Penetration parameter planning method and device suitable for groove welding

Abstract

The application aims to provide a penetration parameter planning method and device suitable for groove welding, establish a filling criterion through experimental data and determine a critical value for realizing penetration, physically quantify the molten pool behavior, combine with welding heat input, construct a penetration probability demarcation model, and plan parameters before welding. The scheme solves the problems of poor interpretability and unclear principle of traditional data-driven black box model through modeling based on clear physical mechanism. The reliability and stability of the welding process are ensured, and high-quality adaptive welding under variable gap conditions is realized.

Description

Technical Field

[0001] This invention relates to the field of welding, and more specifically to the calculation of welding process parameters and the prediction of penetration. Background Technology

[0002] Pulsed gas metal arc welding (P-GMAW) is a common welding method used in the welding of medium and heavy plates. Its basic process is as follows: Before welding, a bevel with a specific angle is machined at the edge of the plate joint requiring complete penetration, forming a groove with a cross-section typically V-shaped or X-shaped. During welding, pulsed gas metal arc welding is used. This process uses welding wire as the consumable electrode, and under the isolation of a shielding gas, a welding arc is generated using periodically varying pulsed currents. The arc provides high energy during the peak pulse period to achieve droplet transfer and deep penetration, while maintaining arc stability and reducing average heat input during the base pulse period. The entire process takes place within the geometrically constrained space of the bevel.

[0003] The distinction between penetration and incomplete penetration is the fundamental standard for evaluating the quality of such welds. Incomplete penetration refers to the welding heat failing to melt the base metal at the root of the groove, or insufficient melting, resulting in unfused gaps or areas at the joint root. Viewed from the back of the weld, it may appear as a completely absent weld or only a localized, discontinuous connection. The hazards of incomplete penetration are extremely serious. This defect forms a natural crack initiation point at the joint root. During the service life of the workpiece, especially under alternating or impact loads, the incomplete penetration area is highly susceptible to crack initiation and propagation, posing a major risk of structural failure and significant quality and safety accidents. Therefore, incomplete penetration is a defect that must be strictly prevented in critical welding involving load-bearing structures.

[0004] In existing technologies, the automation of bevel welding mainly follows two technical paths, but both have inherent limitations. One is the "teach-playback" automated welding with fixed parameters. Before welding, the operator teaches a set of fixed welding trajectories and process parameters for a preset bevel condition. The welding robot strictly replicates these fixed parameters to perform the operation. However, in actual production, errors in upstream processes such as material preparation and assembly are unavoidable, causing the actual bevel assembly gap to fluctuate randomly along the weld length, rather than being a constant value. Fixed welding parameters cannot adapt to this variation, resulting in poor reliability.

[0005] The second approach is an intelligent control method based on the detection of welding process feature information and data-driven analysis. This involves introducing sensing and detection methods such as vision to acquire feature information of the weld pool, such as its length, width, or three-dimensional shape. Models such as artificial neural networks are then established to predict the current penetration state or directly deduce the welding parameters that should be adjusted.

[0006] The second technical solution does indeed obtain welding parameters based on actual welding process information, but it lacks a physical mechanism correlation, resulting in a black box model. Its model construction relies entirely on statistical correlation learning from signals on the molten pool surface to the welding result. The model fails to establish an interpretable connection with the underlying physical mechanism of "how weld penetration is actually formed." This makes the model's decision-making logic uninterpretable in practical applications.

[0007] In practical applications, due to unclear principles, the boundary conditions for technology application cannot be defined at the physical level. A model may be effective within the operating conditions covered by the training data, but its behavior becomes unpredictable and its performance may degrade sharply when faced with new scenarios not fully covered by the training data. This vulnerability makes this type of technology, while active in laboratory research, quite restrictive in industrial use. In actual industrial applications, it is often only used for research on flat plate welding, and under the constraints of bevel joints and the complex conditions of fluctuating assembly gaps in production environments, it is difficult to achieve rapid, stable, and large-scale deployment and application. Summary of the Invention

[0008] The purpose of this invention is to provide a method and equipment for planning penetration parameters suitable for bevel welding. By establishing a filling criterion through experimental data and determining the critical value for achieving penetration, the behavior of the molten pool is physically quantified. Combined with welding heat input, a penetration probability boundary model is constructed, and pre-weld parameter planning is performed. This solution, through modeling based on a clear physical mechanism, completely solves the problems of poor interpretability and unclear principles in traditional data-driven black-box models. It ensures the reliability and stability of the welding process and achieves high-quality adaptive welding under variable gap conditions.

[0009] In a first aspect, the present invention provides a method for planning penetration parameters suitable for bevel welding, comprising the following steps:

[0010] Multiple welding samples are collected. Each welding sample includes groove gap G, welding parameters, and penetration state. The penetration state includes at least an incomplete penetration state and a fully penetrated state.

