An intelligent wire feeding device control method for a welding robot

Abstract

The present application relates to the technical field of control system, and particularly relates to a kind of intelligent wire feeding equipment control method for welding robot, comprising obtaining actual welding trajectory;The curvature of each point in actual welding trajectory is calculated, and the curvature sequence of actual welding trajectory is obtained;Curvature sequence is threshold segmented, a plurality of curvature sections, actual trajectory section corresponding to curvature section and the curvature representative value of each actual trajectory section are determined;In target historical data set, determine the influence value corresponding to the curvature representative value;Determine the wire feeding speed allowable range of each actual trajectory section;Determine the welding direction and the optimal wire feeding speed when welding;The intelligent wire feeding equipment of welding robot is controlled to complete welding.The present application gives wire feeding speed adjustment in advance at trajectory curvature change place by the prediction-compensation idea driven by historical data, is favorable for reducing adjustment lag time, makes wire feeding action and welding movement keep synchronization, and weld forming is more uniform, and spatter is significantly reduced.

Description

Technical Field

[0001] This invention relates to the field of control system technology, and more specifically to a control method for an intelligent wire feeding device used in welding robots. Background Technology

[0002] In trajectory operations on complex curved surfaces such as pipe circumferential welding, aerospace blades, and helical surfaces, the trajectory path is relatively complex, and changes in the welding trajectory will significantly affect the wire feeding stability.

[0003] Existing problems: Traditional methods usually rely on real-time feedback to adjust the wire feed speed, but due to the dynamic changes in the curvature, angle and direction of the welding path, real-time adjustment has a lag, making it difficult to accurately match the wire feed speed with the welding motion, thus affecting the stability of welding quality. Summary of the Invention

[0004] This invention provides a control method for an intelligent wire feeding device for welding robots to solve existing problems.

[0005] The intelligent wire feeding device control method for welding robots according to the present invention adopts the following technical solution:

[0006] An embodiment of the present invention provides a control method for an intelligent wire feeding device for a welding robot. The method includes: acquiring an actual welding trajectory based on point cloud data of a welding area; calculating the curvature of each point in the actual welding trajectory to obtain a curvature sequence of the actual welding trajectory; performing threshold segmentation on the curvature sequence to determine multiple curvature segments, actual trajectory segments corresponding to the curvature segments, and representative curvature values ​​for each actual trajectory segment; determining an influence value corresponding to the representative curvature value in a target historical dataset; wherein the target historical dataset is a historical dataset corresponding to the type of welding wire used in the actual welding; the influence value is used to characterize the degree of influence of the wire feeding speed of the actual trajectory segment on the welding quality; determining the allowable range of wire feeding speed for each actual trajectory segment based on the influence value; determining the welding direction and optimal wire feeding speed during actual welding based on the allowable range of wire feeding speed and the influence value; and controlling the intelligent wire feeding device of the welding robot to complete the welding based on the welding direction and the optimal wire feeding speed.

[0007] Furthermore, the method further includes: dividing historical welding data into multiple historical datasets based on the type of welding wire; determining the curvature sequence of each historical welding trajectory in each historical dataset; performing threshold segmentation on the curvature sequence to determine multiple curvature segments, the historical trajectory segments corresponding to the curvature segments, the representative curvature value of the curvature segments, and the representative wire feed speed value of each historical trajectory segment; determining the quality score of the historical weld image corresponding to each historical trajectory segment; and determining the mapping relationship between the representative curvature value, the influence value, and the allowable range of the wire feed speed based on the representative curvature value, the representative wire feed speed value, and the quality score.

[0008] Furthermore, the method further includes: obtaining a training dataset for the scoring model; wherein the training dataset includes weld images and their corresponding quality scores; and training the scoring model using a mean squared error loss function to obtain the trained scoring model.

[0009] Determining the quality score of the historical weld image corresponding to each historical trajectory segment includes: inputting the historical weld image corresponding to the historical trajectory segment into the trained scoring model, and obtaining the quality score output by the scoring model.

[0010] Further, determining the mapping relationship between the representative curvature value, the influence value, and the allowable range of wire feed speed based on the representative curvature value, the representative wire feed speed value, and the quality score includes: for each representative curvature value, determining the coordinates of the binary pair consisting of the representative wire feed speed value and the quality score in a target coordinate system, and obtaining a scatter plot; wherein the target coordinate system is used to characterize the coordinate axis of the representative wire feed speed value on the horizontal axis and the coordinate axis of the quality score on the vertical axis; calculating a first parameter and a second parameter based on the scatter plot; wherein the first parameter is used to characterize the sensitivity of welding quality to changes in wire feed speed; the second parameter is used to characterize the linear significance of the relationship between the wire feed speed and the welding quality; determining the influence value based on the first parameter and the second parameter; determining the allowable range of wire feed speed based on the maximum and minimum values ​​of the representative wire feed speed value in the scatter plot; and obtaining the mapping relationship between the representative curvature value, the influence value, and the allowable range of wire feed speed.

[0011] Further, the step of calculating the first parameter and the second parameter based on the scatter plot includes: performing principal component analysis on the scatter points in the scatter plot to obtain multiple two-dimensional projection vectors and their corresponding projection values; calculating the first angle between the projection direction of the two-dimensional projection vector corresponding to the maximum projection value and the horizontal axis, and normalizing the first angle value to obtain the first parameter; calculating the second angle between the projection direction of the two-dimensional projection vector corresponding to the minimum projection value and the horizontal axis, and normalizing the second angle value to obtain the second parameter.

[0012] Further, determining the welding direction and optimal wire feed speed during actual welding based on the allowable range of wire feed speed and the influence value includes: obtaining the speed segment corresponding to each actual trajectory segment based on the allowable range of wire feed speed; performing speed segment merging iteration on the target speed segment until all the speed segments are merged to obtain the updated speed segment; and determining the welding direction and optimal wire feed speed during actual welding based on the allowable range of wire feed speed and the influence value of the updated speed segment.

