A welding process parameter reverse optimization method based on target cross-sectional area

Abstract

This invention relates to the field of welding robots and intelligent vision inspection technology, and discloses a method for reverse optimization of welding process parameters based on target cross-sectional area, including the following steps: Step 1: Establish a regression model between welding current, welding speed, and weld cross-sectional area, and convert this model into a quadratic equation with welding current as the unknown and welding speed as the parameter; Step 2: Within a preset feasible range of welding speed, discrete values ​​are taken with a fixed step size, and substituted sequentially into the quadratic equation to solve for the corresponding welding current value. This invention overcomes the limitation of existing technologies that can only predict weld size from process parameters in a forward direction. By converting the regression model into a quadratic equation with welding current as the unknown, and combining discrete traversal and root-finding algorithms, it achieves the reverse and accurate solution of multiple feasible combinations of welding current and welding speed based on any target cross-sectional area, providing a new path for the rapid and scientific selection of process parameters.

Description

Technical Field

[0001] This invention relates to the field of welding robots and intelligent vision inspection technology, and in particular to a reverse optimization method for welding process parameters based on target cross-sectional area. Background Technology

[0002] Currently, with the development of heavy equipment towards larger and more complex sizes, medium-thick plate structural components are increasingly widely used in energy equipment, engineering machinery, and other fields. These components typically feature large plate thickness, complex bevel shapes, and large weld filler volumes, placing higher demands on welding quality consistency and process stability in their manufacturing process. Introducing welding robots to replace manual labor is an important direction for improving manufacturing levels; however, current welding robots in engineering applications still mainly rely on teaching methods for path planning. For multi-layer, multi-pass welding tasks of medium-thick plates, this method has significant shortcomings: on the one hand, it lacks a quantitative correlation between process parameters and weld bead formation (especially weld bead cross-sectional area), making it difficult to achieve precise control of the weld filler process; on the other hand, the layer division and weld bead arrangement lack unified rules, and the planning process relies on repeated debugging, resulting in low efficiency.

[0003] Existing research has preliminarily established regression models between welding process parameters and weld geometry (such as weld width and weld reinforcement), but most of these studies focus on the positive prediction of weld formation results from process parameters. For multi-layer, multi-pass robotic welding, the weld cross-sectional area, as a key parameter connecting process planning and weld filling, is currently only used for result characterization. There is a lack of a method that addresses target filling requirements and can efficiently and backward optimize the best welding process parameters (welding current and welding speed). For example, given the target weld cross-sectional area, how to automatically and quickly solve for the optimal combination of process parameters that meets the area requirement while also considering heat input control and welding efficiency remains an unsolved problem in current technology.

[0004] Regarding the aforementioned technologies, the inventors believe that the following defects exist: the existing technology lacks a method that can automatically perform reverse engineering and global optimization based on a given target weld cross-sectional area within the process window allowed by the welding equipment, so as to quickly obtain the optimal welding current and welding speed that balances low heat input (controlling deformation) and high efficiency (high welding speed). Summary of the Invention

[0005] To address the technical problems mentioned in the background section, this invention provides a reverse optimization method for welding process parameters based on the target cross-sectional area.

[0006] This invention is achieved using the following technical solution: a method for reverse optimization of welding process parameters based on a target cross-sectional area, comprising the following steps: Step 1: Establish a regression model between welding current, welding speed and weld cross-sectional area, and convert the model into a quadratic equation with welding current as the unknown and welding speed as the parameter.

[0007] Step 2: Within the preset feasible range of welding speed, discrete values ​​are taken with a fixed step size, and then substituted into the quadratic equation in sequence to solve for the corresponding welding current value.

[0008] Step 3: Perform range constraint judgment on the obtained welding current value, eliminate invalid combinations that exceed the allowable range of the equipment or have no real solution, and obtain several sets of effective process parameter combinations.

[0009] Step 4: Construct an evaluation function with the objectives of minimizing line energy and maximizing welding speed, score each effective combination of process parameters, and select the combination with the best score as the final output welding current and welding speed.

[0010] Furthermore, the regression model in step 1 is a quadratic regression model, and its expression is:

[0011] in, The cross-sectional area of ​​the weld bead. For welding current, For welding speed, These are the regression coefficients determined through a second-order general rotational combination design experiment.

[0012] Furthermore, after the regression model is established, it undergoes analysis of variance, significance test, and lack of fit test to remove regression terms that have no significant impact on the cross-sectional area. The remaining regression coefficients are then reverse-encoded and converted to obtain the mapping equation between welding current, welding speed, and weld cross-sectional area expressed in terms of actual physical quantities.

[0013] Furthermore, the feasible range of welding speed in step 2 is determined based on welding process experiments, with a lower limit of 25 cm / min and an upper limit of 45 cm / min.