[0011] Based on the bevel gap and welding parameters of each welding sample, the criterion value corresponding to each welding sample is calculated. , ,in, This represents the equivalent average width of the bevel. For welding speed, The cross-sectional area of ​​the welding wire. The criterion value represents the wire feeding speed. It is used to reflect the ability of the electric arc to push aside the molten metal in the molten pool and sink;

[0012] The criterion critical value is derived based on welding result samples of all incomplete penetration states and welding result samples of all penetration states. ;

[0013] A penetration probability calculation model is established. The input data of the model is a binary vector containing the criterion value. The heat input per unit length of the weld is Q, and the output is the penetration probability P; the penetration probability calculation model depends on the discriminant function. The discriminant function The method was established by minimizing the cross-entropy loss function through all welding samples;

[0014] The penetration probability P and groove gap G required for actual welding are used as input values ​​and input into the penetration probability calculation model to obtain the corresponding welding parameters.

[0015] In a second aspect, the present invention provides an electronic device, including a processor and a memory; the processor is connected to the memory.

[0016] The memory is used to store executable program code; the processor reads the executable program code stored in the memory to run a program corresponding to the executable program code, so as to execute the method described above.

[0017] In summary, the present invention has the following beneficial effects:

[0018] 1. By constructing a penetration probability calculation model based on filling criteria and heat input, the welding parameters have been transformed from relying on fixed experience or "black box" predictions to scientific planning based on explicit physical quantities. This method can adaptively output reliable parameters based on the measured bevel gap and the target penetration probability, effectively solving the problem of unstable penetration quality caused by fluctuations in assembly gaps.

[0019] 2. By using orthogonal experimental design to obtain welding samples, a model database can be efficiently established with the fewest number of experiments while ensuring the comprehensiveness and representativeness of the sample data.

[0020] 3. A univariate relationship model was derived from regression analysis of samples from the critical penetration state, establishing a quantitative matching relationship between the bevel gap and welding current, voltage, and wire feed speed. This provides the system with the ability to quickly calculate baseline process parameters, ensuring that the starting point for parameter planning is reasonable and reliable.

[0021] 4. The filling criterion transforms the arc lifting mechanism into a precisely calculable dimensionless number. This criterion's critical value provides a clear quantitative standard for judging the likelihood of penetration, improving the interpretability of the scheme and the transparency of decision-making.

[0022] 5. A weld formation prediction model was constructed, enabling accurate forward prediction of the weld back width and height. This not only provides a predictive basis for assessing weld appearance quality but also provides indispensable key input data for subsequent penetration stability verification.

[0023] 6. A penetration verification step based on heat transfer principles was introduced. By calculating the temperature at the root of the bevel and comparing it with the melting point of the metal, the physical mechanism ensures that the planned parameters have sufficient heat to achieve penetration, providing the first theoretical safety guarantee and fundamentally avoiding the occurrence of incomplete penetration defects.

[0024] 7. Based on the penetration verification, a penetration stability verification is further introduced. This step judges the stability of the weld back formation from the perspective of mechanical equilibrium, which can effectively distinguish between ideal critical penetration and unqualified over-penetration state, providing a second theoretical safety guarantee to prevent weld leakage and burn-through. Attached Figure Description

[0025] Figure 1 A flowchart illustrating several embodiments of the penetration parameter planning method applicable to bevel welding is shown in this specification;

[0026] Figure 2 A schematic diagram of the sub-steps of box 101 is shown;

[0027] Figure 3 A schematic diagram of the structure of an electronic device according to some embodiments of this specification is shown. Detailed Implementation

[0028] The present invention will be further described in detail below with reference to the accompanying drawings.

[0029] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings.

[0030] The terms "first," "second," "third," etc., in the description, claims, and accompanying drawings are used to distinguish different objects, not to describe a particular order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0031] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the described elements without departing from the scope of this specification. Various processes or components may be appropriately omitted, substituted, or added to the examples. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Furthermore, features described with respect to some examples may be combined into other examples.

[0032] like Figure 1 As shown, Figure 1 A flowchart illustrating the penetration parameter planning method and equipment 100 applicable to bevel welding.

[0033] Box 101 is used to construct the statistical model for implementing method 100. Box 101 aims to build a statistical model system based on physical mechanisms and experimental data to provide core decision-making basis for subsequent intelligent process planning. This step quantifies the melt penetration forming conditions and establishes a reliable predictive model through systematic experimental design and data analysis.

[0034] like Figure 2 As shown, box 101 specifically includes multiple sub-steps.

[0035] Frame 1011, Conduct orthogonal experiments on welding processes and statistically analyze the weld formation results.

[0036] An orthogonal test of the system was performed on a standard welding test platform. For example, the substrate was made of 8mm thick Q345 steel, with a V-shaped bevel of 30°, and different assembly gaps were set by tooling.

[0037] The welding process employed pulsed gas metal arc welding (GMAW), with the shielding gas being a mixture of 80% Ar and 20% CO2, and a gas flow rate set at 15 L / min. The welding power source adopted a single-mode control. The bevel assembly gap G, welding speed V, and average welding current I were selected as the three key process factors.