[0013] The Nth speed segment merging iteration includes: determining a target speed segment; the target speed segment is the speed segment whose ranking value is N when the influence value is sorted from largest to smallest among the current speed segments; determining the intersection speed segment of the target speed segment; wherein the intersection speed segment is the speed segment whose allowable wire feeding speed range intersects with the allowable wire feeding speed range of the target speed segment in the preceding and following speed segments; merging the target speed segment and the intersection speed segment to obtain a merged speed segment; wherein the allowable wire feeding speed range of the merged speed segment is the intersection of the allowable wire feeding speed range of the target speed segment and the allowable wire feeding speed range of the intersection speed segment; the influence value of the merged speed segment is the larger of the influence value of the target speed segment and the influence value of the intersection speed segment; and updating the speed segment using the merged speed segment.

[0014] Further, determining the welding direction during actual welding based on the updated allowable range of wire feed speed for the updated speed segment and the influence value includes: determining a speed transition segment based on the updated speed segment; wherein the speed transition segment is used to characterize the speed segment where wire feed speed adjustment is required during actual welding; for each speed transition point in each welding direction, calculating the absolute value of the difference between the influence value of the speed transition segment and the influence value of the next speed segment; determining a transition risk value for each speed transition point based on the normalized absolute value of the difference and the influence value of the speed transition segment; wherein the transition risk characterizes the probability that adjusting the wire feed speed at the speed transition point will lead to welding quality defects; determining the total transition risk value for all speed transition points in each welding direction; and determining a target welding direction; wherein the target welding direction is the welding direction with the smallest total transition risk value.

[0015] Furthermore, based on the updated allowable range of wire feed speed for the speed segment and the influence value, the optimal wire feed speed for actual welding is determined, including: taking the minimum speed variation between adjacent speed segments as the optimization objective, and sequentially determining the optimal wire feed speed for each speed segment along the welding direction.

[0016] Furthermore, the step of controlling the intelligent wire feeding device of the welding robot to complete welding based on the welding direction and the optimal wire feeding speed includes: acquiring in real time the current welding torch position of the intelligent wire feeding device in the welding robot and the actual trajectory segment that the welding robot is about to enter; determining the optimal wire feeding speed corresponding to the actual trajectory segment; and dynamically fine-tuning the rotational speed of the wire feeding motor in the intelligent wire feeding device through a closed-loop feedback mechanism to match the actual wire feeding speed with the optimal wire feeding speed.

[0017] Furthermore, the step of obtaining the actual welding trajectory based on the point cloud data of the welding area includes: extracting the center line or feature edge of the weld from the point cloud data of the welding area; and fitting the center line or feature edge to the actual welding trajectory.

[0018] The beneficial effects of the technical solution of the present invention are:

[0019] In this embodiment of the invention, the actual welding trajectory is obtained based on the point cloud data of the welding area; the curvature of each point in the actual welding trajectory is calculated to obtain the curvature sequence of the actual welding trajectory; the curvature sequence is segmented by thresholding to determine multiple curvature segments, the actual trajectory segments corresponding to the curvature segments, and the representative curvature value of each actual trajectory segment; in the target historical dataset, the influence value corresponding to the representative curvature value is determined; based on the influence value, the allowable range of wire feeding speed for each actual trajectory segment is determined; based on the allowable range of wire feeding speed and the influence value, the welding direction and the optimal wire feeding speed during actual welding are determined; based on the welding direction and the optimal wire feeding speed, the intelligent wire feeding device of the welding robot is controlled to complete the welding. This invention utilizes a historical data-driven prediction-compensation approach to pre-calculate wire feed speed adjustments at points of trajectory curvature change. This reduces adjustment lag time, ensuring synchronization between wire feed and welding motions, resulting in more uniform weld formation and significantly reduced spatter. Furthermore, by employing Otsu multi-threshold segmentation and PCA to quantify the wire feed speed-quality score relationship, it automatically identifies speed-sensitive trajectory segments and prioritizes key segment speeds based on k-value sorting, reducing unnecessary frequent acceleration and deceleration. This reduces mechanical impact and improves the overall weld quality. Additionally, by introducing risk value assessment and bidirectional path comparison when determining the final speed sequence, dynamic programming or a greedy algorithm selects the welding direction with the smoothest change, achieving a gentle wire feed transition at speed transition points and further suppressing defects such as undercut and dents caused by sudden speed changes. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the structure of the intelligent wire feeding device provided in an embodiment of the present invention;

[0022] Figure 2 A flowchart illustrating the intelligent wire feeding device control method for a welding robot provided in an embodiment of the present invention;

[0023] Figure 3 A flowchart illustrating the historical dataset acquisition method provided in this embodiment of the invention;

[0024] Figure 4 This is a schematic diagram of the scatter point distribution in the wire feeding speed-quality scoring coordinate system provided in an embodiment of the present invention;

[0025] Figure 5This is a schematic diagram of the speed segment division when the welding direction is the first direction, provided in an embodiment of the present invention;

[0026] Figure 6 This is a schematic diagram of the speed segment division when the welding direction is the second direction, as provided in an embodiment of the present invention. Detailed Implementation

[0027] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent wire feeding device control method for welding robots proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0029] Before introducing the control method for the intelligent wire feeding device used in welding robots, let's first introduce the welding robot involved in this method and the intelligent wire feeding device for the welding robot:

[0030] Welding robots are mechatronic devices designed for industrial welding scenarios, capable of automatically completing welding operations according to preset programs or real-time instructions. They typically consist of core components such as a robotic arm, control cabinet, welding torch (or laser head), wire feeding device, sensor system, and welding power source. The robotic arm possesses multi-axis linkage capabilities, allowing for flexible posture adjustment in three-dimensional space; the control cabinet is responsible for program parsing and driving the motors of each axis; the welding torch or laser head directly performs the action of melting the workpiece and welding wire. During wire feeding, the high energy of the welding torch or laser head causes the welding wire to melt rapidly, and the molten metal droplets precisely drip onto the weld area, thereby improving the weld formation quality and joint strength; the wire feeding device stably delivers the welding wire according to process requirements; the sensor system (such as laser scanners, vision cameras, current and voltage sensors) monitors the weld position, molten pool state, and arc characteristics in real time; and the welding power source provides the required current and voltage. Under the closed-loop control of software algorithms and hardware, the entire system can operate continuously along complex trajectories, ensuring consistent weld formation and stable penetration depth. This allows it to replace manual labor in completing highly repetitive, high-strength, or precision-required welding tasks, and is widely used in fields such as automobile manufacturing, shipbuilding, aerospace, and pipeline engineering.

