An electric arc additive manufacturing weld bead forming interlayer regulation method and system

Abstract

The application discloses an interlayer regulation method and system for forming a welding bead in electric arc additive manufacturing, and relates to the technical field of electric arc additive manufacturing. The technical scheme is characterized in that high-precision point cloud data is obtained by scanning the surface of a current deposition layer. A neural network prediction module uses a welding speed sequence of a target deposition layer as input, combines a trained neural network model, predicts feature point data of the target deposition layer, and constructs a three-dimensional model of the target deposition layer. An interlayer regulation module adjusts process parameters according to a set threshold value based on the deviation between the prediction result and the target result, so as to realize accurate interlayer shape regulation. Compared with the prior art, the application has strong process adjustment capability, can significantly optimize the forming quality of each layer in the additive manufacturing process, and improves the manufacturing precision of the overall component.

Description

Technical Field

[0001] This invention relates to the fields of robot path planning, additive manufacturing (AM) and intelligent manufacturing technology, and in particular to a method and system for controlling the interlayer formation of weld beads in arc additive manufacturing. Background Technology

[0002] Arc additive manufacturing (AED) is one of the fastest-growing metal additive manufacturing methods in recent years. This technology discretizes complex metal structures into simple planar stacking forms by layering and slicing a three-dimensional model of the metal component and planning its path. Then, an electric arc heat source is used to melt metal wires layer by layer for deposition, significantly reducing the difficulty of forming high-performance metal structural components. Compared with other metal additive manufacturing technologies, AED has significant advantages such as short production cycle, low manufacturing cost, and high deposition efficiency. Furthermore, the formed components are characterized by uniform chemical composition, high density, and excellent mechanical properties. Currently, this technology has become an important driving force in industries such as aerospace, shipbuilding, chemical engineering, and automotive, and its application is gradually developing towards larger and more integrated metal structural components.

[0003] However, the arc additive manufacturing process is affected by various factors (such as deviations in the surface morphology of a single layer, surface deformation caused by high-temperature heating, and random flow of the molten pool). The shape error variation trend between the deposited layer after multi-layer deposition and the ideal model is complex, and it is difficult to achieve the goal of large and complex structures by relying solely on parameter optimization during the design stage. Therefore, how to effectively control the interlayer forming accuracy, reduce the accumulation of deviations, and ensure the final quality of additive components remains an urgent problem to be solved in the field of arc additive manufacturing. Summary of the Invention

[0004] This application provides a method and system for controlling the interlayer forming of weld beads in arc additive manufacturing. The technical purpose is to effectively control the interlayer forming accuracy of weld beads in additive manufacturing, reduce the accumulation of interlayer deviations, and ensure the final quality of additive components.

[0005] The above-mentioned technical objective of this application is achieved through the following technical solution:

[0006] A method for controlling interlayer formation in arc additive manufacturing weld beads, comprising the following steps:

[0007] Step S1: Collect point cloud data of the current weld layer;

[0008] Step S2: Design and train the neural network prediction module, input the target weld layer process parameters into the neural network prediction module, and output the target weld layer feature data; add the target weld layer feature data with the current weld layer point cloud data to obtain the prediction layer point cloud data;

[0009] Step S3: Based on the deviation between the target model and the predicted layer point cloud data, adjust and optimize the process parameters of the target cladding layer, and transmit the optimized process parameters to the arc additive manufacturing equipment to perform the target layer deposition.

[0010] Furthermore, the specific steps in step S2 are as follows:

[0011] Step S21: Train the artificial neural network model to obtain the prediction module;

[0012] Step S22: Input the target weld layer process parameters, i.e. the velocity sequence of the current weld cross section, into the weld cross section prediction unit in the neural network prediction module, and output the target weld layer feature data;

[0013] Step S23: Correct the weld overlap;

[0014] Step S24: Connect the feature points at corresponding positions of different weld cross sections according to the predicted feature points and correction results to obtain a three-dimensional model of the weld. Obtain the target weld layer feature data based on the three-dimensional model of the weld.

