describing the outer contour of the heat-affected zone, respectively. The rotation angle of the fitted ellipse for the isotherm describing the outer contour of the heat-affected zone; This represents the peak temperature of the welding area. This represents the average normal temperature gradient at the boundary of the heat-affected zone. This indicates the transpose symbol. The logical relationship of this formula is that the overall shape characteristics of the heat-affected zone are described by the elliptic parameters, and the heat distribution characteristics of the heat-affected zone are described by the peak temperature and the normal temperature gradient, together forming a 7-dimensional state vector. When the heat-affected zone expands during the welding process, and Increase; when the shape of the heat-affected zone changes from circular to elliptical. and The ratio increases; when the heat-affected zone shifts as a whole in a certain direction... and Changes occur; when the welding heat input increases. Increase Decrease.

[0042] Furthermore, to obtain The observation vector at time t includes:

[0043] Traversal The temperature values ​​of all pixels in the temperature field image at any given time are used to determine the maximum value, which is then taken as the observed peak temperature of the welding area. A temperature threshold is set, and the temperature field image is binarized to obtain a binary image. Connectivity analysis is performed on the binary image, and the boundary of the largest connected component in the binary image is taken as the outer contour of the heat-affected zone. An ellipse is fitted to the outer contour of the heat-affected zone, and the center coordinates, semi-major axis, semi-minor axis, and rotation angle of the fitted ellipse are taken as the observed values ​​of the coordinates of the geometric center of the heat-affected zone. Observed values ​​of the semi-major axis of the fitted ellipse describing the outer contour of the heat-affected zone. Observations of the minor semi-axis Observed values ​​of rotation angle On the outer contour of the heat-affected zone, the temperature gradient of each pixel is calculated along the normal direction of each pixel. The average temperature gradient of all pixels on the outer contour of the heat-affected zone is taken as the observed value of the average normal temperature gradient at the boundary of the heat-affected zone. The above observations are used to construct the observation vector at time t. In this invention, the temperature threshold is set to 85% of the solidus temperature of the material. This temperature threshold can significantly distinguish the temperature of the heat-affected zone from that of the parent material, so that the segmented binary image can clearly distinguish the heat-affected zone from the parent material region, providing an accurate basis for subsequent contour extraction, while effectively suppressing the influence of image noise.

[0044] For example, Figure 3 This is a schematic diagram of a binary image of the temperature field. Figure 4 This is a schematic diagram of the fitted ellipse.

[0045] Furthermore, the observation vector The input is fed into the update module of the extended Kalman filter, and the optimal estimated state at the current time is obtained through iterative calculation based on the prediction and update equations of the standard extended Kalman filter. :

[0046]

[0047]

[0048] In the formula, for The optimal estimated state vector at time t; for The optimal estimated state vector at time t; for The observation vector at time; for The predicted state vector at time step; for The control input vector at time t, , For welding current, Arc voltage The moving speed of the welding torch; , For welding torch in and velocity in the direction; Kalman gain; For time step.

[0049] The system state transition function, constructed based on the heat conduction physical model, includes the following components:

[0050] Position components:

[0051]

[0052]

[0053] In the formula, for The coordinates of the geometric center of the thermally affected zone at any given time; for The coordinates of the geometric center of the thermally affected zone at any given time; For welding torch in and velocity in the direction; The time step is [time step]. The geometric center of the heat-affected zone changes as the welding torch moves, and its moving speed is consistent with the welding torch speed. When moving in direction, Geometric center of the heat-affected zone coordinate The increase is due to this; when the welding torch is directed towards When moving in direction, Geometric center of the heat-affected zone coordinate It increases accordingly.

[0054] Ellipse parameters:

[0055]

[0056]

[0057]

[0058] In the formula, for The major and minor axes of the fitted ellipse of the isotherm that describes the outer contour of the heat-affected zone at any given moment. for The major and minor axes of the fitted ellipse of the isotherm that describes the outer contour of the heat-affected zone at any given moment. for Time and The rotation angle of the fitted ellipse that describes the outer contour of the heat-affected zone at any given moment. These are the sensitivity coefficients of the long semi-axis, short semi-axis, and rotation angle of the heat-affected zone to the welding current, respectively. This refers to the welding current. The reference current; The moving speed of the welding torch; The time step is [not specified]. The changes in the major and minor semi-axis are proportional to the changes in welding current. Greater than the reference current hour, Dimensions of the heat-affected zone and Increase; when welding current Less than the reference current hour, Dimensions of the heat-affected zone and The rotation angle decreases because the welding current directly affects the heat input; an increase in current leads to an increase in heat input and an expansion of the heat-affected zone. The change in rotation angle is directly proportional to the welding torch's moving speed. When it increases, The rate of change increases because during the welding process, as the welding torch moves, heat accumulates in the forward direction, causing the heat-affected zone to become elliptical. The faster the speed, the greater the change in the angle between the major axis of the ellipse and the forward direction of the welding torch.