[0038] For example, based on Table 1 shown below, a 3-factor, 7-level orthogonal array L is constructed. 49 7( 8 Tests were conducted with gap levels of 0.5, 1.0, 1.5, 2.0, and 2.5 mm, welding speeds ranging from 1.33 to 3.33 mm / s, and current levels ranging from 80 to 140 amps. The peak current of the P-GMAW was fixed at 500 A, with a duty cycle of 30%. For each parameter combination, a root pass was performed. After welding, 10 measurement points were evenly selected along the weld length. The width and reinforcement height of the weld back were measured using a video measuring instrument, and the average reinforcement height was calculated.

[0039] Table 1. Orthogonal Experimental Horizontal Setting Table for Melting Penetration Process

[0040] Gap / mm 2.5 0.5 1 1.5 2 1.5 1 <![CDATA[Welding speed / mm·s −1 > 1.33 2.00 2.67 3.33 3.00 2.33 1.67 Current / A 140 120 100 80 130 110 90

[0041] The three factors are groove gap G, welding current I, and welding speed V. weld Among them, the welding speed V weld This refers to the speed at which the welding torch moves during the welding process.

[0042] Based on the back reinforcement height and weld formation, the penetration state was divided into four categories with different evaluation priorities: burn-through was the highest, followed by weld leakage, then critical penetration, and finally incomplete penetration. Critical penetration was defined as an average back reinforcement height greater than 0 mm and less than 1.2 mm; incomplete penetration was defined as a back reinforcement height of less than or equal to 0 mm for more than 70% of the length direction, with its dimension uniformly recorded as 0; weld leakage was defined as a maximum back reinforcement height greater than or equal to 1.2 mm; burn-through was defined as an abnormal interruption of the welding process. Ultimately, 51 sets of tests were completed, including the verification group, yielding valid data for 24 sets of incomplete penetration, 22 sets of critical penetration, 1 set of weld leakage, and 4 sets of burn-through. Specific parameters and results are shown in Table 2 below.

[0043] Table 2. Statistical Analysis of Orthogonal Experimental Parameters for Penetration Process and Weld Formation Results

[0044] 1 2.5 1.33 140 8.05 2.62 burn-through 2 2.5 2.00 120 6.82 4.34 Weld leak 3 2.5 2.67 100 3.95 1.61 burn-through 4 2.5 3.33 80 3.59 1.46 burn-through 5 2.5 3.00 130 3.71 0.50 Critical penetration 6 2.5 2.33 110 3.69 0.65 Critical penetration 7 2.5 1.67 90 4.42 0.86 Critical penetration 8 0.5 1.33 120 0.00 0.00 Incomplete melting 9 0.5 2.00 100 0.00 0.00 Incomplete melting 10 0.5 2.67 80 0.00 0.00 Incomplete melting 11 0.5 3.33 130 0.00 0.00 Incomplete melting 12 0.5 3.00 110 0.00 0.00 Incomplete melting 13 0.5 2.33 90 0.00 0.00 Incomplete melting 14 0.5 1.67 140 0.00 0.00 Incomplete melting 15 1.0 1.33 100 0.00 0.00 Incomplete melting 16 1.0 2.00 80 0.00 0.00 Incomplete melting 17 1.0 2.67 130 3.69 0.58 Critical penetration 18 1.0 2.00 110 3.65 0.27 Critical penetration 19 1.0 3.00 90 0.00 0.00 Incomplete melting 20 1.0 2.33 140 5.66 0.94 Critical penetration 21 1.0 1.67 120 0.00 0.00 Incomplete melting 22 1.5 1.33 80 0.00 0.00 Incomplete melting 23 1.5 2.00 130 5.38 0.94 Critical penetration 24 1.5 2.67 110 3.76 0.56 Critical penetration 25 1.5 3.33 90 3.02 0.66 Critical penetration 26 1.5 3.00 140 4.47 1.07 Critical penetration 27 1.5 2.33 120 3.99 0.90 Critical penetration 28 1.5 1.67 100 0.00 0.00 Incomplete melting 29 2.0 1.33 130 0.00 0.00 Incomplete melting 30 2.0 2.00 110 3.52 0.42 Critical penetration

[0045] Table 2 (continued) Statistical analysis of orthogonal experimental parameters for penetration process and weld formation results

[0046] 31 2.0 2.67 90 0.00 0.00 Incomplete melting 32 2.0 3.33 140 3.40 1.01 Critical penetration 33 2.0 3.00 120 4.50 1.52 burn-through 34 2.0 2.33 100 3.08 0.64 Critical penetration 35 2.0 1.67 80 0.00 0.00 Incomplete melting 36 1.5 1.33 110 0.00 0.00 Incomplete melting 37 1.5 2.00 90 3.79 0.47 Critical penetration 38 1.5 2.67 140 3.96 1.06 Critical penetration 39 1.5 3.33 120 3.11 0.62 Critical penetration 40 1.5 3.00 100 3.48 0.63 Critical penetration 41 1.5 2.33 80 3.67 0.56 Critical penetration 42 1.5 1.67 130 0.00 0.00 Incomplete melting 43 1.0 1.33 90 0.00 0.00 Incomplete melting 44 1.0 2.00 140 0.00 0.00 Incomplete melting 45 1.0 2.67 120 3.81 0.88 Critical penetration 46 1.0 3.33 100 0.00 0.00 Incomplete melting 47 1.0 3.00 80 0.00 0.00 Incomplete melting 48 1.0 2.33 130 0.00 0.00 Incomplete melting 49 1.0 1.67 110 0.00 0.00 Incomplete melting 50 0.5 3.33 150 4.13 1.07 Critical penetration 51 2.5 2.33 140 5.31 1.19 Critical penetration