[0015] Furthermore, the specific steps of step S2 are as follows:

[0016] Step S221: Initialize parameters j = 1, h = 1, define j as the weld bead number, j ∈ {1, 2, ..., j ..., J}, J as the total number of weld beads, h as the slice number of each weld bead, h ∈ {1, 2, ..., h ..., H}, H as the total number of slices of each weld bead;

[0017] Step S222: Input the velocity sequence of the current weld cross section into the trained artificial neural network model, and output the height axis increment of the corresponding positions of these five feature points relative to the current layer substrate;

[0018] The speed sequence includes: the welding speed parameter V0 at the position corresponding to the current j-th pass and h-th slice, and the welding speed parameters V0 at the positions of the two slices preceding the predicted weld cross-section. -2 and V -1 The welding speed parameters V1 and V2 are predicted for the locations of the two slices following the weld bead cross-section; the height axis increment is defined as...

[0019] Step S223: Select the current base contour feature point, add the horizontal coordinate value of the base contour feature point to the horizontal coordinate value of the current welding gun projection point to obtain the horizontal coordinate value of the predicted point; add the height coordinate value of the base contour feature point to the height axis increment to obtain the height coordinate value of the predicted point, and then obtain the coordinates of the predicted point and save the data.

[0020] Step S224: h = h + 1, repeat steps S222-S223 until h = H;

[0021] Step S225: j = j + 1, repeat steps S222-S224 until j = J, thereby obtaining the predicted feature points of the target weld layer of J welds in J*H slices.

[0022] Furthermore, in step S223, the welding torch projection point P j,h The absolute coordinates of the projection of the intersection point of the additive path of the target weld layer and the cross-sectional slice of the weld bead of the target weld layer onto the current weld layer are defined as P. j,h (X P Z P Extract the projection point P on the current weld cross section relative to the welding torch. j,h The horizontal coordinate distance is 0、 The five points are used as the current base contour feature points. and The height direction coordinates of the base contour feature points are defined as follows: The coordinates of the five base contour feature points are as follows: Where d is the additive manufacturing path spacing distance;

[0023] That is, the coordinates of the predicted feature point in the j-th layer h are:

[0024] Furthermore, in step S23, the specific steps for correcting the weld overlap are as follows:

[0025] Step S231: Initialize j = 1, h = 1.

[0026] Step S232: Correct the height direction coordinate value of the overlap portion of the h-th slice section of weld j and weld j+1, as follows:

[0027]

[0028] in, The height coordinate of the fifth weld bead profile feature point on the cross-section of the h-th weld bead of the j-th weld bead is given. The height coordinate of the first weld bead profile feature point on the cross-section of the (j+1)th weld bead and the h-th weld bead section is given. This represents the height coordinate of the overlap between the j-th and j+1-th layers of the h-th slice in the current weld layer.

[0029] Further, step S3 includes:

[0030] Step S31: The target weld layer height is L. Calculate the deviation θ1 between the target model and the predicted layer point cloud data, and compare it with a preset threshold. If the deviation θ1 is less than the preset threshold α1, proceed to step S33; if the deviation θ1 is greater than or equal to the preset threshold α1, calculate the deviation DEV for each weld cross-section. j,h Perform calculations;

[0031] Determine the deviation (DEV) of each weld cross section. j,h Is it greater than the adjustment threshold α2? For the deviation DEV j,h Weld cross sections with values ​​less than or equal to the adjustment threshold α2 maintain the original welding parameters; the deviation DEV for each weld cross section... j,h Weld cross sections exceeding the adjustment threshold are categorized according to deviation DEV. j,h Welding speed V in descending order j,h Make corrections;

[0032] Step S32: Based on the adjusted welding speed, repeat step S2 to obtain the predicted layer point cloud data again, input it into step S31, until the deviation between the target model and the predicted layer point cloud data meets the preset threshold requirement.