[0031] Intelligent wire feeding equipment: Please refer to Figure 1The working principle of the intelligent wire feeding device in a certain application scenario is as follows: the welding wire in the wire hopper is fed to the main wire feeder through the main control pulley wire feeding tube. During the welding process, the wire end sensor monitors the end position of the welding wire in real time to ensure that the welding wire is correctly fed to the welding area. The main control unit coordinates the work of each component, controls the operation of the main wire feeder according to the welding parameters, and receives signals from the main wire feeder through the signal amplification line, thereby adjusting the feeding speed and position of the welding wire. The robot control cabinet is responsible for controlling the robot's movements, enabling the laser head to weld according to the preset trajectory, while maintaining communication with the main control unit to ensure the coordination and synchronization of the welding process. This intelligent wire feeding device can automatically adjust the feeding of the welding wire according to the requirements of the welding task, ensuring the consistency and stability of the welding quality.

[0032] The following description, in conjunction with the accompanying drawings, details a specific scheme for a control method of an intelligent wire feeding device for a welding robot provided by the present invention.

[0033] Please see Figure 2 This illustrates an embodiment of the present invention providing a control method for an intelligent wire feeding device for a welding robot, comprising:

[0034] Step S110: Obtain the actual welding trajectory based on the point cloud data of the welding area.

[0035] Preferably, in one embodiment of the present invention, step S110 includes: extracting the center line or feature edge of the weld from the point cloud data of the welding area; and fitting the center line or feature edge to the actual welding trajectory.

[0036] It should be noted that the point cloud data of the above-mentioned welding area can be obtained by devices such as laser line scanners or structured light projectors. The schemes of extracting the weld center line or feature edges through point cloud processing and the schemes of fitting the center line or feature edges to the actual welding trajectory are both relatively mature and well-known technologies. For specific implementation methods, please refer to the relevant technologies. The embodiments of this invention will not be described in detail.

[0037] Step S120: Calculate the curvature of each point in the actual welding trajectory to obtain the curvature sequence of the actual welding trajectory.

[0038] It should be noted that the curvature mentioned above is a measure of the degree of bending of the trajectory curve at a certain point. The curvature of a straight line is zero, and the smaller the radius of the arc, the greater the curvature. The curvature value is larger at sharp turns or corners. For welding robots, a greater curvature of the trajectory curve means that the welding torch needs to change direction or posture rapidly within a very short distance, which directly affects the flow of the molten pool, the stability of the arc, and the wire feeding resistance. Therefore, curvature changes become an important basis for adjusting process parameters such as wire feeding speed, welding speed, and laser power. For example, in high curvature areas, the welding speed is reduced, and the wire feed needs to be reduced to avoid molten metal accumulation; while in low curvature areas, the welding speed is increased, and the wire feed needs to be increased to prevent weld depressions. If adjustments are not made in time, it will lead to defects such as uneven weld formation and spatter. It is understood that curvature is a relatively mature and well-known technology, and its specific implementation method can be found in related technologies. The embodiments of this invention will not be described in detail here.

[0039] Step S130: Perform threshold segmentation on the curvature sequence to determine multiple curvature segments, the actual trajectory segments corresponding to the curvature segments, and the representative curvature values ​​of each actual trajectory segment.

[0040] It should be noted that step S130 above can use the Otsu multi-threshold segmentation method to perform threshold segmentation on the curvature sequence. Otsu multi-threshold segmentation is a generalization of the classic Otsu algorithm, mainly used to automatically divide a continuous data set (such as a curvature sequence) into several "segments," making the values ​​within the same segment as close as possible and the numerical differences between different segments as large as possible. It is understood that the Otsu multi-threshold segmentation method is a relatively mature and well-known technology; its specific implementation can be found in related technologies, and will not be elaborated further in this embodiment.

[0041] It should be further noted that in practical applications, an empirically reasonable range for the number of segments can be set based on the complexity of the welding trajectory (e.g., setting the number of segments to 3 to 5) to avoid over-segmentation. It is understood that the final segmentation result (the final number of segments and the boundaries of each segment) is automatically calculated by the Otsu multi-threshold segmentation method based on the distribution characteristics of the input curvature sequence, rather than relying on manually set fixed parameters. This approach ensures that the Otsu multi-threshold segmentation method has universality and adaptability for welding trajectories of different shapes.

[0042] Step S140: In the target historical dataset, determine the influence value corresponding to the curvature representative value; wherein, the target historical dataset is the historical dataset corresponding to the type of welding wire used in the actual welding; the influence value is used to characterize the degree of influence of the wire feeding speed of the actual trajectory segment on the welding quality.

[0043] The following describes the methods for obtaining historical datasets:

[0044] Please see Figure 3 Preferably, in one embodiment of the present invention, the above-described intelligent wire feeding device control method for welding robots further includes:

[0045] Step S210: Based on the type of welding wire, divide the historical welding data into multiple historical datasets.

[0046] It should be noted that each historical dataset contains multiple data points, and each data point contains:

[0047] Trajectory image: This refers to the image of the welded area obtained by an industrial camera after the welding process is completed.

[0048] Trajectory curve: Trajectory data refers to the curve formed by the welding trajectory during each welding operation in history. For example, if the welding path is a straight line segment, then the trajectory data is also a line segment of the same length. The method for obtaining this data is as follows: 3D point cloud data of the weld surface is acquired using a laser line scanner or structured light projector. The centerline or feature edges of the weld (such as the ridge line of a V-groove) are extracted through point cloud processing and fitted into an abstract trajectory (such as a V-shaped broken line).

[0049] Step S220: For each historical welding trajectory in each historical dataset, determine the curvature sequence of the historical welding trajectory.

