A dissimilar metal laser welding parameter optimization method based on alloy powder filling

Abstract

The application discloses a kind of laser welding parameter optimization method of dissimilar metal filling based on alloy powder, for the laser welding of DP780 dual-phase steel and 5052 aluminum alloy, the added high-entropy alloy powder composition is FeCoNiCrTi, after designing test scheme and parameter, experimental result is obtained by laser welding test, the mathematical model that laser power, welding speed, defocusing amount, high-entropy alloy addition amount are input, and weld geometry is output is established using support vector machine optimized by raccoon optimization algorithm, and the optimal process parameters and high-entropy alloy addition amount are found in combination with multi-objective cuckoo search algorithm.Laser welding is carried out using the optimized welding process parameters and high-entropy alloy addition amount, which can significantly improve the mechanical properties of DP780 dual-phase steel and 5052 aluminum alloy welded joint, and can provide theoretical basis and data support for improving the performance of steel / aluminum welded joint and the research of high-entropy alloy as welding filler material.

Description

Technical Field

[0001] This invention relates to a method for optimizing laser welding process parameters of dissimilar metals, specifically a method for optimizing process parameters of laser welding of steel / aluminum dissimilar metals based on high-entropy alloy powder filling, belonging to the field of laser welding technology. Background Technology

[0002] With the continuous development of the automotive industry, energy consumption has increased dramatically, and environmental problems have become increasingly serious. Countries worldwide are placing greater emphasis on research into energy-saving and emission-reduction technologies, with automotive lightweighting emerging as a key approach to addressing these issues. Welding dissimilar metals like steel and aluminum can integrate the advantages of both, offering both safety and lightweighting, making it an effective means of achieving automotive lightweighting. However, achieving reliable welded connections between steel and aluminum is extremely difficult. The significant differences in physical properties between steel and aluminum not only hinder fusion but also lead to the formation of brittle intermetallic compounds (IMCs), deteriorating the joint's mechanical properties. IMC formation is primarily controlled by temperature and time. Laser welding, with its high energy density, allows for precise control of heat input and high cooling rates, which helps suppress IMC formation and growth. Therefore, laser welding has become an important method for aluminum / steel welded connections. In existing steel / aluminum laser deep penetration welding technologies, the performance of steel / aluminum joints is mainly improved by optimizing welding process parameters and by adding intermediate layer materials to alter the phase composition of Fe-Al compounds at the steel / aluminum interface.

[0003] In the research on optimizing welding process parameters, certain achievements have been made in optimizing the process parameters of laser welding of the same metal. For example, Ji et al. used response surface methodology to optimize the laser welding process parameters (welding current, welding speed, and laser arc distance) of 5083 aluminum alloy; Gao et al. combined Kriging model and non-dominated sorting genetic algorithm II to optimize the laser welding process parameters (laser power, welding current, laser arc distance, and welding speed) of 316L austenitic stainless steel. Compared with the optimization of laser welding process parameters of the same metal, the current research on optimizing the process parameters of laser welding of dissimilar metals such as steel and aluminum lacks research on combined parameters, which needs to be solved by combining mathematical models and multi-objective optimization algorithms.