[0047] Box 1012, Establish a univariate relationship model for welding process parameters

[0048] Based on the experimental data obtained from box 1011, especially all the weld samples in the "critical penetration" state selected from them, regression analysis was performed to establish the quantitative relationship between the core process parameters. First, linear fitting was performed with the groove gap G as the independent variable and the welding current I as the dependent variable to obtain the relationship: I=k1*G+b. In the data of this embodiment, it was calculated that: I = -26.8 × G + 158.1.

[0049] Furthermore, based on the characteristics of a unified welding power source, the welding voltage U and wire feed speed V are established. f The linear relationship between U and welding current I is: U = K2 * I, = k3 * I, where k2 and k3 are often fixed parameters in a welding machine. In the data of this embodiment, U = 0.15976 × I and V f = 5.208 × 10 -4 × I.

[0050] Therefore, based on the data in box 1011, samples with "critical penetration" are selected. In box 1012, based on the different bevel gaps G, the corresponding welding current I, welding voltage U, and wire feed speed V can be directly obtained. f After obtaining the bevel gap, the system can quickly determine a set of initial values ​​for basic process parameters that are matched in order to achieve critical penetration.

[0051] Box 1013, Calculate and determine the welding fill criterion and its critical value.

[0052] To quantify the impact of the molten pool behavior within the bevel on penetration, a dimensionless number called the "filling criterion" is defined. The calculation formula is as follows: .

[0053] in, This represents the equivalent average width of the bevel. For welding speed, The cross-sectional area of ​​the welding wire. Wire feed speed. Welding speed. As already stated in box 1011 above, the cross-sectional area of ​​the welding wire is a known quantity in the welding process, and the wire feed speed has been obtained in box 1012. The equivalent average width of the bevel can be directly calculated from the gap width G, such as... Where h is the plate thickness and θ is the bevel angle, both of which are known values ​​in the welding process.

[0054] The introduction of the filling criterion is a key technical improvement of this invention. The physical meaning of this criterion is the ratio of the volume of the groove that needs to be filled per unit time to the volume of molten metal that the welding wire can provide, directly reflecting the filling rate of the molten pool or the level of the molten metal. It reflects the ability of the arc to push aside the molten metal in the pool and allow it to sink. When... When the surface area is small, the free surface of the liquid metal remains high, and the electric arc is supported by the liquid metal, making it difficult to affect the root of the bevel. When When the value increases to near a certain critical value, the supporting effect of the molten metal in the molten pool weakens, and the electric arc has the potential to sink under the geometric constraints of the bevel.

[0055] Substitute the process parameters of all test groups in box 1011 into this formula for calculation, and then compare the results of each group. Compared to the actual melt penetration state, the critical value is derived from the data in this embodiment. The value is 200, representing the difference between fully melted and incompletely melted. The critical value is 200. It should be noted that the penetration mentioned here includes critical penetration, weld leakage, and weld burn-through.

[0056] In box 1013, the key physical mechanism of the molten pool metal's support for the electric arc was creatively quantified into a precisely calculable dimensionless index, and its critical threshold was clearly defined. This transforms the judgment of complex penetration conditions from relying on qualitative experience or complex image features into a mathematical criterion with clear physical meaning. The establishment of this criterion provides a core physical principle for understanding and controlling the penetration process, and is the foundation for subsequent intelligent models to move beyond a black box and possess strong interpretability. The mechanism of this criterion stems from the fact that when there is too much filler metal, the arc is raised to a high position within the groove, making it difficult for heat to be transferred to the root of the groove and thus preventing penetration. Therefore, this criterion was designed.

[0057] Box 1014, Establish a model for calculating the probability of melt penetration.

[0058] To more accurately predict and plan processes, a filling criterion is constructed. The model uses a binary probabilistic prediction model with the heat input per unit length of welding as a joint variable. The heat input per unit length Q is a commonly used parameter in welding, and its calculation method is generally accepted: welding power / welding speed, or UI / .

[0059] Box 1014 is used to construct a penetration probability calculation model, which is a binary probability prediction model that can comprehensively quantify the "penetration probability". Its core is to recognize that whether the weld can penetrate is not determined by a single condition of filling criterion, but by the combined effect of the molten pool filling state and the overall thermal energy input.

[0060] The specific steps are as follows: First, based on all sample data obtained from the orthogonal experiment of box 1011, calculate two key input features: the filling criterion defined in box 1013. And the welding line energy Q, which reflects the total heat. Each sample is characterized as a feature vector. [Q], and use its corresponding penetration status as a label. Penetration is marked as 1, and non-penetration is marked as 0.

[0061]

[0062] Among them, the penetration welds include critical penetration and over-penetration weld samples.

[0063] Given the eigenvector x, the penetration probability P is:

[0064] .

[0065] Among them, feature vector .