[0014] Furthermore, the fixed step size in step 2 is 0.5 cm / min.

[0015] Furthermore, in step 2, after substituting each discrete welding speed value into a quadratic equation, the corresponding welding current value is calculated using the quadratic formula:

[0016] in, The coefficient function for welding speed is obtained from the regression model.

[0017] Furthermore, in the equipment allowable range in step 3, the lower limit of the welding current is 200 A, and the upper limit is 300 A; Invalid combinations in step 3 also include cases where the welding current solution is a complex or negative number.

[0018] Furthermore, the evaluation function in step 4 is specifically as follows:

[0019] in, and is a dimensionless weighting coefficient.

[0020] Furthermore, the weighting coefficients are set according to the welding process requirements: when welding deformation is controlled with priority, the weight of the line energy term is increased; when production efficiency is improved with priority, the weight of the welding speed term is increased.

[0021] Furthermore, the cross-sectional area of ​​the target weld bead The result is derived from the layer planning of multi-layer and multi-pass welding. Specifically, after determining the number of welding layers and the number of weld passes in each layer, the single-pass target area is obtained by evenly distributing the theoretical total filling area of ​​each layer to each weld pass in that layer.

[0022] The theoretical total filling area is obtained by integrating the V-shaped bevel width function:

[0023] in, , The first The lower boundary height and the upper boundary height of the layer.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention breaks through the limitation of existing technologies that can only predict weld size from process parameters. By transforming the regression model into a quadratic equation with welding current as the unknown, and combining discrete ergodic and root-finding algorithms, it realizes the reverse and accurate solution of multiple feasible combinations of welding current and welding speed based on any target cross-sectional area, providing a new path for the rapid and scientific selection of process parameters.

[0025] This invention constructs a comprehensive evaluation function with the objectives of minimizing line energy (controlling heat input and reducing deformation) and maximizing welding speed (improving efficiency). By scoring and selecting the best among multiple feasible solutions, the optimal set can be intelligently selected from numerous parameter combinations that meet the area requirements. This effectively balances and controls welding deformation and improves production efficiency while ensuring weld filling quality.

[0026] The method of this invention directly serves the layer planning results of multi-layer, multi-pass welding. By taking the target cross-sectional area of ​​each pass required for each layer as input, the optimal welding process parameters are automatically output, realizing a seamless and automatic conversion from "geometric filling requirements" to "physical process instructions". This significantly reduces the reliance on manual experience-based adjustments and provides key technical support for building a fully automatic, teach-free welding robot path planning system.

[0027] This invention employs computationally efficient and logically clear algorithms such as discrete traversal, root-finding formulas, and linear weighted evaluation functions. It eliminates the need for complex iterative optimization or numerical simulation, enabling rapid convergence to the optimal solution. Furthermore, the weighting coefficients can be flexibly adjusted according to actual engineering needs, demonstrating strong adaptability and engineering practicality. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of an automated welding testing system in an embodiment of the present invention, wherein... The FAOYIWYG FR5 six-DOF collaborative robot. For Megmeet DEX500M fully digital gas shielded welding machine;

[0029] Figure 2 These are macroscopic metallographic images of weld sections corresponding to different welding process parameters in the embodiments of the present invention; Figure 3 This is a comparison chart of measured and predicted values ​​of weld bead forming dimensions (cross-sectional area, weld width, and weld height) and relative error analysis in an embodiment of the present invention. Figure 4 This is a scatter plot showing the correlation between measured and predicted values ​​of weld formation dimensions (cross-sectional area, weld width, and weld reinforcement) in an embodiment of the present invention. Figure 5 This is a contour plot of the weld cross-sectional area in an embodiment of the present invention with respect to the process parameters of welding current and welding speed; Figure 6 This is a contour plot of the weld width with respect to the process parameters of welding current and welding speed in an embodiment of the present invention; Figure 7 This is a contour plot of the weld reinforcement height in this embodiment of the invention, relating to the process parameters of welding current and welding speed. Figure 8 This is a schematic diagram of the geometric feature parameters of the V-shaped bevel section in an embodiment of the present invention; Figure 9 This is a schematic diagram of the V-shaped bevel equal-height filling strategy and layer channel planning in an embodiment of the present invention. 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. Detailed Implementation

[0031] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should be noted that these embodiments are only for explaining the present invention and do not constitute a limitation on the scope of protection of the present invention. Various modifications or variations that can be made by those skilled in the art based on an understanding of the technical solution of the present invention without creative effort still fall within the scope of protection of the present invention.