[0050] Step S230: Perform threshold segmentation on the curvature sequence to determine multiple curvature segments, the historical trajectory segments corresponding to the curvature segments, the representative curvature values ​​of the curvature segments, and the representative wire feeding speed values ​​of each historical trajectory segment.

[0051] It should be noted that the purpose of the aforementioned intelligent wire feeding device control method for welding robots is to obtain a current welding speed adjustment scheme by analyzing historical welding data. Considering the significant differences in the deposition characteristics, thermal conductivity, and wire feeding resistance of different materials, the required process parameters for the same welding trajectory—that is, the wire feeding speed—are material-dependent. During the welding process, when the curvature of the trajectory changes (e.g., from a straight line to an arc), the wire feeding speed must be adjusted accordingly. For example, in high-curvature areas, the welding speed decreases, requiring a reduction in the wire feed to avoid molten metal accumulation; while in low-curvature areas, the welding speed increases, requiring an increase in the wire feed to prevent weld depressions. Failure to adjust in time will lead to defects such as uneven weld formation and spatter.

[0052] Therefore, the historical welding data first needs to be divided into different datasets based on the type of welding wire. Then, for each data point in the trajectory curve within each dataset, the curvature of each point on the curve is calculated, resulting in a curvature sequence. This curvature sequence is then segmented using the Otsu multi-threshold method to obtain multiple curvature segments, with curvature values ​​within the same segment being similar. For each curvature segment, the absolute value of the difference between each curvature value in the sequence and all other curvature values ​​is calculated. The smallest absolute value of the difference and the corresponding curvature value are used as the representative curvature value for that segment. Finally, the wire feed speed curve corresponding to each curvature segment is obtained, and the speed value with the highest frequency in this curve is taken as the wire feed speed for that corresponding curvature.

[0053] It should be further noted that the selection criterion for the aforementioned representative curvature value can be: selecting the point within the curvature segment that has the smallest difference from all other curvature points within the segment, i.e., the "most representative center point". The specific calculation basis is as follows:

[0054] (1) For a certain curvature segment, there are m curvature values: ;

[0055] (2) Calculate the curvature value for each segment. To all other curvature values ​​within the segment The sum of absolute distances (i.e., the sum of absolute differences): ,in, ,and ;

[0056] (3) Compare all The value to choose is... The smallest curvature value This is the representative value of curvature for this segment.

[0057] The aforementioned curvature representative value can reflect the point with the "smallest overall difference" or "most central" within the segment, thus reflecting the overall curvature characteristics of the segment to the greatest extent.

[0058] Step S240: Determine the quality score of the historical weld image corresponding to each historical trajectory segment.

[0059] Preferably, in one embodiment of the present invention, the above-mentioned intelligent wire feeding device control method for welding robots further includes: acquiring a training dataset of a scoring model; wherein, the training dataset includes weld images and their corresponding quality scores; and training the scoring model using a mean squared error loss function to obtain a trained scoring model. Step S240 includes: inputting historical weld images corresponding to historical trajectory segments into the trained scoring model to obtain the quality scores output by the scoring model. For example, in this implementation: for each weld segment corresponding to a curvature segment (the area corresponding to the curvature segment on the weld), an image of that area is obtained and denoted as the weld image corresponding to the curvature segment. A neural network for scoring is trained, and the training dataset consists of multiple welding images, with experts assigning 0-1 ratings to the welding conditions, and training using a mean squared error loss function to obtain a trained scoring neural network. Each weld image of a curvature segment is input into the scoring neural network to obtain a quality score. Segments with scores less than 0.8 in each dataset are removed, thereby obtaining multiple (wire feeding speed, quality score) pairs corresponding to each curvature. It is understandable that multiple curvatures can be obtained on each trajectory curve, and different trajectory curves may have the same curvature, but the corresponding wire feeding speeds may not be the same.

[0060] It should be further noted that the above scoring model can use a neural network. The architecture and training scheme of neural networks are relatively mature and well-known technologies. For specific implementation methods, please refer to the relevant technologies. The embodiments of this invention will not be described in detail here.

[0061] Step S250: Based on the representative value of curvature, the representative value of wire feeding speed, and the quality score, determine the mapping relationship between the representative value of curvature, the influence value, and the allowable range of wire feeding speed.

[0062] Preferably, in one embodiment of the present invention, step S250 includes: for each representative value of curvature, determining the coordinates of the binary pair consisting of the representative value of wire feed speed and the quality score in the target coordinate system, and obtaining a scatter plot; wherein, the target coordinate system is used to characterize the horizontal axis as the representative value of wire feed speed and the vertical axis as the coordinate axis of quality score; based on the scatter plot, calculating a first parameter and a second parameter; wherein, the first parameter is used to characterize the sensitivity of welding quality to changes in wire feed speed; the second parameter is used to characterize the linear significance of the relationship between wire feed speed and welding quality; based on the first parameter and the second parameter, determining the influence value; based on the maximum and minimum values ​​of the representative value of wire feed speed in the scatter plot, determining the allowable range of wire feed speed; and obtaining the mapping relationship between the representative value of curvature, the influence value, and the allowable range of wire feed speed.

[0063] It should be noted that the implementation method for establishing the mapping relationship between the representative value of curvature and the allowable range of wire feed speed described above may include:

[0064] (1) Data collection: Collect a large amount of historical successful welding process data. Each data includes: welding trajectory, curvature of each point on the trajectory, and wire feeding speed value corresponding to each point on the trajectory that has been verified as high quality in practice.

[0065] (2) Curvature segmentation and representative value extraction: For each historical trajectory, curvature segmentation is performed according to the Otsu multi-threshold segmentation method described above, and a representative curvature value is determined for each curvature segment.

[0066] (3) Statistical speed range: For the same representative value of curvature (allowing for small deviations), find all the wire feeding speed values ​​corresponding to the representative value in all historical data to form a speed set.

[0067] (4) Determine the allowable range: Perform statistical analysis on the velocity set (e.g., calculate the mean ± 3). Alternatively, one can directly take the range from the minimum to the maximum value to determine the allowable range of wire feed speed corresponding to the representative value of curvature. .