[0066] Subsequently, the mathematical framework of logistic regression was used to fit the complex relationship between the "melt-through probability" and these two variables.

[0067] Its core is the penetration probability and a complex discriminant function. The function is The higher-order polynomials of Q contain individual terms, interaction terms, and higher-order terms of both, to accurately describe the complex nonlinear relationship between them and the possibility of melting through.

[0068]

[0069] Using all experimental data, i.e., N independent samples, By minimizing the cross-entropy loss function, the optimal discriminant function parameter β is found.

[0070] For example, the cross-entropy loss function is In the experimental data of this embodiment, the values ​​of β0 to β9 were calculated by minimizing the loss function, as follows:

[0071]

[0072] After obtaining all the specific values ​​of β, we can use the standard logic function to construct the penetration probability model as shown in the formula below, and map the value of the discriminant function S to a probability value p between 0 and 1, representing the probability of penetration. .

[0073] Furthermore, to better illustrate the drawing of the fill criterion... The relationship between heat input Q and heat input Q was plotted in the following phase diagram, and the penetration probability cutoff curves for p = 0.25, p = 0.50 and p = 0.75 were marked in the figure.

[0074]

[0075] The discriminant function value is mapped to a penetration probability p between 0 and 1 using the logistic function p = 1 / (1 + exp(-S(Π_fill, Q))). After model training, a "penetration probability contour line," also known as an equal probability boundary curve, can be plotted on the Π_fill-Q two-dimensional plane. For example, see the attached document. Figure 3 The diagram clearly shows the penetration probability cutoff lines for p'=0.25, 0.50, and 0.75.

[0076] The penetration probability calculation model constructed in box 1014 outputs a continuous probability value reflecting the certainty of penetration. This aligns with the fluctuating nature of engineering practice, enabling decision-making to shift from qualitative to quantitative methods, significantly improving its scientific rigor and precision. The model's two input variables are the filling criterion. Both the welding line energy Q, which reflects the total heat, are measurable and calculable quantities with clear physical meaning, and their decision-making logic is based on traceable mathematical functions. This overcomes the black-box nature of traditional neural network models based on visual features, which suffer from unclear principles and weak correlations. This makes every prediction of the model reliable and enhances the reliability of the technology and its potential for trust in industrial scenarios.

[0077] In box 1015, a weld formation prediction model is constructed, which is a feedforward artificial neural network (ANN) prediction model, used to predict the forming size of the weld back under given process parameters.

[0078] Using the orthogonal experimental data and statistical results in box 1011 as the dataset, an ANN prediction model for weld penetration forming size is constructed.

[0079] The model input feature vector is The output vector is Among them, W back H represents the width of the weld back. back The height is the back of the weld. The number of neural network layers and hidden layer neurons is unlimited, with the goal of obtaining the best prediction result.

[0080] For example, the input layer of this model can contain four nodes, corresponding to welding current I, welding voltage U, and welding speed v. weld And the bevel gap G. The output layer contains 2 nodes, corresponding to the prediction target: weld back width W. back He Yu Gao H back The network contains four hidden layers with 64, 128, 64, and 32 neurons respectively. All hidden layers use the SiLU activation function, and Dropout is set to 0.15 to prevent overfitting. Model weights are initialized using Kaiming Normal, and biases are initialized to 0. The network is trained using all 51 sets of experimental data obtained in box 1011, with mean squared error as the loss function. Parameters are adjusted through an optimization algorithm to enable the network to learn the complex nonlinear mapping relationship between input process parameters and output back dimension.

[0081] This model can quickly and accurately predict the width and height of the weld back side based on the planned or set process parameters.

[0082] At this point, all five sub-steps of box 101 are complete. Box 101 ultimately yields two models: one is a penetration probability calculation model, which visualizes and probabilizes the process window; the other is a weld formation prediction model, used to predict the specific forming dimensions of the weld back side. Together, they transform the welding process from an experience-based operation relying on fixed parameters or "black box" image recognition into a calculable, predictable, and customizable scientific decision-making process based on clear physical and statistical laws.

[0083] In box 102, process parameter planning for method 100 is implemented.

[0084] Box 102, building upon the established statistical model, performs automated process parameter planning before actual welding. Its purpose is to calculate and output segmented welding parameters that can adapt to varying gap conditions and ensure the target penetration quality, based on real-time measured bevel conditions and preset quality requirements, using the model system established in Box 101. The specific implementation process is as follows:

[0085] First, a laser vision sensor integrated into the front of the welding torch is used to scan along the trajectory of the bevel to be welded, measuring and acquiring real-time data on the continuous variation of the bevel assembly gap along the weld length. For example, for a 430 mm long variable gap bevel, it is divided into 7 continuous segments based on its gap distribution characteristics, and the average gap value of each segment is calculated as the representative value of that segment, as shown in the table below.

[0086]

[0087] Meanwhile, based on the quality grade requirements of the weld, the target confidence level for penetration is set to p′=0.96 in the system. Generally, a larger p′ is selected when the weld penetration quality requirements are high, and a smaller p′ is selected when the weld penetration quality requirements are low.