[0059] Peak temperature:

[0060]

[0061] In the formula, for Time and Peak temperature of the welding area at any given time; This is the sensitivity coefficient of peak temperature to heat input; This refers to the welding current. It is the arc voltage; This is due to heat loss; The time step is [not specified]. The change in peak temperature is proportional to the net heat input (heat input minus heat loss). Indicates the welding heat input power. This indicates the heat loss per unit time. This represents the net heat input. When the net heat input is greater than 0, the peak temperature increases; when the net heat input is less than 0, the peak temperature decreases. During welding, as the welding current and voltage increase, the heat input increases, and the peak temperature rises; when the material dissipates heat more quickly, the peak temperature decreases. As the temperature increases, the rate of increase in peak temperature slows down.

[0062] Mean normal temperature gradient:

[0063]

[0064] In the formula, for The average normal temperature gradient at the boundary of the heat-affected zone at any given moment; for Peak temperature of the welding area at any given time; The ambient temperature; for The semi-major axis of the fitted ellipse of the isotherms that describe the outer contour of the heat-affected zone at any given time. In heat conduction theory, the temperature gradient is inversely proportional to the size of the heat-affected zone; the larger the size of the heat-affected zone, the smaller the temperature gradient. The higher the peak temperature, the larger the temperature gradient. Assuming that the temperature decreases linearly from the peak temperature to the ambient temperature, the average normal temperature gradient is approximately equal to the difference between the peak temperature and the ambient temperature divided by the size of the heat-affected zone.

[0065] It should be noted that, These are the sensitivity coefficients of the long semi-axis, short semi-axis, rotation angle, and peak temperature of the heat-affected zone to the welding current, respectively. They are obtained by conducting multiple welding experiments, keeping other process parameters constant, and only changing the welding current. The morphological parameters of the heat-affected zone (HAZ) under different currents were recorded. These parameters included the major and minor axes, rotation angle, and peak temperature of the HAZ. The experimental data were fitted using the least squares method to establish a linear relationship between the welding current and the HAZ morphological parameters. The slope of the fitted linear relationship divided by the time step was used as the sensitivity coefficient of the HAZ morphological parameters to the welding current.

[0066] The reference current is determined as follows: through welding procedure qualification experiments, the current is gradually increased until a weld that meets the basic penetration depth requirements is obtained, and the current value at this point is recorded as the reference current. .

[0067] Heat loss includes radiation loss, convection loss and conduction loss, and is calculated using the heat balance equation.

[0068] Let be the observation function, satisfying:

[0069] in, for The observation vector at time; For observation noise: , These are, respectively, location observation noise, elliptic parameter observation noise, peak temperature observation noise, and normal temperature gradient observation noise.

[0070] It should be noted that during the welding process, due to the movement of the welding torch, heat accumulates in the forward direction and cools rapidly in the backward direction. This dynamic heat conduction characteristic causes the heat-affected zone to typically appear as an elongated ellipse along the forward direction of the welding torch. Therefore, this invention uses ellipse fitting to describe the main geometric characteristics of the heat-affected zone.

[0071] S3. Construct an ideal temperature field image based on the optimal state estimation, and conduct spatial anomaly assessment based on the residual temperature field between the actual temperature field image and the ideal temperature field image to obtain spatial anomaly indicators characterizing the severity of welding anomalies.

[0072] It should be noted that during welding, the heat-affected zone typically exhibits an elliptical distribution. Furthermore, local anomalies during welding usually do not disappear instantly but gradually expand or weaken over time. For example, the presence of a micro-oxide layer on the material surface can lead to abnormal localized heat absorption, which gradually increases with the welding process. Therefore, this invention reconstructs an ideal temperature field image through optimal state estimation, analyzes the residual temperature field between the actual and ideal temperature field images, thereby determining the severity and potential threat of the anomaly and intervening before defects form.

[0073] Specifically, based on the optimal state estimation vector at each moment, an ideal temperature field image is reconstructed. The ideal temperature field image perfectly conforms to the elliptical geometric model and temperature distribution law described by the optimal state estimation vector. The residual temperature field at that moment is obtained by subtracting the actual temperature field image collected at each moment from the ideal temperature field image reconstructed at that moment.