[0004] In research on adding intermediate layer materials, researchers have tried adding Sn, Ni, Mn, etc., and although some progress has been made, a common problem is that many Fe-Al binary brittle phases still exist in the weld zone, which seriously affects the mechanical properties of the welded joint. High entropy alloys (HEAs) (composed of five or more main elements with equal or near-equal atomic ratios, and each component having an atomic fraction between 5% and 35% and a simple crystal structure) have high mixing entropy due to their multi-component nature. The constituent elements are freely and disorderly distributed, forming simple disordered solid solutions rather than complex phases or compounds, which can suppress the formation of IMC (high entropy effect). In addition, the hysteresis diffusion effect of high entropy alloys is due to the complexity of the chemical composition and the severity of lattice distortion, which makes atomic diffusion within the alloy extremely difficult. This effect can prevent the formation of a thick IMC layer at the interface during dissimilar metal welding. Although existing technologies report that adding high-entropy alloys during welding can suppress the formation of IMC, there is limited research on the addition of high-entropy alloys in the welding of dissimilar metals such as steel and aluminum, and corresponding theoretical research reports are scarce. On the other hand, adding a large amount of high-entropy alloy not only increases costs but also affects the weld penetration, leading to a decrease in mechanical properties. Conversely, adding a small amount of high-entropy alloy results in an insignificant effect in suppressing IMC formation. The lack of research on how to accurately control the amount of high-entropy alloy added and how to determine the optimal trade-off to achieve the best overall goal remains a challenge in the industry. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method for optimizing dissimilar metal laser welding parameters based on alloy powder filling. Specifically, for laser welding of DP780 duplex steel and 5052 aluminum alloy, the added high-entropy alloy powder composition is FeCoNiCrTi. The optimized parameters achieve the best weld pool depth-to-width ratio, reducing the formation of brittle intermetallic compounds and improving the mechanical properties of the weld joint. This provides a theoretical basis and data support for obtaining the optimal combination of process parameters for steel / aluminum dissimilar metal laser welding.

[0006] To achieve the above objectives, this method for optimizing dissimilar metal laser welding parameters based on alloy powder filling is applied to the laser welding of DP780 duplex steel and 5052 aluminum alloy. The added high-entropy alloy powder has the composition FeCoNiCrTi, and the method includes the following steps:

[0007] Step 1: Using three process parameters—laser power, welding speed, and defocusing amount—as well as the amount of high-entropy alloy added, as experimental factors, and with the minimum weld width and maximum weld depth as optimization objectives, an experimental scheme and parameters were designed, and laser welding experiments were conducted. The weld width and weld depth of the corresponding weld geometry were then measured.

[0008] Step 2: Using the data samples provided by the experimental scheme, the penalty parameter c and kernel function parameter g in SVM are selected as the search individuals of the raccoon population. The average relative error of the training samples is used as the objective function to optimize the penalty parameter c and kernel function parameter g, obtain the optimal hyperparameter combination, and establish a COA-SVM model that describes the nonlinear functional relationship between weld geometry, laser welding process parameters, and high-entropy alloy addition.

[0009] Step 3: The prediction functions provided by the COA-SVM model for laser power, welding speed, defocusing amount, and high-entropy alloy addition amount in relation to weld width and weld depth are used as the two objective functions of the MOCS algorithm to obtain the optimal solution set for laser power, welding speed, defocusing amount, and high-entropy alloy addition amount.

[0010] Step four: Finally, the CRITIC method is used to comprehensively evaluate each optimal solution using a comprehensive evaluation model to obtain the optimal solution.

[0011] Furthermore, in step one, the experimental design and parameters adopt orthogonal experimental design.

[0012] Furthermore, in step five, the expression for the comprehensive evaluation model is as follows:

[0013]

[0014] In the formula: Z represents the comprehensive evaluation value of each group of test schemes, W j y represents the weight value of the j-th evaluation indicator. ij Let y represent the value of the j-th evaluation index for the i-th scheme. i1 It represents the penetration depth of the welded joints in each group of test schemes, y i2 It is the weld width of the welded joint in each group of test schemes.

[0015] Compared with existing technologies, this method for optimizing dissimilar metal laser welding parameters based on alloy powder filler is applied to the laser welding of DP780 duplex steel and 5052 aluminum alloy. The added high-entropy alloy powder composition is FeCoNiCrTi. After designing the experimental scheme and parameters, experimental results were obtained through laser welding experiments. A mathematical model was established using a support vector machine (SVM) optimized by the Raccoon Optimization Algorithm (COA), with laser power, welding speed, defocusing amount, and high-entropy alloy addition amount as inputs and weld geometry as output. The optimal process parameters and high-entropy alloy addition amount were found by combining the multi-objective Cuckoo Search (MOCS) algorithm. This method can better predict the nonlinear relationship between laser welding process parameters, high-entropy alloy addition amount, and weld geometry. Using the optimized welding process parameters and high-entropy alloy addition amount, laser welding of DP780 duplex steel and 5052 aluminum alloy can significantly improve the mechanical properties of the laser-welded joint of DP780 duplex steel and 5052 aluminum alloy. This method can provide theoretical basis and data support for improving the performance of steel / aluminum welded joints and for research on high-entropy alloys as welding filler materials. Attached Figure Description