[0088] Subsequently, the planning system automatically performs calculations based on the above inputs. For each segment, the system first substitutes its average gap value G_seg into the univariate relation model established in box 1012 (I = -26.8 × G + 158.1) to calculate the reference welding current I_seg for that segment. It then further derives the corresponding welding voltage U_seg and wire feed speed V. f _seg. Then, the gap G_seg, the calculated basic parameters I_seg, U_seg, and V are... f The _seg and the target confidence level p'=0.96 are used as inputs and substituted into the melt penetration probability calculation model constructed by box 1014.

[0089] The system uses this model to solve inversely, calculating the key variable that enables the penetration probability to reach 96%—the welding speed v of this segment.weld Finally, the system planned a set of coordinated welding current, voltage, wire feed speed and welding speed for each of the seven segments. The complete planning results are shown in Table 3.

[0090] Table 3 Results of Welding Current Selection and Process Parameter Derivation in Penetration Process Planning Experiment

[0091] Welding current / A 150 150 145 145 125 125 90 Welding voltage / V 23.96 23.96 23.17 23.17 19.97 19.97 18.00 <![CDATA[Wire feeding speed / m•s -1 > 0.078 0.078 0.076 0.076 0.065 0.065 0.047 <![CDATA[Welding speed / mm•s -1 > 3.57 3.40 3.20 3.08 2.57 2.43 1.58

[0092] Box 103 is used to verify the process mechanism of method 100.

[0093] Box 103 is a crucial step in verifying the theoretical safety and reliability of the planned parameters after the process parameter planning in Box 102 is completed and before actual welding is performed. For example, the verification includes two aspects: first, welding penetration verification, i.e., whether the bottom of the bevel can be effectively melted; second, welding stability verification, i.e., whether the back of the weld is stably formed and meets quality requirements.

[0094] Through penetration verification, incomplete penetration and full penetration can be distinguished, while through stability verification, critical penetration and over-penetration can be distinguished, thereby ensuring the formation of critical penetration welds with stable welding process.

[0095] This will be explained in detail below.

[0096] For example, in the weld penetration verification, the penetration probability calculation based on the principle of heat transfer is performed on the process parameters planned for each segment of box 102.

[0097] The method for verifying weld penetration is to compare the temperature in the root region of the bevel. The magnitudes of the values ​​of the liquidus temperature and the liquidus temperature of the metal.

[0098] like If the temperature is greater than the liquidus temperature of the metal, i.e. the melting point of the metal, it indicates that the weld can achieve full penetration under these process parameters.

[0099] like If the temperature is lower than the liquidus temperature of the metal, it indicates that the weld cannot achieve full penetration under these process parameters, resulting in incomplete penetration defects. Further optimization of the process parameters is required.

[0100] Since the liquidus temperature of the metal is a known value, the key to this calculation process is the temperature in the bevel root region. The calculation.

[0101] .

[0102] in, T0 represents the temperature at the root region of the bevel; T0 represents room temperature. The top heat source, i.e., the quasi-steady-state temperature rise formed when the electric arc is conducted to the root of the bevel, is calculated by the following Rosenthal moving point heat source formula.

[0103] .

[0104] Where η is the arc thermal efficiency, κ is the thermal conductivity, and α is the thermal diffusivity, all of which are known values. zh is the heat source loading height, which can be calculated from the bevel size and the filler wire amount. weld For welding speed, the average power of the electric arc heat source is Pmean = I•U.

[0105] This refers to the transient local temperature rise in the root region caused by the electric arc tail flame during the pulse peak phase.

[0106] Based on experimental observations, the arc tail flame and its heating effect on the root of the groove are related to the groove gap size, the filling effect, the total energy during the peak arc phase, and the welding speed. Considering the heating effects of different variables on the root of the groove, the ratio of the total energy during the peak arc phase (taking the gap factor into account) to the temperature rise effect of the characteristic region volume of the root of the groove varying with welding speed, after dimensional unification, yields the temperature rise of the characteristic region of the root of the groove caused by the arc tail flame. The specific calculation formula is as follows:

[0107]

[0108] Where, η tail For the thermal efficiency of the tail flame, ρ is the peak total power of the electric arc. l C is the density of liquid metal. p δ represents the specific heat capacity of the metal and the duty cycle.

[0109] A is the gap enhancement factor, characterizing the degree to which the bevel gap scale enhances the effect of arc exhaust energy efficiency. root H represents the characteristic area at the root of the bevel related to the gap. tail This is the bevel tail flame effect function.

[0110] These three values ​​are obtained using the following formula:

[0111] ,

[0112] ,

[0113] ,

[0114] Among them, G ref The reference gap is m, the gap enhancement index is lz This refers to the height of the characteristic zone at the root of the bevel. The critical value for bevel filling is... The results have been obtained through experiments and statistical analysis in box 1013.

[0115] The following example demonstrates how a series of inherent welding process attribute parameters are retrieved from the process database or preset. Apart from setting the process parameters, these welding inherent attribute parameters are obtained from literature and industry-standard parameter queries. See Table 4 for details.