[0068] (5) Forming a mapping table / function: Repeat the above process to establish the corresponding allowable range of wire feed speed for all different curvature representative values, and finally form a mapping lookup table of "curvature representative value - allowable range of wire feed speed" or fit a mapping function. In actual welding, the system can quickly determine the upper and lower limits of the current wire feed speed by looking up the table or calculating the function based on the curvature representative value calculated in real time.

[0069] Preferably, in one embodiment of the present invention, the above-mentioned calculation of the first parameter and the second parameter based on the scatter plot includes: performing principal component analysis on the scatter points in the scatter plot to obtain multiple two-dimensional projection vectors and their corresponding projection values; calculating the first angle between the projection direction of the two-dimensional projection vector corresponding to the maximum projection value and the horizontal axis, and normalizing the first angle value to obtain the first parameter; calculating the second angle between the projection direction of the two-dimensional projection vector corresponding to the minimum projection value and the horizontal axis, and normalizing the second angle value to obtain the second parameter.

[0070] It should be noted that, since a single wire feeding speed may correspond to multiple quality scores, this embodiment of the invention does not use similarity calculation. Instead, it uses a coordinate system to represent the binary pairs, thereby analyzing the influence relationship between wire feeding speed and quality. For each curvature value in each dataset, a corresponding wire feeding speed-quality score coordinate system can be constructed, thus obtaining the coordinate points corresponding to each (wire feeding speed, quality score) binary pair, forming a scatter plot. All coordinate points in the scatter plot are used as input to the Principal Component Analysis (PCA) algorithm to obtain multiple two-dimensional projection vectors and the corresponding projection value for each vector. The projection vector represents the projection direction, and the corresponding projection value represents the projection length of all data points in the corresponding projection direction. The larger the projection length, the wider the distribution range of the data points in that direction. Specifically:

[0071] Please see Figure 4 In the wire feed speed-quality score coordinate system, the horizontal axis represents the wire feed speed, and the vertical axis represents the quality score. Figure 4 The diagonal line in the diagram represents the projection direction corresponding to the maximum projection value. The closer the angle between the directions of the maximum and minimum projection values ​​is to 90°, the closer the data distribution of the coordinate points is to a linear shape. Under this curvature, the more clearly the influence of wire feed speed on welding quality is defined. Conversely, the closer the direction of the maximum projection value is to 90°, the more significant the change in wire feed speed will lead to a larger change in wire feed quality under this curvature; that is, the greater the influence of wire feed speed on welding quality. Let 'a' be the arctangent angle of the ratio of the second element to the first element in the two-dimensional projection vector corresponding to the maximum projection value. 'a' represents the angle between the projection direction corresponding to the maximum projection value and the horizontal coordinate in the coordinate system. Similarly, obtain the angle value 'b' corresponding to the minimum projection value. Then, obtain the acute angle between these two angles in the coordinate system, denoted as 'c'. Finally, obtain the ratio of 'c' to 90°, denoted as 'd'. The larger the ratio 'd', the more clearly the influence of wire feed speed on welding quality is under this curvature; that is, a larger change in wire feed speed is more likely to cause a change in welding quality. The ratio of 'a' to 90° is denoted as 'a0'. The larger 'a0' is, the greater the influence of wire feed speed on weld quality at that curvature. For each curvature value in each dataset, the corresponding 'd' and 'a0' are obtained. The product of these two, 'k', is taken as the degree of influence of wire feed speed on weld quality at that curvature, i.e., the influence value mentioned above. The larger the value of K, the higher the requirement for wire feed speed to obtain higher weld quality at that curvature; in other words, a higher wire feed quality can only be obtained within a relatively small range of wire feed speeds.

[0072] It should be further noted that the Principal Component Analysis (PCA) algorithm described above is a relatively mature and well-known technology. For its specific implementation, please refer to the relevant technologies. The embodiments of this invention will not be described in detail here.

[0073] Step S150: Based on the influence value, determine the allowable range of wire feeding speed for each actual trajectory segment.

[0074] It should be noted that: before actual welding, the type of welding wire to be welded is first obtained, and then the dataset of the corresponding welding wire type from historical welding data is obtained; the welding trajectory corresponding to the actual welding is obtained, that is, the abstract representation of the weld shape, denoted as the actual trajectory. Multiple curvature segments are calculated for the actual trajectory using the same method, denoted as actual trajectory segments. For each actual trajectory segment, a representative curvature value is calculated, and then the influence value k corresponding to the same representative curvature value in historical data is obtained. The larger the k value, the higher the welding quality required for that actual trajectory segment, and the stricter the requirements for the wire feeding speed range. In addition, step S250 above determines the mapping relationship between the representative curvature value, the influence value, and the allowable range of wire feeding speed. After determining the influence value in step S140 above, the allowable range of wire feeding speed for each actual trajectory segment can be determined through the above mapping relationship.

[0075] Step S160: Based on the allowable range of wire feed speed and its influence value, determine the welding direction and the optimal wire feed speed for actual welding.

[0076] Preferably, in one embodiment of the present invention, step S160 includes: obtaining the speed segment corresponding to each actual trajectory segment based on the allowable range of wire feeding speed; performing speed segment merging iteration on the target speed segment until all speed segments are merged to obtain the updated speed segment; and determining the welding direction and optimal wire feeding speed during actual welding based on the allowable range of wire feeding speed and the influence value of the updated speed segment.

[0077] The Nth speed segment merging iteration includes: determining the target speed segment; the target speed segment is the speed segment with a sorted value of N when the influence value is sorted from largest to smallest in the current speed segment; determining the intersection speed segment of the target speed segment; wherein, the intersection speed segment is the speed segment in the preceding and following speed segments of the target speed segment whose allowable wire feeding speed range intersects with the allowable wire feeding speed range of the target speed segment; merging the target speed segment and the intersection speed segment to obtain the merged speed segment; wherein, the allowable wire feeding speed range of the merged speed segment is the intersection of the allowable wire feeding speed range of the target speed segment and the allowable wire feeding speed range of the intersection speed segment; the influence value of the merged speed segment is the larger of the influence value of the target speed segment and the influence value of the intersection speed segment; and updating the speed segment using the merged speed segment.