[0074] It should be noted that the residual temperature field reflects the difference between the actual temperature distribution and the ideal temperature model during welding. When the actual temperature is exactly the same as the ideal temperature, the residual value is 0; when the actual temperature deviates from the ideal temperature, the residual value increases. During welding, due to the complex heat conduction characteristics at the material interface, the residual temperature field can effectively reveal local hot spots and temperature distortions that are difficult to detect using traditional methods. For example, it can reveal micropores or crack precursors that may form at the interface. These anomalies often appear as bright areas in the residual temperature field.

[0075] Furthermore, the current travel speed vector of the welding torch is read in real time from the welding machine control system. Its direction represents the welding forward direction, and the die length represents the travel speed.

[0076] The observed values ​​of the geometric center of the heat-affected zone at the current moment. Using the coordinates of the molten pool center as a reference, the residual temperature field at the current moment is segmented using the Otsu threshold method to obtain a binary residual image. Connectivity analysis is then performed on the binary residual image, and each resulting connected component is treated as a residual region. A centroid vector is constructed for each residual region, with its direction pointing from the molten pool center to the centroid of the residual region, and its magnitude being the Euclidean distance between the centroid of the residual region and the molten pool center.

[0077] For each residual region, the positional threat weight of the residual region is determined based on the cosine of the angle between the centroid vector of the residual region and the welding torch travel speed vector:

[0078]

[0079] In the formula, for The first moment Location threat weights for each residual region; for The first moment The cosine of the angle between the centroid vector of the residual region and the welding torch travel speed vector; It is a natural exponential function. During welding, anomalies in front of the molten pool can cause cracks in the unsolidified weld, while anomalies in the cooling zone have a smaller impact on weld quality. Therefore, when the residual region is located in the welding forward direction, the residual region is the region in front of the molten pool. At this time, the angle between the centroid vector of the residual region and the welding torch travel speed vector is 0°, making... , The residual region poses the greatest threat; when the residual region is located in the welding initiation direction, it is a cooled region, and the angle between the centroid vector of the residual region and the welding torch travel speed vector is 180°, making... , The residual region poses a relatively low threat level.

[0080] Furthermore, based on the area change and temperature change of the residual region at adjacent time points, the dynamic rate of change of the residual region is determined:

[0081]

[0082] In the formula, for The first moment The dynamic rate of change of each residual region; for The first moment The area of ​​each residual region; for The first moment The average residual temperature of each residual region; for The first moment Each residual region in The area of ​​the residual region in the residual temperature field at time t; for The first moment Each residual region in The average residual temperature of the corresponding residual region in the residual temperature field at time t; This represents the time step. The first moment Each residual region in The method for obtaining the residual region corresponding to the residual temperature field at time t is as follows: based on The location of the centroid of the residual region at any given time. Search for a circular region centered at the centroid of the residual region and with a radius equal to the welding speed multiplied by the time step within the residual temperature field at time step 2. Find the region with the largest area within this circular region that is also the same as the time step. The connected region with the highest shape similarity to the residual region at any given time is taken as the corresponding residual region. The shape similarity is obtained by the Hausdorff distance. The smaller the Hausdorff distance, the higher the shape similarity.

[0083] In the formula, Reflects the first Thermal anomaly intensity and dynamic rate of change of each residual region This represents the rate of change of the intensity of thermal anomalies in the residual region per unit time. During welding, the presence of contamination or oxide layers on the material surface can lead to localized abnormal heat absorption. These anomalies typically expand gradually over time, hence the dynamic rate of change. This reflects an abnormal development trend; the larger the value, the faster the abnormal development. When, it indicates that the abnormality is expanding, when When the time is right, it indicates that the abnormality is weakening.

[0084] Furthermore, based on the location threat weights, areas, residual temperatures, and dynamic rates of change of all residual regions in the residual temperature field at the current moment, spatial anomaly indicators are obtained for the current moment:

[0085]

[0086] In the formula, for Spatial anomaly indicators at any given time; for The first moment Location threat weights for each residual region; for The first moment The area of ​​each residual region; for The first moment The average residual temperature of each residual region; for The first moment The dynamic rate of change of each residual region; The hyperbolic tangent function is used to normalize the dynamic rate of change to... Within the range; express The number of residual regions at each time step; This represents the maximum area of ​​all residual regions across all historical welding tasks involving the same material. This represents the maximum average residual temperature across all residual regions in all historical welding tasks involving the same material. Used for Normalize.