[0016] Figure 1 These are diagrams of tensile test specimens, where (a) is a dimensional diagram of the tensile specimen and (b) is a schematic diagram of the tensile test.

[0017] Figure 2 The results of COA-optimized SVM are shown in the figure, where (a) is the COA iteration process diagram and (b) is a comparison diagram of the COA melt width prediction value and the common 5-fold cross-validation SVM melt width prediction value with the actual melt width value.

[0018] Figure 3 This is a comparison chart of the predicted and actual values ​​of the COA-SVM model, where (a) is the melt width chart and (b) is the melt depth chart.

[0019] Figure 4 It is the Pareto optimal frontier plot;

[0020] Figure 5 The image shows the EDS composition analysis of the weld joint, where (a) is the scan site without high-entropy alloy, (b) is the scan site with high-entropy alloy, (c) is the element at position I, and (d) is the element at position II.

[0021] Figure 6 This is an EDS analysis location diagram of the welded joint, where (a) is without the addition of high-entropy alloy and (b) is with the addition of high-entropy alloy;

[0022] Figure 7 These are XRD analysis diagrams of welded joints, where (a) shows the joint without high-entropy alloy and (b) shows the joint with high-entropy alloy.

[0023] Figure 8These are displacement-tension diagrams of welded specimens before and after the addition of high-entropy alloy powder, where (a) is without high-entropy alloy and (b) is with high-entropy alloy.

[0024] Figure 9 These are fracture morphology images of tensile specimens before and after the addition of high-entropy alloy powder, where (a) is without the addition of high-entropy alloy and (b) is with the addition of high-entropy alloy. Detailed Implementation

[0025] This paper presents an optimization method for dissimilar metal laser welding parameters based on alloy powder filling, focusing on the laser welding of DP780 duplex steel and 5052 aluminum alloy. The study investigates the influence of pre-placed high-entropy alloy (FeCoNiCrTi) powder on the aluminum alloy surface on the microstructure and mechanical properties of the steel / aluminum dissimilar metal laser weld joint. A support vector machine (SVM) optimized by the Coati Optimization Algorithm (COA) is used to establish a mathematical model with laser power, welding speed, defocusing amount, and high-entropy alloy addition amount as inputs, and weld geometry as output. The optimal process parameters and high-entropy alloy addition amount are then found using a multi-objective cuckoo search (MOCS) algorithm. This method can provide a reference for improving the performance of steel / aluminum weld joints and for research on high-entropy alloys as welding filler materials. It also provides theoretical basis and data support for obtaining the optimal combination of process parameters for steel / aluminum dissimilar metal laser welding.

[0026] I. Test Materials and Test Methods

[0027] The experiment used DP780 duplex steel and 5052 aluminum alloy plates (90mm×65mm×1.3mm) for laser lap welding of dissimilar metals (steel on top, aluminum on the bottom). The added high-entropy alloy powder composition was FeCoNiCrTi. The laser used in the experiment was a JHM-1GX-1000F fiber laser welding machine with a maximum output power of 1000W, a laser wavelength of 1064nm, and a frequency of 50Hz. 99.9% argon gas was used as the shielding gas at a flow rate of 10L / min.

[0028] Before the experiment, the oxide film on the surface of the aluminum alloy plate was sanded off, and the oil stains on the surface of the aluminum alloy plate and steel plate were removed with acetone. The cleaned plates were assembled into lap joints. High-entropy alloy powder was weighed using an electronic scale with a range of 500g and an accuracy of 0.001g, dissolved in acetone solution, and thoroughly shaken and mixed until there was no precipitate. Then, it was evenly brushed onto the surface of the aluminum alloy plate.