[0032] I. Experimental Platform and Material Preparation This embodiment uses, as follows: Figure 1 The automated welding testing system shown is a FR5 six-DOF collaborative robot, which serves as the core execution unit of the system. Figure 1 (a) The robot has a repeatability of 0.02 mm and provides an open TCP / IP communication interface, enabling it to receive and execute spatial trajectory commands generated by the host computer. The welding energy system uses a Megmeet DEX500M fully digital gas-shielded welding machine. Figure 1 (b) The device can interact with the robot control system via digital communication to achieve real-time adjustment of parameters such as welding current, arc voltage and wire feeding speed.

[0033] The base material used in the experiment was Q235B low-carbon steel. After assembly, the welded specimens underwent surface treatment, with an angle grinder used to grind the bevel and its two sides to remove oxide scale, rust, and oil. ER90-6 solid welding wire with a diameter of 1.2mm was used as the welding filler material. The shielding gas was a mixture of 82% Ar and 18% CO2, with a gas flow rate controlled at 25L / min to ensure stable arc combustion and reduce welding spatter.

[0034] II. Welding process parameter range and coding The welding current is set in the range of 200-300A, the arc voltage is controlled at approximately 30V, the welding speed is controlled in the range of 25-45cm / min, and the wire extension is set to 13mm and kept constant. The specific process parameter ranges are shown in Table 2.1.

[0035] Table 2.1 Welding process parameter range ; To avoid the impact of differences in the dimensions and ranges of welding current and welding speed on the estimation of regression coefficients, the natural variables are converted into dimensionless coded variables. Based on the two-factor experimental system (k=2), the asterisk arm γ is calculated using the following formula:

[0036] Natural variables With encoded variables The conversion relationship between them is defined as follows:

[0037] In the formula, The zero level value of the factor. The step size of the factor change.

[0038] welding current The investigation range of 200–300 A is mapped to the encoding interval [-1.414, 1.414]. The zero level of the current factor is calculated. For 250A, varying step size Approximately 35.36A; similarly, welding speed The observation range is 25–45 cm / min, with zero level. The speed is 35 cm / min, and the step size is... The velocity is approximately 7.07 cm / min. The factor level coding is shown in Table 2.2.

[0039] Table 2.2 Welding Test Factor Level Coding Table ; III. Secondary General Rotational Combination Experimental Design Following the structural principles of a quadratic general rotational combination design, this experimental matrix contains 13 independent welding test samples: 4 sets of two-level full-factor experiments (coded as ±1), 4 sets of asterisk points (coded as ±1.414), and 5 sets of completely repeated center points (coded as 0,0). Detailed experimental protocols are shown in Table 2.3. The welding process was strictly performed according to the principle of complete randomization.

[0040] Table 2.3 Experimental Matrix for Secondary Universal Rotational Combination Design ; IV. Weld Section Measurement After welding, a milling machine is used to process the weld and base material, removing irregular sections of the weld morphology from the arc initiation and termination stages, retaining the uniform and stable weld section in the center. Macroscopic metallographic photographs of the weld cross-section are taken using a camera; the weld cross-sectional morphology corresponding to different welding process parameters is shown below. Figure 2 As shown.

[0041] ImageJ image processing software was used for quantitative analysis of weld cross-sectional images. First, the area containing the scale was selected, and the correspondence between pixels and actual lengths was established using the scale calibration function. After calibration, the geometric dimensions of the weld cross-section were measured using a linear measurement tool along both the transverse and vertical directions of the weld: the transverse dimension characterizes the weld bead width. The vertical dimension is used to characterize the weld reinforcement height. To obtain the cross-sectional area of ​​the weld, the image was converted to grayscale, and the weld region was separated from the background using a threshold segmentation method. The cross-sectional area of ​​the weld was then calculated using region analysis. The measured data are shown in Table 2.4.

[0042] Table 2.4 Test results of weld width, reinforcement height and area ; V. Establishing the Regression Model A second-order polynomial response surface model was used for fitting. Response variables These refer to the weld width, reinforcement height, and cross-sectional area, respectively. The independent variable is... and These correspond to the dimensionless coded values ​​of welding current and welding speed, respectively. The general formula for the functional relationship is:

[0043] Because this experiment is a two-factor experiment ( Specifically, it can be expanded as follows:

[0044] The regression coefficients are estimated using the following formula:

[0045]

[0046]

[0047]

[0048] In the formula, , , , , , In the above formula , , , The characteristic constants for the design can be found in Table 2.5.

[0049] Table 2.5 Calculation Parameters for Secondary Universal Rotational Combination Design ; Based on the above calculation method, the weld width can be calculated separately. , Yu Gao and cross-sectional area A quadratic regression equation relating welding current and welding speed.

[0050] This is used to establish the cross-sectional area of ​​the weld. The regression equation is used as an example for detailed explanation; the derivation process for the other height and weld width is similar.