[0116] Table 4 Calculation parameters for temperature rise combination criterion at the root of the bevel

[0117]

[0118] Subsequently, according to This formula calculates the filling criteria for each of the seven segments and the temperature of the bevel root region. As shown in Table 5, it can be seen that the planned process parameters can all raise the temperature of the root region of the bevel to above the liquidus temperature of 1798 K, thus determining that a full penetration weld can be obtained.

[0119] Table 5. Verification results of segmented calculation of penetration stability

[0120]

[0121] At this point, the first verification, namely the weld penetration verification, has been completed. The second verification, namely the penetration stability verification, will then proceed. The core verification idea is to determine whether the free interface at the bottom of the molten pool remains stable.

[0122] In this verification, based on the confirmation that the heat is sufficient, it is necessary to further perform mechanical calculations on the weld formation stability under the fusion state.

[0123] Its core verification formula is ,in, Let γ be the external load acting on the bottom of the molten pool, γ be the surface tension maintaining interface stability, and L be the span of the free interface at the bottom of the molten pool. The derivation of this formula and the data sources are explained in detail below.

[0124] First, the external load on the bottom of the molten pool. The surface tension γ that maintains interface stability is described by the energy functional form of the Young-Laplace equation:

[0125] ,

[0126] The constant term does not affect the stability analysis in the variational calculus, therefore only the surface area increment is considered. and volume increment Substituting into the energy functional, we obtain the following formula:

[0127] ,

[0128] Let the first derivative of the energy functional with respect to the molten pool collapse deflection ω be... The extremum condition of the energy functional can be obtained as shown in the following equation: If the equation has real solutions, it indicates that the energy functional has a stationary point, and the corresponding free surface has a quasi-static equilibrium state that satisfies the Young-Laplace equation. If the equation has no real solutions, it indicates that the interface has undergone severe deformation, that is, the free surface is unstable.

[0129] By setting the discriminant for the equation to have real solutions ≥ 0, we can obtain the stability criterion as follows: The above is the derivation process of the discriminant. This equation reflects the maximum span allowed to exist on the free surface under a given external load. If this condition holds, the equation has real solutions, the free surface morphology is stable, and the weld pool is either incompletely penetrated or critically penetrated. If this condition does not hold, the equation has no real solutions, the free surface morphology cannot exist stably, and weld leakage occurs in the weld pool.

[0130] The parameters in the above stability criterion come from the following sources:

[0131] The external load ∆p has an arc pressure p arc Electromagnetic force p em Plasma flow force p plasma The static pressure p generated by the gravity of the molten pool g and droplet impact force p d ,

[0132] .

[0133] Where α is the equivalent coefficient, determined by the width W of the weld face. front With the width W of the back back The division yields a value representing the offsetting effect of the bevel on the molten pool support force during welding. Since the molten droplet acts at the center of the molten pool and is an intermittent force, it cannot offset the bevel support force, so no equivalent calculation was performed.

[0134] ,

[0135] ,

[0136] ,

[0137] ,

[0138] ,

[0139] In the above formula, μ0 is the free permeability, and r a Let r be the equivalent radius of the electric arc. m Let ρ be the equivalent radius of the conductive region of the molten pool, and take the arc radius as ρ. p v is the plasma density. p ρ is the axial jet velocity. l Let g be the density of the molten metal in the pool, g be the acceleration due to gravity, and v be the velocity of the molten metal. d This represents the droplet impact velocity.

[0140] Therefore, its core verification formula is: , Using the above methods, it can be calculated that γ is generally a constant in the industry or preset by engineers, while L is directly taken as the weld back width W obtained in box 1015. back .

[0141] The following examples will also use specific parameters.

[0142] First, the weld formation prediction model established in box 1015 is called, and the planned process parameters (I, U, v_weld, G) of each segment are input into the network to predict the forming width W_back and the excess height H_back of the weld back of the corresponding segment. The prediction results are shown in Table 6.

[0143] Table 6. Verification results of segmented calculation of penetration stability

[0144] 1 4.426 1.224 2 4.656 1.601 3 4.677 2.020 4 4.899 2.531 5 3.923 0.717 6 4.067 0.749 7 4.560 1.321

[0145] The applied load value calculated from the planned process parameters and the weld back width obtained from the weld formation prediction model are input into the formula. The calculation results obtained from the criterion model are shown in Table 7. It can be seen that the stability criterion results obtained from the segmented calculation are all greater than 0, that is, the welding process is stable and there will be no weld leakage or weld burn-through defects.

[0146] Table 7. Verification results of segmented calculation of penetration stability

[0147] 1 1221.350 0.994 2 1210.368 0.765 3 1159.211 0.978 4 1153.787 0.748 5 953.424 2.660 6 946.869 2.549 7 706.312 3.179

[0148] Weld penetration verification can distinguish between incomplete penetration and complete penetration. Furthermore, penetration stability verification can be used to verify the stability of the calculated complete penetration weld samples to distinguish between critical penetration and over-penetration, thereby ensuring the formation of a stable critical penetration weld during the welding process.

[0149] When the verification results show defects such as incomplete penetration or weld leakage, the process returns to the planning stage (box 102), adjusts the welding current, and outputs the result again for verification. This process is repeated multiple times until the verification is successful.