[0029] Welding was performed using the laser welding process parameters and high-entropy alloy addition amount determined by the COA-optimized SVM model and MOCS optimization algorithm. A set of joints with the same welding process parameters but without the addition of high-entropy alloy powder was welded as a control group.

[0030] Post-weld wire-cut samples were polished and then analyzed using a field emission electron microscope equipped with an energy dispersive spectroscopy (EDS) to examine the interface morphology and main elements, and to analyze the influence of the hysteretic diffusion effect of the high-entropy alloy powder. The main phases of the weld joint IMC were analyzed using a D8 X-ray diffractometer (XRD). Wire-cut tensile specimens (e.g., according to GB / T 228.1-2021) were used. Figure 1 (As shown in a). During the test, shims of appropriate thickness are added to both ends of the specimen to ensure that the specimen is subjected to coaxial force (e.g., ...). Figure 1 (As shown in b), the loading speed was 0.5 mm / min. The tensile fracture morphology was observed using an electron microscope.

[0031] II. Selection of Process Parameters and High-Entropy Alloy Addition Amount

[0032] 1. Orthogonal experimental design

[0033] The weld depth-to-width ratio is the ratio of weld penetration depth to weld width. Under certain conditions, a larger depth-to-width ratio indicates better metallurgical bonding between the weld metal and the base metal, and higher mechanical properties of the joint. Therefore, weld penetration depth and weld width are selected as optimization targets. Three process parameters that significantly affect the quality of laser welding of dissimilar metals (steel / aluminum) are selected as experimental factors: laser power (x1, W), welding speed (x2, mm / s), and defocusing amount (x3, mm). Since the addition of high-entropy alloys absorbs some welding energy, resulting in a shallower weld penetration and decreased mechanical properties, the amount of high-entropy alloy added (x4, mg / mm) is controlled. 2 This was also considered as an experimental factor. The levels of each parameter were determined using a trial-and-error method. Within this range, no significant welding defects were observed with any combination of welding process parameters. The experimental factors and their levels are shown in Table 1 below.

[0034] Table 1. Experimental parameters and levels

[0035]

[0036] Based on the factors and levels in Table 1, L was selected. 25 (5 4 Welding process parameters were designed using an orthogonal matrix table. Laser welding of dissimilar steel / aluminum metals was performed with reference to the process parameters and high-entropy alloy addition amount in the orthogonal table. The weld geometry of the laser-welded joint was measured using ImageJ software to obtain the corresponding weld penetration (y1, mm) and weld width (y2, mm). The experimental parameters and results are shown in Table 2 below.

[0037] Table 2 Experimental parameters and results

[0038]

[0039]

[0040] 2. Establishment of the COA-SVM model

[0041] SVM is an important machine learning method that exhibits many unique advantages in solving small-sample, nonlinear, and high-dimensional pattern recognition problems. It overcomes the drawbacks of neural network learning methods, such as difficulty in determining network structure, slow convergence speed, overlearning, underlearning, and the need for large amounts of data during training. However, due to the small number of groups in this experiment, the average relative error obtained by the SVM model using the common 5-fold cross-validation method was only 17%, which is clearly insufficient and requires optimization of the model parameters.

[0042] COA, an optimization algorithm for simulating raccoon hunting behavior proposed by Dehghani Mohammad et al. in 2023, is characterized by strong evolutionary ability, fast convergence speed, and high convergence accuracy. The penalty parameter c and kernel function parameter g in the SVM are selected as the search individuals in the raccoon population. The average relative error of the training samples is used as the objective function to construct a COA-SVM model to iteratively optimize c and g. The COA iteration process is illustrated in the image below. Figure 2 As shown in Figure a. The 25 experimental groups were divided into training and test sets in an 8:2 ratio. The predicted values ​​(melt width) of the SVM model using COA parameter selection and common 5-fold cross-validation parameter selection were compared with the actual values. Figure 2 As shown in b. The error values ​​of the two models before and after optimization are shown in Table 3 below. As can be seen from Table 3, the mean absolute error (MAE) of the SVM model optimized using COA decreased by 68%, the root mean square error (RMSE) decreased by 67.5%, and the mean relative error decreased by 56.7%.