[0051] Substituting the 13 sets of measured cross-sectional area data recorded in Table 2.4 one by one into the basic cumulative formula, the following can be calculated: , , , , , .

[0052] Subsequently, the above basic data and corresponding constant parameters are substituted into the estimation formula for the regression coefficients to calculate the regression coefficients in the cross-sectional area model: , , , , , .

[0053] Substituting the calculated coefficients into the two-factor polynomial, a preliminary quadratic regression equation for the weld cross-sectional area with respect to the welding current and welding speed coding values ​​can be established:

[0054] Using a similar derivation and calculation process, the initial regression equations for weld width and reinforcement height can be calculated separately.

[0055] VI. Testing the Regression Equation After establishing the preliminary quadratic general rotational regression equation, it is necessary to perform analysis of variance and statistical tests on the model's fit. Using the latest obtained cross-sectional area... Taking the experimental data and preliminary regression equation as an example, the testing process is divided into the significance test of the regression equation, the goodness-of-fit test, and the significance test of the regression coefficient.

[0056] (1) Significance test of the regression equation The significance test is used to determine whether the obtained regression surface model is meaningful for interpreting the experimental data as a whole. Based on the measured data of the 13 sets of cross-sectional areas in the table, the calculation process for the sum of squares and degrees of freedom is as follows: First, the total experimental cross-sectional area. .

[0057] Total Sum of Squares The calculation formula is:

[0058] Substituting the data into the calculation yields Its degrees of freedom .

[0059] Sum of Squares of Residuals This is the sum of squares of the differences between the experimentally measured values ​​and the model predictions:

[0060] Substitute each point into the equation, calculate, and sum to obtain the result. Its degrees of freedom .

[0061] Sum of squares of regression The sum of squares of total deviations is the difference between the sum of squares of residuals.

[0062] Degrees of freedom of the regression equation .

[0063] Perform an F-test on the entire equation and calculate... value:

[0064] Find the critical value from the F-distribution table. Because it was obtained through calculation. This shows that in At the given level, the established cross-sectional area quadratic regression model is highly significant overall.

[0065] (2) Goodness-of-fit test of the regression equation To determine whether the equation has good predictive ability throughout the entire parameter design interval, a failure-of-fit test is needed. This involves calculating the sum of squared residuals. Decomposed into the sum of squared pure errors of repeated trials at the center point with mismatched sum of squares :

[0066] The experimental design matrix shows that groups 9 through 13 are repeated trials with a zero-level center point. The area mean of these 5 groups... .

[0067] Pure error sum of squares for:

[0068] Degrees of freedom of pure error:

[0069] Loss-of-squares for:

[0070] Deficient degrees of freedom:

[0071] Perform a goodness-of-fit test test:

[0072] Find the critical value from the table. .because The lack of fit term is not significant. This indicates that the regression equation fits the actual experimental data very well, truly reflecting the relationship between various process parameters and cross-sectional area.

[0073] (3) Significance test of regression coefficients To further clarify the relationship between each term in the equation and the cross-sectional area The extent of the influence requires significance testing for each coefficient individually. The mean square of the residuals is known. Substituting the values ​​into the formula for calculating each F value, the specific results are as follows:

[0074]

[0075]

[0076]

[0077]

[0078] At a given significance level, the critical value can be obtained by looking up the table. Comparing the test results obtained from the above calculations with the critical values ​​reveals that all linear terms, quadratic terms, and their interaction terms in the equation... The test values ​​are all significantly greater than the critical threshold of 5.59. This fully demonstrates that not only do welding current and welding speed have a significant dominant influence on cross-sectional area, but the strong interaction between the two and their respective boundary curvatures are also not negligible.

[0079] Since all terms in the model exhibit statistically significant characteristics, all parameter variables in the initial model are retained according to the regression optimization principle. The final high-precision cross-sectional area prediction model for process optimization is as follows:

[0080] Reference cross-sectional area The derivation process allows for the direct construction of an initial mathematical model of the quadratic response surface for weld width and reinforcement height.

[0081] After substituting the 13 sets of experimental data into the regression coefficient solution matrix, the initial regression equation for the weld width was obtained as follows:

[0082] The obtained initial regression equation for the remaining height is:

[0083] The regression equations for melt width and excess height were analyzed using ANOVA and statistical tests, as shown in Table 2.6. Table 2.6 Significance and goodness-of-fit tests of regression equations ; According to the statistical data in Table 2.6, at the set significance level, the overall test values ​​for both melt width and excess height far exceed the overall critical test values, confirming that these two initial regression equations are statistically significant at the macroscopic level. Meanwhile, the lack of fit of both models... All values ​​are less than the critical value for good fit. The lack of significant misfit indicates that the regression equation fits the actual experimental data very well, truly reflecting the nonlinear mapping relationship between various process parameters and dimensional morphology, without omitting other key influencing variables.