[0087] When there is contamination or an oxide layer on the surface of a material, it can lead to abnormal localized heat absorption. This reflects the intensity of the abnormal thermal anomaly. This reflects the severity of the anomaly, and its value range is [value range missing]. When the anomaly expands relative to the previous moment, It is positive at this time. The greater the expansion, the more severe the anomaly, and the more immediate the intervention required; when the anomaly weakens relative to the previous moment, It is negative at this time. As the weakening effect increases, the anomalous impact gradually decreases, and the threat level reduces. It should be noted that in the gamma transform, when the base is at... When the index is within the range, Within the range, the result of the gamma transform is greater than or equal to the base, and when the exponent is... The smaller the range, the larger the result of the gamma transform; when the base is in the range... When the index is within the range, When the exponent is within the range, the result of the gamma transform is less than or equal to the base, and when the exponent is within the range... The larger the value of the inner value, the smaller the result of the gamma transform. Therefore, this invention uses the form of gamma transform to... As the base, As an index, to obtain abnormal indicators ,when When it is the right time, ,right Increase the size to improve the sensitivity of abnormal indicators, and when When it is larger, exist The smaller the range, the better. The greater the increase; when When it is negative, ,right To reduce the sensitivity of abnormal indicators, and when The smaller, exist The larger the range, the better. The greater the reduction; when hour, , At this point, the abnormal indicators only reflect the severity of the current abnormality.

[0088] In the formula, the positional threat weight This invention reflects the potential impact of the position of the residual region relative to the welding advance direction on weld quality. The further forward the residual region is, the greater its impact on the unsolidified weld, potentially leading to crack formation. The invention incorporates a comprehensive positional threat weighting. and abnormal indicators To obtain spatial anomaly indicators, the value of spatial anomaly indicators is greater when the residual region is large, the location is dangerous, and the changes are rapid.

[0089] S4. By integrating optimal state estimation and spatial anomaly indicators, welding parameters are dynamically adjusted to achieve control over the welding process.

[0090] It should be noted that welding quality depends simultaneously on the accuracy of macroscopic heat distribution and the effective suppression of microscopic local thermal disturbances; a single control logic cannot achieve both. During welding, due to the significant differences in the thermophysical properties of the two materials, it is necessary to simultaneously control the overall morphology of the heat-affected zone to ensure sufficient weld depth and width, while also promptly suppressing local hot spots to prevent crack and porosity formation. Therefore, this invention adjusts the welding machine operating parameters based on optimal state estimation and spatial anomaly indicators.

[0091] Specifically, the optimal state estimate As the input to the main control loop, the base current regulation is calculated by the PID controller. Spatial anomaly indicators Input to a threshold comparator: If Then a proportional to walking speed increment ;otherwise .Will and The current is output to the welding power supply and the motor driver respectively, so as to realize the synchronous adjustment of welding current and walking speed.

[0092] It should be noted that, The abnormal threshold is determined as follows: First, multiple welding experiments are conducted, and spatial anomaly indicators during the welding process are recorded. Time series and final weld quality; analysis of welding defects and The correspondence between values ​​is used to determine the critical value that leads to a significant increase in welding defects. Based on the system response characteristics and control stability requirements, Set as The system should be kept at 80%-90% of its capacity to ensure timely response before defects occur, while avoiding frequent adjustments due to oversensitivity.

[0093] For example, in welding experiments, statistical analysis of a large amount of experimental data revealed that when spatial anomaly indicators... When the weld strength exceeds 4.5, the weld crack rate and porosity increase significantly, indicating that 4.5 is a critical value that leads to a significant increase in welding defects. To avoid the system from being too sluggish, 80% of this critical value is used as the anomaly threshold. ,Right now In practical applications, The value can be adjusted according to specific welding conditions and quality requirements. Fine-tune within the range to achieve the best control effect.

[0094] It should be noted that when spatial anomaly indicators exceed the anomaly threshold... When this occurs, it indicates a local anomaly requiring attention. Increasing the walking speed at this point can shorten the time the welding torch remains in that area, reducing heat input and effectively suppressing local overheating. The increase in walking speed is related to... Proportional, the more severe the abnormality, The larger the size, the greater the increase in walking speed.

[0095] It should be further explained that the increase in walking speed is related to... proportionality coefficient The determination method is as follows: First, conduct multiple sets of welding experiments and record the spatial anomaly indicators during the welding process. With walking speed increment The correspondence and final weld quality; analysis of welding defects and The correspondence between values ​​is used to determine the critical value that causes excessive fluctuations in walking speed. Then, based on the system response characteristics and control stability requirements, Set as The percentage is 50%-70% to ensure that the system can suppress local anomalies while avoiding uneven melting depth caused by excessive changes in walking speed.