[0033] Step S33: Input the adjusted target cladding layer process parameters into the arc additive manufacturing equipment to perform target cladding layer deposition.

[0034] Furthermore, the deviation DEV of each weld cross section in step S31 j,h The calculation formula is as follows:

[0035]

[0036] in, Z represents the height coordinate value of the i-th feature point on the cross-section of the j-th and h-th weld bead sections. target DEV represents the coordinate value in the height direction of the target weld layer. j,h This indicates the deviation between the cross section of the j-th weld bead and the target weld layer.

[0037] Furthermore, regarding the welding speed V j,h The formula for correction is:

[0038] V′ j,h =V j,h ±DEV j,h *α

[0039] Among them, V' j,h V represents the adjusted welding speed of the h-th weld bead cross section in the j-th pass. j,h This represents the original welding speed of the h-th weld bead section in the j-th pass, and α represents the correction coefficient.

[0040] A neural network-based additive manufacturing weld bead forming interlayer control system, comprising:

[0041] The scanning module acquires point cloud data of the current weld layer surface;

[0042] The neural network prediction module outputs target weld layer feature data based on the input target weld layer process parameters.

[0043] The accumulation module adds the target weld layer feature data to the current weld layer point cloud data to obtain the prediction layer point cloud data;

[0044] The interlayer control module calculates the deviation between the target model and the predicted layer point cloud data. If the deviation exceeds the preset threshold, the target cladding layer process parameters are adjusted until the deviation meets the preset threshold requirement. The adjusted target cladding layer process parameters are then input into the arc additive manufacturing equipment to perform the target cladding layer deposition.

[0045] A robotic arc additive manufacturing apparatus includes a welding robot, an argon arc welding torch fixed to the end effector of the welding robot, a 3D scanner, an industrial control computer, an electric welding machine, and a robot controller. The 3D scanner collects current point cloud data of the arc additive manufacturing structure, and the industrial control computer receives and processes the data.

[0046] The beneficial effects of this application are as follows:

[0047] (1) Interlayer control of weld bead forming height in arc additive manufacturing is achieved: This application predicts the interlayer morphology during the additive manufacturing process, obtains three-dimensional morphology prediction data of the weld bead, and adjusts the welding speed corresponding to each cross-section of each weld bead in real time, which can reduce the morphology deviation caused by the cumulative effect of interlayer errors. Through this precise interlayer control, the accuracy loss caused by layer-by-layer stacking in traditional methods can be effectively avoided, thereby improving the overall quality and reliability of the formed component.

[0048] (2) Welding speed sequence as input, considering the spatial influence of the target weld layer process parameters: This application innovatively uses the welding speed sequence of each cross section of each weld pass as the neural network input, and fully considers the spatial correlation of the target weld layer process parameters in the same weld pass. This enables fine-tuning of the welding speed process parameters at multiple locations along the additive manufacturing path, thereby improving the operational flexibility and adaptability in the additive manufacturing process and further enhancing manufacturing accuracy.

[0049] (3) Combining neural network prediction with 3D model construction to accurately predict weld bead morphology: Unlike traditional techniques that only predict the height and width of the weld section, this application proposes a strategy that combines neural network prediction and 3D modeling, which can more accurately simulate and reconstruct the real 3D morphology of the weld bead. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the robotic arc additive manufacturing apparatus in one of the embodiments of this application;

[0051] Figure 2 This is a schematic diagram of the neural network prediction module in an embodiment of this application;

[0052] Figure 3 This is a flowchart illustrating the interlayer control process for additive manufacturing weld bead formation in an embodiment of this application.

[0053] Figure 4 This is a schematic diagram showing the selection of the welding torch projection point in an embodiment of this application;

[0054] Figure 5 This is a schematic diagram of the structural component planning path in the embodiments of this application;

[0055] Figure 6 This is a schematic diagram of the three-dimensional modeling process of the weld layer in an embodiment of this application;

[0056] Figure 7 This is a schematic diagram of the interlayer control module in an embodiment of this application.