[0084] The significance of the regression coefficients is shown in Table 2.7: Table 2.7 Significance Test of Regression Coefficients ; Comparative screening of the data in Table 2.7 revealed that, for the melt width model, the coefficient of the quadratic term... The test value was 2.28, which is less than the critical value of 5.59. This indicates that within this range of process parameters, the quadratic surface effect of welding speed has no significant impact on the weld width, and should be eliminated according to the principles of regression analysis; while the remaining first-order, second-order, and interaction terms... All values ​​are greater than the critical value, significantly affecting the weld width, and are therefore retained. After removing the insignificant terms, the F-values ​​for the significance test and the goodness-of-fit test of the weld width are 205.88 and 0.95, respectively, still meeting the test requirements.

[0085] For the Yu Gao model, the coefficient of the linear term With the coefficient of the quadratic term None of the terms reached the significance threshold, indicating that the linear change in welding speed and the quadratic effect of welding current have a weak impact on weld reinforcement height and contribute little to the model response. Therefore, these insignificant terms were removed from the initial regression equation to improve the model's simplicity and stability. All other terms in the model passed the significance or extremely significant tests, indicating that they have a relatively stable influence on weld reinforcement height and were retained. By screening and removing insignificant terms, redundant variables were avoided from interfering with the model's accuracy, allowing the regression model to maintain its fitting ability while possessing better generalization performance.

[0086] After eliminating and optimizing the insignificant terms, redundant variables in the initial model were removed, and a clear and highly confident prediction equation for weld morphology and size was finally established.

[0087] The optimized weld width regression equation is:

[0088] The optimized regression equation for weld reinforcement is:

[0089] VII. Reverse Encoding Conversion of Regression Model To enable the regression model to be directly applied to the robot welding control system and to achieve the conversion from mathematical model to actual process parameters, it is necessary to modify the coded variables used in the experimental design. , Converted to actual physical quantities, namely welding current With welding speed While the introduction of coded variables is beneficial for regression modeling and significance analysis, they need to be transformed into executable process instructions in practical engineering applications.

[0090] Based on the parameter mapping rules of the rotational combination design, the linear transformation relationship between the encoded value and the actual physical quantity is as follows:

[0091] Substituting the above conversion formulas into the cross-sectional areas that have undergone significance testing and elimination optimization, respectively... , melt width And Yu Gao In the quadratic regression equation, after polynomial expansion and merging like terms, a mathematical model of the true physical mapping between actual process parameters and weld formation dimensions can be obtained:

[0092] VIII. Comparison and Verification of Calculated and Measured Values ​​from the Prediction Model To verify the generalization prediction accuracy of the decoded process model, the actual welding current and welding speed from the previous 13 experimental schemes were re-substituted into the aforementioned physical mapping model to obtain the corresponding calculated predicted values, which were then compared and analyzed with the actual measured values ​​of the dimensions extracted from metallographic experiments. Specific measured and predicted values ​​are shown in Table 2.8, and the fitting effect and error distribution trend of each response index are shown in [Table 2.8]. Figure 3 .

[0093] Table 2.8 Comparison of calculated and measured values ​​of weld bead forming dimensions ; Depend on Figure 3 It can be seen that the relative errors of each response index are all within 10%. A scatter plot of the correlation between the measured values ​​and the model calculated values ​​is drawn, as shown below. Figure 4 As shown, the discrete data points for the three topographic dimensions all converge tightly to both sides of the 45° diagonal, confirming that the established quadratic mapping model has good predictive performance for weld formation characteristics.

[0094] IX. Contour Analysis of Weld Morphology and Process Parameter Response Use MATLAB software to plot the cross-sectional area. , melt width And Yu Gao A two-dimensional contour map.

[0095] Figure 5 Showing cross-sectional area The trend of process parameters changes. Cross-sectional area shows high sensitivity to both welding current and speed. Observation along the same contour line reveals that in order to maintain a constant metal filler volume, the welding current must be increased simultaneously with the welding speed. For a specific target cross-section, there are not unique combinations of process parameters.

[0096] Figure 6 and Figure 7 They respectively showed the melt width and Yu Gao Distribution patterns with varying welding parameters. As welding current increases, both weld width and reinforcement height tend to increase; while increasing welding speed significantly decreases weld width, but has a relatively smaller impact on reinforcement height. The contour distribution trends of weld width and reinforcement height are related to the cross-sectional area. The evolutionary patterns exhibit good consistency in terms of physical mechanisms.