[0043] Table 3 shows the error values ​​of the two models before and after optimization.

[0044]

[0045] Two COA-SVM models (using COA optimization to obtain hyperparameter SVM models) were established with laser power, welding speed, defocusing amount, and high-entropy alloy addition amount as input variables, and weld width and weld depth as outputs, respectively. The trained COA-SVM models were analyzed, and the prediction results were compared with actual test results. The results are as follows: Figure 3 As shown.

[0046] The errors between the two models and the true values ​​can be calculated from the prediction results of the COA-SVM model, as shown in Table 4 below. Table 4 shows the R-squared values ​​of the two models. 2 All values ​​are greater than 0.9, indicating that the model has good learning ability. The root mean square error is no greater than 0.1, indicating that the model has low dispersion. The average relative error is less than 8%, which can accurately express the nonlinear relationship between laser welding process parameters and weld morphology.

[0047] Table 4. Accuracy Verification of COA-SVM Model

[0048]

[0049] 3. Multi-objective Cuckoo Search Algorithm

[0050] The Cuckoo Search Algorithm (CSA) is a bio-inspired intelligent optimization algorithm proposed by Yang et al. in 2010. This algorithm draws inspiration from the roosting and breeding behavior of cuckoos and the levy flight behavior of fruit flies. It has a simple structure, is easy to implement, and possesses excellent global search capabilities. Subsequently, in 2013, the Multi-Objective Cuckoo Search (MOCS) algorithm was proposed and applied to the field of multi-objective optimization.

[0051] To enable the optimization algorithm to simultaneously find the minimum of the objectives, the reciprocal of the weld penetration prediction function and the weld width prediction function are used as the objective functions of the algorithm. The prediction functions between welding process parameters and weld morphology geometry are provided by the COA-SVM model. The compromise solutions calculated by MOCS are all non-dominated solutions, meaning that no other solution is superior to these non-dominated solutions in both optimization objectives simultaneously. Figure 4 The diagram shows the Pareto optimal front composed of these non-dominated solutions.

[0052] 4. CRITIC Overall Evaluation

[0053] Figure 4 There are 50 Pareto optimal solutions in total, and the optimal solution needs to be selected by comprehensively evaluating each group of solutions. The CRITIC method, proposed by Diakoulaki, is an indicator weighting method based on data contrast and contradiction indicators. It is suitable for data where there is a certain correlation between indicators and factors and the volatility is small.

[0054] Assigning appropriate weights to each Pareto optimal solution, the expression for the comprehensive evaluation model is as follows:

[0055]

[0056] In the formula: Z represents the comprehensive evaluation value of each group of test schemes, W j y represents the weight value of the j-th evaluation indicator. ij Let y represent the value of the j-th evaluation index for the i-th scheme. i1 It represents the penetration depth of the welded joints in each group of test schemes, y i2 It is the weld width of the welded joint in each group of test schemes.

[0057] Fifty Pareto optimal solutions were obtained from MOCS. Based on the CRITIC comprehensive evaluation score of each solution, solution group 23 had the lowest comprehensive evaluation score among the 50 groups, corresponding to a weld width of 0.64 mm and a weld depth of 1.13 mm. The laser welding process parameters at this point were: laser power 649.7 W, welding speed 30.5 mm / s, defocusing amount 5.24 mm, and high-entropy alloy addition amount 0.34 mg / mm². 2 .