[0078] Preferably, in one embodiment of the present invention, step S160 includes: determining a speed transition segment based on the updated speed segment; wherein, the speed transition segment is used to characterize the speed segment in which wire feed speed adjustment is required during actual welding; for each speed transition point in each welding direction, calculating the absolute value of the difference between the influence value of the speed transition segment and the influence value of the next speed segment; determining the conversion risk value of each speed transition point based on the normalized absolute value of the difference and the influence value of the speed transition segment; wherein, the conversion risk is used to characterize the probability that wire feed speed adjustment at the speed transition point will lead to welding quality defects; determining the total conversion risk value of all speed transition points in each welding direction; determining the target welding direction; wherein, the target welding direction is the welding direction with the minimum total conversion risk value.

[0079] It should be noted that, considering the frequent acceleration and deceleration of the wire feed speed can easily lead to unstable welding results, this embodiment of the invention first identifies the actual trajectory segment with strict speed requirements. The speed of each actual trajectory segment is then determined sequentially according to the degree of strictness of the speed requirement, from highest to lowest. This results in fewer wire feed speed changes and thus higher quality welding results. The weld has two endpoints: endpoint A and endpoint B. Since the wire feed speed needs to be adjusted when moving from one speed segment to another, the speed adjustment varies depending on the welding direction. For example, the speed adjustment for welding from left to right differs from that for welding from right to left. Therefore, the endpoint from which welding begins to obtain the least speed change is selected, resulting in a more stable and higher quality welding result. Specifically:

[0080] Each trajectory segment corresponds to an allowable wire feed speed range, denoted as a speed segment. First, find the actual trajectory segment with the largest k-value. Calculate the intersection of the wire feed speed of this trajectory segment with the left and right side trajectory segments. If an intersection exists, obtain the intersection speed and merge it with the speed segment on that side to obtain a merged speed segment. Then, find the actual trajectory segment with the second-largest k-value. Calculate the intersection of the wire feed speed of this trajectory segment with the left and right side trajectory segments. If an intersection exists, obtain the intersection speed and merge it with the speed segment on that side to obtain a merged speed segment. Repeat this iterative process until all speed segments are merged. For each merged speed segment, the larger k-value of the two merged trajectory segments is taken as the k-value of the merged speed segment. The larger the k-value, the higher the priority should be given to determining the speed value of this trajectory segment, thus achieving higher welding quality. This results in the updated k-value for each speed segment. In the speed adjustment scheme, the speed value of the speed segment with the larger k-value should be given priority, and other segments should be adjusted based on this speed value. Subsequent speed segments include merged speed segments.

[0081] Let's assume the welding direction from point A to point B is denoted as the first direction, and the welding direction from point B to point A is denoted as the second direction. Along the first direction, moving from one speed segment to another inevitably requires adjusting the wire feed speed within the current segment. This is to avoid problems with wire feeding and affect quality when passing through the turning point due to the turning point structure, caused by using a wire feed speed that is not suitable for the next segment.

[0082] Please see Figure 5 and Figure 6 , Figure 5 and Figure 6 The diagram illustrates the speed ranges corresponding to the first and second welding directions. Different grayscale values ​​in the diagram represent the impact of wire feed speed on quality within each speed range; darker grayscale values ​​indicate a greater impact. Figure 5 For example, along this welding direction, to avoid significant risks when passing the transition point between two speed segments, the speed conversion process is placed in the first segment. The light-colored squares below the numbered blocks in the diagram represent the speed conversion process. The speed segment in which the square is located indicates the speed segment in which the conversion from the current speed segment to the next speed segment is completed. When moving from one speed segment to another, there is a significant risk that the welding quality may be poor due to an inappropriate wire feed speed at the transition point. Therefore, in this invention, when calculating the quality stability for different welding directions, the difference in k values ​​between the two speed segments is used to calculate the potential risks at the transition point. Taking the trajectory path in the first direction as an example, the absolute value of the difference between the k value of each speed conversion segment (the segment in which the speed conversion is implemented is the speed conversion segment) and the next speed segment is calculated. This difference is then normalized to obtain s. The smaller s is, the lower the probability of quality risk at the transition point. Taking the trajectory path in the first direction as an example, the risk value of each turning point is represented by the product of the k value of the previous speed transition segment and the s values ​​of the two speed segments corresponding to the turning point. This allows us to obtain the risk values ​​of all turning points. The sum of these values ​​is then used as the total risk value for the trajectory path in the first direction, representing the impact of wire feed speed changes on weld quality when welding along this path, or the probability of weld quality defects caused by adjusting the wire feed speed at speed transition points. The same method can be used to obtain the total risk value for the trajectory path in the opposite direction. The direction corresponding to the smaller total risk value is taken as the actual welding direction. When welding along this direction, the impact of wire feed speed changes on weld quality is relatively small.

[0083] Preferably, in one embodiment of the present invention, step S160 includes: taking the minimization of speed variation between adjacent speed segments as the optimization objective, and sequentially determining the optimal wire feed speed for each speed segment along the welding direction. This implementation, for example, employs a dynamic programming algorithm or a greedy algorithm, taking the minimization of speed variation between adjacent speed segments as the objective, and sequentially determining the optimal wire feed speed value for each speed segment starting from the starting point, ensuring that the speed is always within the allowable range and the overall variation is smooth; the final generated speed sequence can adapt to the wire feed requirements of different curvature segments while avoiding drastic speed fluctuations, thereby ensuring the stability and quality of the welding process.

[0084] It should be noted that the sum of squared differences between adjacent velocities can be used as the objective function of the dynamic programming algorithm or greedy algorithm described above. The objective function can be expressed as:

[0085]

[0086] in, and The first The speed segment and the first The optimal wire feeding speed for each speed range; This represents the total speed range.

[0087] Constraints when solving for the optimal wire feed speed may include:

[0088] (1) Velocity boundary constraints:

[0089]

[0090] in, This is the minimum allowable wire feeding speed, expressed in m / s, and its value can be set to 0.1. This is the maximum allowable wire feeding speed, expressed in m / s, and can be set to 2.0.