[0097] 10. Back-mapping and optimization algorithm for process parameters based on target cross-sectional area In multi-layer, multi-pass welding path planning, the target cross-sectional area that needs to be filled for the current single-pass weld is first calculated based on the geometry of the groove and the set path layout strategy. To enable the robot to automatically match the most suitable welding current without human intervention. With welding speed This requires establishing a reverse mapping and optimization algorithm from the target area to process parameters. This includes the following steps: (1) Discrete solution of the target cross-sectional area Extracting the cross-sectional area physical mapping model established earlier, when the target cross-sectional area When the welding current is known, the prediction model can be directly converted into a prediction model. One-variable quadratic nonlinear equation:

[0098] To facilitate implementation in computer programs, let the coefficients of each term in the equation be:

[0099] During actual robot operation, welding speed Strictly limited to the set equipment process window, the effective physical range for this experiment is 25 cm / min to 45 cm / min. Welding speed set. Using discrete traversal variables and a fixed control step size of 0.5 cm / min, a one-dimensional traversal search is performed within this defined window. Each discrete variable... Substitute the values ​​into the above coefficient formula and calculate the corresponding welding current using the quadratic formula. :

[0100] In the program code calculation, cases where the discriminant within the square root is less than zero and there are no real solutions must be eliminated. Invalid roots whose calculation results exceed the effective current range of the 200A to 300A equipment must also be filtered out. Through discrete solving and range boundary filtering, the system can obtain a series of results that accurately meet the target filling area requirements. Effective process parameter combination set .

[0101] (2) Calculation of the global optimal combination of process parameters The parameter set obtained through discrete solution contains multiple effective combinations that can achieve the required cross-sectional area. For the welding of thick plates in heavy equipment, the core engineering objective of optimizing process parameters is to reduce heat input to minimize thermal deformation of the overall structure while ensuring that the single-pass filling amount meets the requirements, and at the same time, to maximize welding speed to improve production line efficiency.

[0102] Linear energy input per unit length of weld and its ratio A significant positive correlation is observed. Combining the aforementioned dual engineering objectives, a comprehensive evaluation penalty objective function is constructed, oriented towards minimizing linear energy input and maximizing operating speed:

[0103] In the formula, and These are dimensionless weighting coefficients, which can be assigned values ​​based on the different emphases on controlling thermal deformation and pursuing high efficiency at the work site. Each set of coordinates in the effective parameter set... Substitute each value into the objective function, compare and select the one that makes the objective function more efficient. The associated coordinates that yield the minimum value are used as the optimal process command for the final output. .

[0104] Specific example: Assume the target cross-sectional area of ​​a certain weld bead is obtained from the layer planning. Traverse the velocity range [25, 45] with a step size of 0.5, and for each... Solving for the current yields multiple effective combinations, for example: hour,(

[0105] hour,(

[0106] hour,(

[0107] Set weighting coefficients ( (Prioritize heat input control), (Taking efficiency into account), calculate the ( ) of each combination. value:

[0108]

[0109]

[0110] Select The combination with the smallest value ( As the optimal output.

[0111] (3) Source of the target cross-sectional area This embodiment establishes as follows: Figure 8 The local Cartesian coordinate system shown Set the center point of the bottom end of the gap at the root of the bevel as the origin of the coordinate system. The direction perpendicular to the surface of the base material and upward is defined as... The positive axis direction is used to characterize the stacking height; it is defined as the direction along the width of the base material plate and perpendicular to the welding advance direction. The axis is used to characterize the lateral arrangement trajectory of multi-layer, multi-pass welding. Within this coordinate system, the typical V-groove geometry is determined by the base material thickness. Angle with bevel Two independent parameters uniquely determine the model. The model assumes rigid contact at the bottom of the joint, and that the sidewalls extend linearly outwards at a constant slope from the origin until they reach the top surface of the base material. The key to accurately calculating the required filler metal volume lies in obtaining the horizontal cross-sectional width of the bevel. Regarding height coordinates explicit function expression Observing the geometric topology of this pointed-bottom V-shaped bevel, it can be seen that within the domain... The inner sidewall profile is represented by a linear ray passing through the origin. Due to the relationship between the two sidewalls of the bevel... The axis is strictly symmetrical, and the inclination angle of one side wall relative to the vertical axis is half of the bevel angle, that is... According to the trigonometric relationship of the tangent, at any height... The horizontal width on one side of the bevel is the product of the height and the half-angle tangent. Therefore, the cross-sectional width function of this bevel... It can be simplified to a direct proportional function:

[0112] The above function clearly shows that the cross-sectional width of the planning layer is linearly positively correlated with the current floor height. This simple geometric relationship greatly reduces the computational complexity of subsequent algorithms. For any given planning floor height... function value The maximum physical window for the lateral oscillation of the welding torch in this layer is strictly limited. Furthermore, to determine the total filler volume for the multi-layer, multi-pass plan, the total cross-sectional area of ​​the entire bevel area to be filled needs to be calculated. In mathematics, Equivalent to the width function The definite integral in the height direction. Through analytical integration, the total area can be expressed as:

[0113] Solving the above definite integral yields a practical algebraic expression for engineering applications:

[0114] The above formula quantifies the total amount of deposited metal required to complete the welding of the joint under ideal assembly conditions, providing an absolute total benchmark for subsequent planning of welding speed and allocation of the filling area of ​​each layer.