[0057] Among them, 1-welding robot, 2-argon arc welding gun, 3-electric arc additive manufacturing platform, 4-electric arc additive manufacturing structural component, 5-3D scanner, 6-industrial control computer, 7-electric welding machine, 8-robot controller. Detailed Implementation

[0058] To enhance understanding of the present invention, the following will be discussed in conjunction with the accompanying drawings. Figure 1-7 The present invention will be further described in detail below. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of this application.

[0059] To reduce the impact of interlayer error accumulation on the forming of arc additive manufacturing structural parts, this application provides a method for controlling interlayer formation in arc additive manufacturing weld beads. This method involves, as follows: Figure 1 The illustrated robotic arc additive manufacturing apparatus includes a welding robot 1, an argon arc welding torch 2 fixed to the end effector of the welding robot 1, a substrate 3, a 3D scanner 5, an industrial computer 6, a welding machine 7, and a robot controller 8. The 3D scanner 5 acquires the current point cloud data of the arc additive manufacturing structure 4. The industrial computer 6 receives and processes the data, obtains a regulated welding speed sequence using the arc additive manufacturing weld bead forming interlayer control method described in this application, and sends the corrected additive welding process parameters to the robot controller 8 and the welding machine 7.

[0060] For arc additive manufacturing structural component 4 (length 80mm * width 80mm * height 80mm), the layer thickness is set to Lmm; XYZ is the Cartesian coordinate system for arc additive manufacturing, where X is the horizontal direction, Y is the vertical direction, and Z is the height direction.

[0061] like Figure 2 As shown, the additive manufacturing path is divided into J additive manufacturing paths along the X-axis, with an interval of dmm between the additive manufacturing paths, and the additive manufacturing direction is the positive Y-axis direction; ER308L stainless steel with a diameter of 1.0mm is selected as the welding wire, the welding current is 115A, the wire feeding speed is 0.6m / min, and the welding speed adjustment range is set to 20-30cm / min.

[0062] like Figure 3 As shown, the method for controlling interlayer formation of weld beads in arc additive manufacturing according to this application includes:

[0063] Step S1: Collect point cloud data of the current weld layer using a 3D scanner.

[0064] The number of slice sections H is calculated based on the weld length and the interval distance d of the additive path. Each weld is sliced ​​along the Y-axis of the additive direction based on the number of slice sections H, and the cross-sectional profile of each slice is obtained.

[0065] Define j as the weld bead number, j∈{1,2,.j..,J}, J as the total number of weld beads, h as the slice number of each weld bead, h∈{1,2,.h..,H}, H as the total number of slices of each weld bead;

[0066] like Figure 4 As shown, the projection point P of the welding torch j,h The absolute coordinates of the intersection point of the additive path of the target weld layer and the cross-sectional slice of the weld bead of the target weld layer projected onto the current weld layer. The welding torch projection point indices p∈{P} for different weld beads and different slices are assigned. 1,1 P 1,2 , ..., P j,h ,...,P J,H Let P be the set of all welding torch projection point coordinate data, and let P be the set of all welding torch projection point coordinate data. The weld bead section base contour line is the contour line of the weld bead section slice on the base point cloud data.

[0067] In this embodiment, the number of slices for each weld pass is H, and the target weld layer passes are J. Therefore, the additive path consists of a dataset of H*J slices representing the weld gun projection point data. Read and save. Simultaneously, obtain the base contour line where the welding torch projection point is located through 3D scanning, which will be used to subsequently correct the feature points of the weld cross-section. Proceed to step S2;

[0068] Step S2: Design and train the neural network prediction module, input the target weld layer process parameters into the neural network prediction module, and output the target weld layer feature data.