[0091] (2) Smoothness constraint:

[0092]

[0093] in, This represents the maximum allowable change between adjacent speed ranges, expressed in m / s, and its value can be set to 0.15.

[0094] (3) Iterative convergence criterion:

[0095] The convergence criterion for iteration can be set as follows: the change in the objective function value is less than a preset threshold for the change in the objective function in two consecutive iterations, that is, the following condition is met for two consecutive iterations:

[0096]

[0097] in, This represents the number of iterations. For the first The objective function value calculated in the next iteration; The threshold value for the change of the preset objective function can be set to 0.0001.

[0098] The convergence criterion for iterative convergence can also be set to the maximum number of iterations, that is:

[0099]

[0100] in, This represents the maximum number of iterations.

[0101] It should be noted that the above-mentioned dynamic programming algorithm and greedy algorithm are both well-known and mature technologies. For their specific implementation methods, please refer to the relevant technologies. The embodiments of this invention will not be described in detail here.

[0102] Step S170: Based on the welding direction and optimal wire feeding speed, control the intelligent wire feeding device of the welding robot to complete the welding.

[0103] Preferably, in one embodiment of the present invention, step S170 includes: acquiring in real time the current welding torch position of the intelligent wire feeding device in the welding robot and the actual trajectory segment that the welding robot is about to enter; determining the optimal wire feeding speed corresponding to the actual trajectory segment; and dynamically fine-tuning the rotational speed of the wire feeding motor in the intelligent wire feeding device through a closed-loop feedback mechanism to match the actual wire feeding speed with the optimal wire feeding speed. For example, in the actual welding control process, the system will achieve precise control through the following steps based on the pre-divided trajectory segments (speed segments) and their corresponding optimal wire feeding speed parameters: First, the welding control system reads the current welding torch position and the trajectory segment information to be entered in real time, and calls the preset wire feeding speed parameters for that segment; then, through a closed-loop feedback mechanism (such as an encoder monitoring the actual wire feeding speed and a current sensor detecting the arc state), the rotational speed of the wire feeding motor is dynamically fine-tuned to ensure that the actual wire feeding speed matches the theoretical value; simultaneously, the system will predict the trajectory curvature change trend in advance and initiate a smooth transition of the wire feeding speed at the end of the previous trajectory segment to avoid abrupt speed changes.

[0104] It should be noted that during the actual welding process, welding quality data (such as molten pool shape and spatter) will be continuously recorded throughout the welding process. This data will be used to optimize the speed range division and wire feeding parameters, forming a control loop for continuous improvement.

[0105] It should be further noted that the aforementioned closed-loop feedback mechanism is a control strategy that adjusts the input by real-time monitoring of the system output and comparing it with the desired value to achieve precise control. It is understood that the closed-loop feedback mechanism is a relatively mature and well-known technology; for its specific implementation, please refer to relevant technologies, and this embodiment of the invention will not elaborate further.

[0106] This invention is now complete.

[0107] In summary, in this embodiment of the invention, the actual welding trajectory is obtained based on the point cloud data of the welding area; the curvature of each point in the actual welding trajectory is calculated to obtain the curvature sequence of the actual welding trajectory; the curvature sequence is segmented by thresholding to determine multiple curvature segments, the actual trajectory segments corresponding to the curvature segments, and the representative curvature values ​​of each actual trajectory segment; in the target historical dataset, the influence values ​​corresponding to the representative curvature values ​​are determined; based on the influence values, the allowable range of wire feeding speed for each actual trajectory segment is determined; based on the allowable range of wire feeding speed and the influence values, the welding direction and the optimal wire feeding speed during actual welding are determined; based on the welding direction and the optimal wire feeding speed, the intelligent wire feeding device of the welding robot is controlled to complete the welding. This invention employs a historical data-driven prediction-compensation approach to pre-calculate wire feed speed adjustments at points of trajectory curvature change. This reduces adjustment lag time, ensuring synchronization between wire feed and welding motions, resulting in more uniform weld formation and significantly reduced spatter. Furthermore, by utilizing Otsu multi-threshold segmentation and PCA to quantify the wire feed speed-quality score relationship, it automatically identifies speed-sensitive trajectory segments and prioritizes key segment speeds based on k-value sorting, reducing unnecessary frequent acceleration and deceleration. This reduces mechanical impact and improves the overall weld quality. Additionally, by introducing risk value assessment and bidirectional path comparison when determining the final speed sequence, it selects the welding direction with the smoothest change through dynamic programming or a greedy algorithm, achieving a gentle wire feed transition at speed transition points and further suppressing defects such as undercut and dents caused by sudden speed changes.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A control method for an intelligent wire feeding device for a welding robot, characterized in that, The method includes: The actual welding trajectory is obtained based on the point cloud data of the welding area; Calculate the curvature of each point in the actual welding trajectory to obtain the curvature sequence of the actual welding trajectory; The curvature sequence is segmented by a threshold to determine multiple curvature segments, the actual trajectory segments corresponding to the curvature segments, and the representative curvature value of each actual trajectory segment; In the target historical dataset, an influence value corresponding to the curvature representative value is determined; wherein, the target historical dataset is the historical dataset corresponding to the type of welding wire used in the actual welding; the influence value is used to characterize the degree of influence of the wire feed speed of the actual trajectory segment on the welding quality; Based on the aforementioned influence values, the allowable range of wire feeding speed for each of the actual trajectory segments is determined; Based on the allowable range of wire feeding speed and the influence value, the welding direction and optimal wire feeding speed during actual welding are determined. The step of determining the welding direction and optimal wire feed speed during actual welding based on the allowable range of wire feed speed and the influence value includes: Based on the allowable range of wire feeding speed, obtain the speed segment corresponding to each actual trajectory segment; Perform speed segment merging iterations on the target speed segment until all speed segments have been merged, and obtain the updated speed segment; Based on the updated allowable range of wire feed speed for the speed range and the influence value, the welding direction and optimal wire feed speed during actual welding are determined. The determination of the optimal wire feed speed for actual welding based on the updated allowable range of the wire feed speed within the speed range and the influence value includes: With the goal of minimizing the speed variation between adjacent speed segments, the optimal wire feed speed for each speed segment is determined sequentially along the welding direction. The Nth velocity segment merging iteration includes: Determine the target speed segment; the target speed segment is the speed segment whose ranking value is N when the influence values ​​are sorted from largest to smallest in the current speed segment. Determine the intersection speed segment of the target speed segment; wherein, the intersection speed segment is the speed segment in the preceding and following speed segments of the target speed segment where the allowable range of the wire feeding speed intersects with the allowable range of the wire feeding speed of the target speed segment; The target speed segment and the intersection speed segment are merged to obtain a merged speed segment; wherein, the allowable range of the wire feeding speed of the merged speed segment is the intersection of the allowable range of the wire feeding speed of the target speed segment and the allowable range of the wire feeding speed of the intersection speed segment; the influence value of the merged speed segment is the larger of the influence value of the target speed segment and the influence value of the intersection speed segment. The speed segment is updated using the merged speed segment; based on the welding direction and the optimal wire feeding speed, the intelligent wire feeding device of the welding robot is controlled to complete the welding.