[0115] The foundation of multi-layer, multi-pass welding path planning lies in discretizing the continuous deposition process along the depth direction, dividing it into a series of ordered physical layers. This discretization process essentially involves a comprehensive trade-off between deposition efficiency, arc stability, and interlayer bonding quality, exhibiting typical multi-objective optimization characteristics. Considering the high sensitivity of arc morphology to arc extension stability in robotic automated welding, maintaining the relative uniformity of single-layer thickness is crucial for ensuring the robustness of the welding process. Based on the requirement of process stability, the layer planning adopts an "equal-height filling" strategy, that is, at the algorithm level, it is assumed that after the first root pass welding is completed, all subsequent fill and cap layers have uniform height characteristics.

[0116] The determination of the single-layer planning height depends on the actual forming capacity of the weld bead under specific welding process conditions. In multi-layer, multi-pass welding process design, the forming height of the weld bead is usually selected based on engineering practice experience. Generally, there are significant differences between root pass and fill pass welding in terms of process parameters, heat input, and forming requirements; therefore, the single-pass weld height corresponding to the two types of weld bead is not the same. Based on on-site production experience, the root pass weld bead height and the reference layer height of the fill pass weld bead are usually given separately as the basis for subsequent weld bead planning and layer calculation, such as... Figure 9 As shown.

[0117] In an ideal V-groove structure, the first root pass is primarily used to fill the pointed triangular area at the bottom of the groove. This weld not only achieves root penetration and forms the back weld, but also provides a stable reference surface for subsequent filler layers. Considering the requirements for penetration quality and back weld formation in the root pass, its welding process typically employs lower heat input parameters. Therefore, it is treated separately in the height direction, and the actual deposition height of the root pass is defined as... This height affects both the quality of the root weld and the starting position of the subsequent filler weld layer.

[0118] After determining the root pass height, the remaining depth to be filled within the groove is: For subsequent filler layers, the height of each single layer is determined based on an empirically established reference layer height. Planning is then carried out. The number of filler layers can be determined by the relationship between the remaining fill depth and the reference layer height, and rounded up to ensure complete fill of the bevel. Therefore, the total number of welding layers, including the root pass, is calculated. The following mathematical formula is derived:

[0119] The rounding operation in the above calculation means that if the original baseline layer height is strictly followed... ,but The cumulative height of the weld layers will exceed the target plate thickness. To eliminate this potential error of excessive redundant height, the total number of layers in integer form was determined. Next, the actual fill layer height must be globally fine-tuned and corrected. The corrected, unified actual layer height... By distributing the remaining fill depth evenly to the determined Layer filling layer obtained:

[0120] Through this rigorous correction procedure, the V-shaped bevel edge in the physical space... The axial direction is precisely discretized as A horizontal slice with defined geometric boundaries. layer( The top surface height coordinates of ) were precisely determined as This strategy of reverse-deriving geometric layering based on process model constraints effectively avoids the risk of uncontrolled filling volume caused by setting layer heights based on experience, ensuring that the deposition thickness of each layer is always strictly within the range allowed by the process window.

[0121] After determining the height range of each floor Subsequently, the task of multi-layer, multi-pass planning was transformed into further calculating the number of weld passes and the cross-sectional area distribution within each layer.

[0122] Assume the first The required geometric filler cross-sectional area for a layer of bevel cut should be strictly equal to the sum of the cross-sectional areas of the weld metal deposited in all actual weld beads of that layer. Therefore, the theoretical cross-sectional area of ​​each filler layer can be obtained by integrating the bevel width function. It is known that the bevel cross-sectional width varies linearly with height, and its functional expression is: Therefore, the first Theoretical filling area of ​​the layer for:

[0123] After calculating the cross-sectional area of ​​a single layer, the number of weld passes required for that layer is determined based on the principle of equal area. Since the number of weld passes in a single layer must be an integer during actual welding, directly dividing the area based on geometric dimensions can easily lead to mismatches in area calculations. Therefore, the effective cross-sectional area of ​​a single pass under baseline process parameters is introduced. For reference only. A set of benchmark welding currents and welding speeds with good arc stability were selected through welding procedure qualification tests and obtained using a weld bead prediction model. The first... Theoretical total area of ​​the layer Based on the single-lane area The planned number of weld passes for that layer can be obtained by rounding the result to the nearest integer. :

[0124] After obtaining the specific number of weld passes, the cross-section of this layer needs to be geometrically discretized. Considering that the V-groove sidewalls have a fixed inclination angle, the horizontal stacking of adjacent weld passes is constrained by the geometric features of the sidewalls. Under the conventional process of sequentially arranging weld passes from one side of the groove to the other, the forming surface of the previous weld pass becomes the lateral support for the next weld pass. In this stacking pattern, the previous... The ideal cross-section of a single weld bead is approximated as a rhombus parallel to the bevel sidewall angle, while the last sealing weld bead, close to the other sidewall, has a trapezoidal cross-section. This combination of cross-sections more realistically reflects the physical accumulation pattern of the weld bead within the bevel.