[0069] Step S21: Train the artificial neural network model to obtain the prediction module;

[0070] (1) Sample set construction:

[0071] The target weld layer process parameters are selected for input to the neural network, specifically including: the welding speed parameter V0 at the predicted weld cross-section location, and the welding speed parameter V at the locations of the two preceding slices of the predicted weld cross-section. -2 and V -1 The system predicts the welding speed parameters V1 and V2 at the locations of the two sections following the weld cross section (if the current section does not meet the condition of having two cross sections before and after it, the welding speed parameters at the locations of the missing cross sections are filled with 0). This incorporates the spatial influence of process parameters during arc additive manufacturing, making it possible to finely control the welding speed parameters at the corresponding locations of each cross section.

[0072] Select the cross-sectional feature points of the output neural network: Slice each weld bead along the additive manufacturing direction Y-axis according to the number of slice sections H, and obtain the cross-sectional contour of each slice. Construct a coordinate system and project the welding torch point P of the current slice. j,h As the origin, select point P on the current base contour relative to the welding torch projection. j,h Five points with horizontal offsets of -d / 2, -d / 4, 0, d / 4, and d / 2 were designated as base contour feature points. Five points on the predicted weld bead cross-section profile corresponding to the X-coordinates of these base contour feature points were defined as weld bead cross-section feature points. The weld bead height increment ΔZ relative to the base contour feature points was used as output data. Selecting five feature points of the weld bead cross-section, compared to predicting only weld bead height and width, more accurately characterizes the geometric morphology of the weld bead cross-section, making the model more closely resemble actual additive manufacturing effects. Furthermore, using the weld bead height increment as an output parameter effectively depicts the correspondence between the target weld layer and the current layer point cloud data, thus accurately reflecting the cumulative effect of interlayer errors.

[0073] (2) Neural Network Model Construction

[0074] The construction of an artificial neural network structure includes an input layer, an output layer, and a hidden layer. The hidden layer contains several neurons. The input process parameters and the weld cross-section feature points obtained after processing the point cloud data are taken as a whole sample set. The sample set is then divided into a training set and a test set. The artificial neural network is trained by substituting into the training set, and the test set is used to verify the generalization ability of the constructed artificial neural network model.

[0075] The rationality and accuracy of the artificial neural network structure were verified using mean square error and coefficient of determination R².

[0076] The structure of the neural network prediction module is as follows: Figure 5 As shown, the neural network prediction module includes a weld cross-section prediction unit and a 3D modeling unit for the weld layer.

[0077] Step S22: Input the target weld layer process parameters into the weld cross-section prediction unit in the neural network prediction module, and output the target weld layer feature data:

[0078] Step S221: Initialize parameters, j = 1, h = 1.

[0079] Step S222: Input the velocity sequence of the current weld bead (the current weld bead is the predicted weld bead) cross-section into the trained artificial neural network model, and output the height (Z) axis increment of the corresponding positions of these five feature points relative to the current layer substrate; wherein the velocity sequence of the current weld bead cross-section includes: the welding velocity parameter V0 at the position corresponding to the h-th slice of the j-th weld bead, and the welding velocity parameters V0 at the positions of the first two slices of the predicted weld bead cross-section. -2 and V -1 Predict the welding speed parameters V1 and V2 at the locations of the two slices following the weld cross section; (If the current slice does not meet the condition of having two cross sections before and after it, fill in the missing welding speed parameters at the locations of the missing cross sections with 0).

[0080] Input the velocity sequence of the current weld cross-section into the trained artificial neural network model, and output the height (Z-axis increment of the corresponding positions of these five feature points relative to the current layer substrate, defined as...

[0081] Step S223: Define the current welding torch projection point coordinates as P j,h (X P Z P Extract the projection point P on the current weld cross section relative to the welding torch. j,h The horizontal (X) coordinate distance is 0、 The five points are taken as the current base contour feature points, and the Z coordinate of the base contour feature points is defined as... Then the coordinates of each point are Where d is the additive manufacturing path spacing distance.