[0096] For example, in welding experiments, statistical analysis of a large amount of experimental data revealed that when the proportionality coefficient... When the value exceeds 0.002, the walking speed fluctuation exceeds 30% of the base walking speed, resulting in a melting depth fluctuation exceeding 15%; while when When the value is less than 0.0005, the system's response to local anomalies is insufficient, and the crack rate exceeds 5%. This indicates that 0.002 is the critical value that causes excessive fluctuations in walking speed. To balance the anomaly suppression effect with system stability, 50% of this critical value is taken as... Value, that is In practical applications, The value can be adjusted according to specific welding conditions and quality requirements. Fine-tune within the range to achieve the best control effect.

[0097] This invention also discloses an intelligent welding machine operating parameter control system, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent welding machine operating parameter control method according to the present invention is implemented.

[0098] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method of controlling operating parameters of an intelligent welding machine, characterized by, The method comprises the following steps: real-time acquisition of temperature field images of a welding area; construction and estimation of a state vector representing the geometric shape and heat distribution of a heat-affected zone based on an extended Kalman filter, the state vector including the geometric center coordinates of the heat-affected zone, the major semi-axis, minor semi-axis and rotation angle of a fitted ellipse of isotherms describing the outer contour of the heat-affected zone, the peak temperature of the welding area and the average normal temperature gradient at the boundary of the heat-affected zone; construction of an ideal temperature field image based on optimal state estimation, spatial anomaly evaluation based on the residual temperature field between the actual temperature field image and the ideal temperature field image, and acquisition of a spatial anomaly index representing the severity of welding anomalies, including: acquisition of residual areas in the residual temperature field; for each residual area, determining the position threat weight of the residual area according to the cosine value of the included angle between the centroid vector of the residual area and the welding torch walking speed vector; determining the dynamic change rate of the residual area according to the area change and residual temperature change of the residual area at adjacent time; acquiring the spatial anomaly index according to the position threat weight, area, residual temperature and dynamic change rate of all residual areas in the residual temperature field; fusing the optimal state estimation and the spatial anomaly index to dynamically adjust the welding parameters and realize the control of the welding process; wherein the residual temperature field is subjected to Otsu threshold segmentation to obtain a residual binary image, and the residual binary image is subjected to connected domain analysis, and each connected domain obtained is regarded as a residual area; the direction of the centroid vector of the residual area is the direction from the center of the molten pool to the centroid of the residual area, and the modulus is the Euclidean distance between the centroid of the residual area and the center of the molten pool; the direction of the welding torch walking speed vector is the welding advancing direction, and the modulus is the walking speed; The position threat weight satisfies the expression: ; is the position threat weight of the mth residual region at the time t; is the position threat weight of the mth residual region at the time t; is the cosine value of the angle between the centroid vector of the mth residual region at the time t and the welding gun walking speed vector; is a natural exponential function;​​​ The dynamic rate of change of the residual region satisfies the expression: ; for The first moment The dynamic rate of change of each residual region; for The first moment The area of ​​each residual region; for The first moment The average residual temperature of each residual region; for The first moment Each residual region in The area of ​​the residual region in the residual temperature field at time t; for The first moment Each residual region in The average residual temperature of the corresponding residual region in the residual temperature field at time t; For time step; Spatial anomaly indicators satisfy the expression: ; for Spatial anomaly indicators at any given time; It is the hyperbolic tangent function; express The number of residual regions at each time step; This represents the maximum area of ​​all residual regions across all historical welding tasks involving the same material. The maximum value of the average residual temperature of all residual regions across all historical welding tasks involving the same material; The optimal state estimation and the space anomaly index are fused, and the welding parameters are dynamically adjusted, including: taking the optimal state estimation as the input of the main control loop, calculating the basic current adjustment amount through the PID controller ; inputting the space anomaly index to a threshold comparator: in response to , generating a walking speed increment proportional to , otherwise ; outputting and to the welding power supply and the motor driver respectively, so as to realize the synchronous adjustment of the welding current and the walking speed, is an anomaly threshold value.

2. The method of claim 1, wherein, the method for obtaining the residual temperature field is as follows: subtracting the temperature field image actually acquired at each time from the ideal temperature field image at the time to obtain the residual temperature field at the time.

3. An intelligent welding machine operating parameter control system characterized by, The method comprises the following steps: a processor and a memory, the memory storing computer program instructions, when the computer program instructions are executed by the processor, realizing a kind of intelligent welding machine operating parameter control method according to any one of claims 1-2.

Images