[0058] 5. Optimize result verification

[0059] The Pareto solution set predictions and the results before and after optimization are shown in Table 5 below. The errors between the test results and the Pareto optimal solution are 3.03% and 3.67%, respectively, both below 4%, indicating that the optimization method is relatively accurate. The optimized weld width is 0.66 mm and the weld depth is 1.09 mm. Before optimization (process parameters: laser power 700 W, welding speed 34 mm / s, defocusing amount 10 mm, high-entropy alloy addition 0.4 mg / mm), the results were significantly different. 2 The weld width was 0.89 mm and the weld depth was 0.84 mm. Compared with the original result, the weld width decreased by 28.06% and the weld depth increased by 29.76%, improving the aspect ratio and verifying the feasibility of the optimization method.

[0060] Table 5 Comparison and Analysis of Optimization Results

[0061]

[0062] III. The Influence of High-Entropy Alloys on Laser Welded Joints of Steel / Aluminum Dissimilar Metals

[0063] 1. EDS Analysis

[0064] Elemental line scan results of steel / aluminum welded joints before and after adding high-entropy alloy powder are as follows: Figure 5 As shown. From I ( Figure 5 c) and II ( Figure 5 Line scan results at position d) show that due to the high-entropy effect and hysteresis diffusion effect of the high-entropy alloy, the relative intensity of aluminum diffusion towards the steel side decreases from approximately 300 to approximately 200, and the width of inter-diffusion between steel and aluminum decreases from approximately 330 μm to approximately 300 μm. The magnified image of the steel / aluminum interface shows a significant reduction in the thickness of the brittle needle-like IMC layer between steel and aluminum, which will improve the mechanical properties of the joint.

[0065] 2. Phase structure of the interface layer

[0066] To analyze the phase composition of the steel / aluminum welded joint interface, energy dispersive spectroscopy (EDS) analysis was performed on different regions of the steel / aluminum interface before and after the addition of the high-entropy alloy. The analysis locations are shown in the figure. Figure 6As shown in Table 6, the results indicate that the atomic ratio of Fe to Al in region A is close to 3:1, in region B it is close to 1:1, and in region C it is close to 1:5. Based on the Fe-Al phase diagram, it is inferred that the Fe3Al phase forms in region A, the FeAl phase forms in region B, and the FeAl phase forms in region C. 3.2 It is composed of FeAl6; the atomic ratio of Fe to Al in region D is close to 2:3, in region E it is close to 2:5, and in region F it is close to 2:9. Based on the Fe-Al phase diagram, it is inferred that region D is composed of FeAl and FeAl2, and region E is composed of FeAl6. 3.2 It is composed of FeAl2 and FeAl6 and Fe4Al. 13 composition.

[0067] Table 6. EDS Point Analysis of Steel / Aluminum Interface Elements

[0068]

[0069]

[0070] To further determine the phase structure type of the welded joint, such as Figure 7 As shown, X-ray diffraction (XRD) was performed on the steel / aluminum welded specimens. Without the addition of high-entropy alloy powder, intermetallic compounds such as FeAl and Fe3Al (e.g., ...) were formed at the joint interface. Figure 7 a) Adding high-entropy alloy powder causes FeAl2 and Fe4Al to form at the joint interface. 13 Al5FeNi,Co3Fe, and Al 0.983 Cr 0.017 ( Figure 7 b) This is basically consistent with the energy dispersive spectroscopy (EDS) analysis results of the steel / aluminum joint interface layer mentioned above. The complexity of the chemical composition and the severity of lattice distortion in high-entropy alloys make atomic diffusion within the alloy extremely difficult. This reduces the bonding between Fe and Al, and some Fe / Al reacts with elements in the high-entropy alloy to form Co3Fe and Al. 0.983 Cr 0.017 Isophase. Some high-entropy alloying elements participate in the reaction between Fe and Al to form new phases such as Al5FeNi, which are formed by Ni replacing the iron element in the original brittle Fe2Al5. Compared with brittle phases such as FeAl and Fe2Al5, these new phases have increased ductility and can improve the mechanical properties of steel / aluminum welded joints.