[0150] Box 104 is used to implement the welding execution of method 100.

[0151] Welding is automatically performed according to the planned process parameters to obtain a critical penetration weld.

[0152] Figure 3 A block diagram of an electronic device 300 that can implement various embodiments of the present disclosure is shown. For example... Figure 3 As shown, device 300 includes a processor 301, which can perform various appropriate actions and processes based on computer program instructions loaded into random access memory (RAM) 303 according to computer program instructions stored in read-only memory (ROM) 302. RAM 303 may also store various programs and data required for the operation of device 300. The processor 301, ROM 302, and RAM 303 are interconnected via bus 304. Input / output (I / O) interface 305 is also connected to bus 304.

[0153] The various processes and procedures described above, such as method 100, can be executed by processor 301. For example, in some embodiments, method 100 may be implemented as a software program tangibly contained in a machine-readable medium. In some embodiments, part or all of the software program may be loaded and / or installed on device 300 via ROM 302. When the software program is loaded into RAM 303 and executed by processor 301, one or more actions of method 300 described above may be performed.

[0154] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0155] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. Furthermore, although operations are depicted in a specific order, this should be understood as requiring that such operations be performed in the specific order shown or in sequential order, or requiring that all illustrated operations be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the foregoing discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0156] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for planning penetration parameters applicable to bevel welding, characterized in that, It includes the following steps: Multiple welding samples are collected. Each welding sample includes groove gap G, welding parameters, and penetration state. The penetration state includes at least an incomplete penetration state and a fully penetrated state. Based on the bevel gap and welding parameters of each welding sample, the corresponding criterion value for each welding sample is calculated. , in, This represents the equivalent average width of the bevel. For welding speed, The cross-sectional area of ​​the welding wire. The criterion value represents the wire feeding speed. It is used to reflect the ability of the electric arc to push aside the molten metal in the molten pool and sink; The criterion critical value is derived based on welding result samples of all incomplete penetration states and welding result samples of all penetration states. ; A penetration probability calculation model is established. The input data of the model is a binary vector containing the criterion value. The heat input per unit length of welding is Q, and the output is the penetration probability P; the penetration probability calculation model depends on the discriminant function. The discriminant function The method was established by minimizing the cross-entropy loss function through all welding samples; The penetration probability P and groove gap G required for actual welding are used as input values ​​and input into the penetration probability calculation model to obtain the corresponding welding parameters.

2. The penetration parameter planning method for bevel welding according to claim 1, characterized in that: Multiple welding samples were obtained through orthogonal experiments, and the welding parameters included at least the welding speed V. weld And welding current I.

3. The penetration parameter planning method for bevel welding according to claim 2, characterized in that: The penetration state includes critical penetration state, weld leakage state, and weld burn-through state. The criterion value for each welding sample is calculated accordingly. During the process, all welding samples at the critical penetration state were first screened out and regression analysis was performed. With the groove gap G as the independent variable, the corresponding welding current I, welding voltage U, and wire feed speed V were calculated. f .

4. The penetration parameter planning method for bevel welding according to claim 3, characterized in that: The criterion value The calculation method is as follows ,in, This represents the equivalent average width of the bevel. For welding speed, The cross-sectional area of ​​the welding wire. For wire feeding speed, criterion value The smaller the value, the stronger the lifting effect of the liquid metal on the electric arc; the criterion value. The larger the value, the weaker the lifting effect of the liquid metal on the electric arc.

5. The penetration parameter planning method for bevel welding according to claim 1, characterized in that: Using the data from the welding samples, the cross-entropy loss function is minimized. To calculate the optimal discriminant function parameters β0-β9, the discriminant function is then constructed. Finally, the penetration probability was calculated. .

6. The penetration parameter planning method for bevel welding according to claim 3, characterized in that: Using the welding samples as a dataset, a weld formation prediction model is constructed. The input data are welding current I, welding voltage U, and welding speed V. weld And the bevel gap G, the output data is the weld back width W back and weld back height H back The weld formation prediction model is configured to predict the width and height of the weld back by inputting the above input data.

7. The penetration parameter planning method for bevel welding according to claim 6, characterized in that: After obtaining the corresponding welding parameters through the penetration probability calculation model, the welding parameters need to be verified through penetration. The penetration verification is defined as the temperature in the root region of the bevel. It needs to be higher than the liquidus temperature of the metal. ,in, T0 represents the temperature at the root region of the bevel; T0 represents room temperature. It serves as the top heat source.

8. The penetration parameter planning method for bevel welding according to claim 7, characterized in that: After the penetration verification is passed, a further penetration stability verification is required. The penetration stability verification is configured to verify whether the back of the weld is stably formed, thereby distinguishing between critical penetration and over-penetration.

9. The penetration parameter planning method for bevel welding according to claim 8, characterized in that: The verification formula for the penetration stability verification is as follows: ,in, γ is the external load on the bottom of the molten pool, γ is the surface tension that maintains interface stability, and L is the span of the free interface at the bottom of the molten pool.

10. An electronic device, comprising a processor and a memory; the processor being connected to the memory; The memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to perform the method as described in any one of claims 1 to 9.

Images