[0125] The aforementioned rounding operation will cause the total planned weld area to differ from the theoretical area of ​​the trapezoidal slice. Numerical deviations arise between these parameters. If welding is performed directly using the baseline process parameters, the accumulation of these area deviations can easily lead to forming problems such as excessive interlayer height or incomplete bevel filling. To correct the volumetric errors caused by geometric discretization, the theoretical total area is... Redistribute equally among the already determined The corrected target cross-sectional area is obtained from the weld beads. .

[0126] After obtaining the target cross-sectional area Then, it is substituted into the inverse mapping algorithm from the target area to the process parameters. Within the allowable operating range of the welding machine, by setting different welding speeds and performing calculations, multiple sets of welding currents that meet the area requirements can be recalculated. With welding speed Combination. This calculation process not only determines the specific number of lanes. It also enables welding parameters to be dynamically adjusted according to changes in bevel size, providing reliable data support for subsequent planning of the spatial orientation of the welding torch.

[0127] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0128] In conclusion, 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 spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for reverse optimization of welding process parameters based on a target cross-sectional area, characterized in that, Includes the following steps: Step 1: Establish a regression model between welding current, welding speed and weld cross-sectional area, and convert the model into a quadratic equation with welding current as the unknown and welding speed as the parameter; Step 2: Within the preset feasible range of welding speed, discrete values ​​are taken with a fixed step size, and substituted into the quadratic equation in sequence to solve for the corresponding welding current value. Step 3: Perform range constraint judgment on the obtained welding current value, eliminate invalid combinations that exceed the allowable range of the equipment or have no real solution, and obtain several sets of effective process parameter combinations; Step 4: Construct an evaluation function with the objectives of minimizing line energy and maximizing welding speed, score each effective combination of process parameters, and select the combination with the best score as the final output welding current and welding speed.

2. The method according to claim 1, characterized in that, The regression model in step 1 is a quadratic regression model, and its expression is: ； in, The cross-sectional area of ​​the weld bead. For welding current, For welding speed, These are the regression coefficients determined through a second-order general rotational combination design experiment.

3. The method according to claim 2, characterized in that, After the regression model was established, it underwent analysis of variance, significance test and lack of fit test to remove regression terms that did not significantly affect the cross-sectional area. The remaining regression coefficients were then reverse-encoded to obtain the mapping equation between welding current, welding speed and weld cross-sectional area expressed in terms of actual physical quantities.

4. The method according to claim 1, characterized in that, The feasible range of welding speed in step 2 is determined based on welding process experiments, with a lower limit of 25 cm / min and an upper limit of 45 cm / min.

5. The method according to claim 1 or 4, characterized in that, The fixed step size in step 2 is 0.5 cm / min.

6. The method according to claim 1, characterized in that, In step 2, each discrete welding speed value is substituted into a quadratic equation, and the corresponding welding current value is calculated using the quadratic formula: ； in, , The coefficient function for welding speed is obtained from the regression model.

7. The method according to claim 1, characterized in that, In step 3, the allowable range of the equipment includes a lower limit of 200 A and an upper limit of 300 A for the welding current. Invalid combinations in step 3 also include cases where the welding current solution is a complex or negative number.

8. The method according to claim 1, characterized in that, The evaluation function in step 4 is specifically as follows: ； in, and is a dimensionless weighting coefficient.

9. The method according to claim 8, characterized in that, The weighting coefficients are set according to the welding process requirements: when welding deformation is controlled in a priority manner, the weight of the line energy term is increased; when production efficiency is improved in a priority manner, the weight of the welding speed term is increased.

10. The method according to claim 1, characterized in that, The cross-sectional area of ​​the target weld bead The result is derived from the layer planning of multi-layer and multi-pass welding. Specifically, after determining the number of welding layers and the number of weld passes in each layer, the target area of ​​a single pass is obtained by distributing the theoretical total filling area of ​​each layer to each weld pass in that layer. The theoretical total filling area is obtained by integrating the V-shaped bevel width function: ； in, , The first The lower boundary height and the upper boundary height of the layer.

Images