[0071] 3. Analysis of the mechanical properties of the joint

[0072] Due to the presence of brittle intermetallic compounds (IMCs) between steel and aluminum, all welded specimens fractured at the steel-aluminum interface of the weld. Displacement-tensile force diagrams of the welded specimens before and after the addition of high-entropy alloy powder are shown below. Figure 8 As shown. By Figure 8As can be seen from a, the maximum tensile force without the addition of high-entropy alloy powder is 603 N, and the absence of a yield stage indicates brittle fracture; [the text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Figure 8 As can be seen from b, after adding high-entropy alloy powder, the maximum tensile force can reach 815N. Compared with the absence of high-entropy alloy powder, the maximum tensile force of the welded joint increased by 35.2%, and the yielding stage appeared, indicating that the welded joint exhibits toughness characteristics.

[0073] Fracture morphology of tensile specimens before and after adding high-entropy alloy powder is as follows: Figure 9 As shown, without the addition of high-entropy alloy powder, the microstructure of the fracture surface at the joint exhibits a river-like pattern, displaying brittle fracture characteristics. Figure 9 a) Adding high-entropy alloy powder resulted in a small number of dimples and tear ridges on the joint fracture surface, exhibiting quasi-cleavage fracture characteristics. Figure 9 b).

[0074] In summary, when laser welding DP780 duplex steel to 5052 aluminum alloy, adding high-entropy alloy powder with the composition FeCoNiCrTi can effectively suppress the interdiffusion of steel and aluminum, reduce the formation of brittle IMC, and simultaneously generate structurally stable Al5FeNi,Co3Fe, and Al 0.983 Cr 0.017 The isoductile new phase can significantly improve the mechanical properties of the laser-welded joint between DP780 duplex steel and 5052 aluminum alloy. At the same time, the SVM model optimized by COA can better predict the nonlinear relationship between laser welding process parameters, high-entropy alloy addition, and weld geometry, thereby determining the optimal trade-off to achieve the best overall goal. This can provide a theoretical basis and data support for obtaining the optimal combination of process parameters for laser welding of dissimilar metals such as steel and aluminum.

Claims

1. A method for optimizing parameters in dissimilar metal laser welding based on alloy powder filling, characterized in that, For laser welding of DP780 duplex steel and 5052 aluminum alloy, the added high-entropy alloy powder composition is FeCoNiCrTi, and the specific steps include: Step 1: Using three process parameters—laser power, welding speed, and defocusing amount—as well as the amount of high-entropy alloy added, as experimental factors, and with the minimum weld width and maximum weld depth as optimization objectives, an orthogonal experimental design was adopted to design the experimental scheme and parameters, and laser welding experiments were conducted. The weld width and weld depth of the corresponding weld geometry were then measured. Step 2: Using the data samples provided by the experimental scheme, the penalty parameter c and kernel function parameter g in SVM are selected as the search individuals of the raccoon population. The average relative error of the training samples is used as the objective function to optimize the penalty parameter c and kernel function parameter g to obtain the optimal hyperparameter combination. A COA-SVM model is established to describe the nonlinear functional relationship between the weld geometry and the laser welding process parameters and the amount of high-entropy alloy added. Step 3: The prediction functions provided by the COA-SVM model for laser power, welding speed, defocusing amount, and high-entropy alloy addition amount in relation to weld width and weld depth are used as the two objective functions of the MOCS algorithm to obtain the optimal solution set for laser power, welding speed, defocusing amount, and high-entropy alloy addition amount. Step four: Finally, the CRITIC method is used to comprehensively evaluate each optimal solution using a comprehensive evaluation model to obtain the optimal solution. The expression for the comprehensive evaluation model is as follows: In the formula: This represents the comprehensive evaluation value of each group of test schemes. Representing the The weight values ​​of each evaluation indicator, Indicates the first Group Scheme No. The values ​​of each evaluation indicator, It represents the penetration depth of the welded joints in each group of test schemes. It is the weld width of the welded joint in each group of test schemes.

Images