2. The intelligent wire feeding device control method for welding robots according to claim 1, characterized in that, The method further includes: Based on the type of welding wire, the historical welding data is divided into multiple historical datasets. For each historical welding trajectory in each of the historical datasets, determine the curvature sequence of the historical welding trajectory; The curvature sequence is segmented by thresholding to determine multiple curvature segments, the historical trajectory segments corresponding to the curvature segments, the representative curvature values ​​of the curvature segments, and the representative wire feeding speed values ​​of each historical trajectory segment. Determine the quality score of the historical weld image corresponding to each of the aforementioned historical trajectory segments; Based on the representative value of curvature, the representative value of wire feeding speed, and the quality score, a mapping relationship is determined between the representative value of curvature, the influence value, and the allowable range of wire feeding speed.

3. The intelligent wire feeding device control method for welding robots according to claim 2, characterized in that, The method further includes: Obtain the training dataset for the scoring model; wherein, the training dataset includes weld images and their corresponding quality scores; The scoring model is trained using the mean squared error loss function to obtain the trained scoring model; The process of determining the quality score of the historical weld image corresponding to each of the historical trajectory segments includes: Input the historical weld image corresponding to the historical trajectory segment into the trained scoring model, and obtain the quality score output by the scoring model.

4. The intelligent wire feeding device control method for welding robots according to claim 2, characterized in that, The step of determining the mapping relationship between the representative curvature value, the influence value, and the allowable range of the wire feed speed based on the representative curvature value, the representative wire feed speed value, and the quality score includes: For each of the curvature representative values, the coordinates of the binary pair consisting of the yarn feed speed representative value and the quality score are determined in the target coordinate system, and a scatter plot is obtained; wherein, the target coordinate system is used to characterize the horizontal axis as the yarn feed speed representative value and the vertical axis as the quality score coordinate axis; Based on the scatter plot, a first parameter and a second parameter are calculated; wherein, the first parameter is used to characterize the sensitivity of the welding quality to changes in wire feed speed; and the second parameter is used to characterize the linear significance of the relationship between the wire feed speed and the welding quality. The influence value is determined based on the first parameter and the second parameter; Based on the maximum and minimum values ​​of the representative values ​​of the wire feeding speed in the scatter plot, the allowable range of the wire feeding speed is determined; Obtain the mapping relationship between the representative curvature value, the influence value, and the allowable range of wire feeding speed.

5. The intelligent wire feeding device control method for welding robots according to claim 4, characterized in that, The calculation of the first parameter and the second parameter based on the scatter plot includes: Principal component analysis is performed on the scatter points in the scatter plot to obtain multiple two-dimensional projection vectors and their corresponding projection values; Calculate the first angle between the projection direction of the two-dimensional projection vector corresponding to the maximum projection value and the horizontal coordinate axis, and normalize the first angle value to obtain the first parameter; Calculate the second angle between the projection direction of the two-dimensional projection vector corresponding to the minimum projection value and the horizontal coordinate axis, and normalize the second angle value to obtain the second parameter.

6. The intelligent wire feeding device control method for welding robots according to claim 1, characterized in that, The determination of the welding direction during actual welding based on the updated allowable range of the wire feed speed within the speed range and the influence value includes: Based on the updated speed range, a speed transition range is determined; wherein, the speed transition range is used to characterize the speed range in which wire feed speed adjustment is required during actual welding. For each speed transition point in each welding direction, calculate the absolute value of the difference between the influence value of the speed transition segment and the influence value of the next speed segment; Based on the normalized absolute value of the difference and the influence value of the speed transition segment, a transition risk value is determined for each speed transition point; wherein, the transition risk is used to characterize the probability that adjusting the wire feed speed at the speed transition point will lead to welding quality defects. Determine the total risk of transitioning to all speed inflection points in each of the aforementioned welding directions; Determine the target welding direction; wherein the target welding direction is the welding direction with the minimum total conversion risk value.

7. The control method for an intelligent wire feeding device for a welding robot according to any one of claims 1-6, characterized in that, The intelligent wire feeding device that controls the welding robot to complete the welding based on the welding direction and the optimal wire feeding speed includes: The current position of the welding torch in the intelligent wire feeding device of the welding robot and the actual trajectory segment that the welding robot is about to enter are obtained in real time. Determine the optimal wire feeding speed corresponding to the actual trajectory segment; Through a closed-loop feedback mechanism, the rotational speed of the wire feeding motor in the intelligent wire feeding device is dynamically fine-tuned so that the actual wire feeding speed matches the optimal wire feeding speed.

8. The control method for an intelligent wire feeding device for a welding robot according to any one of claims 1-6, characterized in that, The acquisition of the actual welding trajectory based on the point cloud data of the welding area includes: Extract the centerline or feature edge of the weld from the point cloud data of the welding area; The centerline or the feature edge is fitted to the actual welding trajectory.

Images