[0082] Read the projection point P of the welding torch j,h coordinates (X) P Z P )as well as Will and Z P , Add them together, and then add the horizontal (X) coordinate values ​​of the weld cross-section projection point to the welding torch projection point P. j,h Horizontal coordinates (X) P The coordinates of the predicted feature point in the j-th layer and h-th layer are obtained by adding them together: The height direction coordinate value of the i-th predicted feature point in the h-th layer of the j-th channel is defined as... (i = 1, 2, 3, 4, 5), save the data.

[0083] Step S224: h = h + 1, repeat steps S222-S223 until h = H;

[0084] Step S225: j = j + 1, repeat steps S222-S224 until j = J, thereby obtaining the predicted feature points of the target weld layer of J welds in J*H slices.

[0085] Step S23: Correct the weld overlap;

[0086] Step S231: Initialize j = 1, h = 1.

[0087] Step S232: Correct the height direction (Z) coordinate value of the overlap portion of slices j and j+1 h, as shown below:

[0088]

[0089] in, The Z-coordinate value of the fifth weld bead profile feature point on the cross-section of the h-th weld bead of the j-th weld bead. The Z-coordinate value of the first weld bead profile feature point on the cross-section of the h-th weld bead of the (j+1)-th weld bead. This is the Z-coordinate value of the overlap between the j-th and j+1-th layers of the h-th slice on the current weld layer.

[0090] Step S224: h = h + 1, repeat step S232 until h = H;

[0091] Step S225: j = j + 1, repeat steps S232-S233 until j = J - 1, thereby obtaining the predicted point cloud data after the target cladding layer is corrected.

[0092] Step S24: As Figure 6 As shown, based on the predicted feature points and the correction results, the feature points at corresponding positions of different weld cross sections are connected to obtain a three-dimensional weld model. The target weld layer feature data are then obtained based on the three-dimensional weld model. Proceed to step S3;

[0093] Step S3: By calculating the deviation between the target model and the predicted layer point cloud data, when the deviation exceeds a preset threshold, the system dynamically adjusts the process parameters of the target weld layer until the accuracy requirements are met. The optimized process parameters are then transmitted to the arc additive manufacturing equipment to perform the target layer deposition. The control strategy based on weld cross-sectional characteristics makes process parameter adjustment more precise and flexible, enabling targeted optimization for specific errors in each weld cross-section, effectively solving the problem of insufficient component forming accuracy caused by the accumulation of interlayer errors.

[0094] Specifically, step S3 includes:

[0095] Step S31: The target cladding layer height is L. Calculate the deviation θ1 between the target model and the predicted layer point cloud data, and compare it with a preset threshold. If the deviation θ1 is less than the preset threshold α1, proceed to step S33.

[0096] If the deviation θ1 is greater than or equal to the preset threshold α1, such as Figure 7 As shown, the deviation DEV for each weld cross section j,h Calculate the deviation (DEV). j,h The calculation formula is:

[0097]

[0098] Deviation of each weld cross section (DEV) j,h The weld cross section is judged to be greater than or equal to the adjustment threshold α2; if it is less than or equal to the adjustment threshold α2, the original welding parameters are maintained.

[0099] Deviation of each weld cross section (DEV) j,h Weld cross sections exceeding the adjustment threshold are categorized according to deviation DEV. j,h Welding speed V in descending order j,h The correction is expressed as:

[0100] V′ j,h =V j,h ±DEV j,h *α

[0101] in, Z represents the Z-value of the i-th feature point on the cross-section of the j-th and h-th weld beads. target DEV represents the Z-value of the target weld layer. j,h V' represents the deviation between the cross section of the j-th weld bead and the target weld layer. j,h V represents the adjusted welding speed of the h-th weld bead cross section in the j-th pass. j,h This represents the original welding speed of the h-th cross section of the j-th weld pass, and α represents the correction factor.

[0102] Step S32: Based on the adjusted welding speed, repeat step S2 to obtain the predicted layer point cloud data again, input it into step S31, until the deviation between the target model and the predicted layer point cloud data meets the preset threshold requirement.

[0103] Step S33: Input the adjusted target cladding layer process parameters into the arc additive manufacturing equipment to perform target cladding layer deposition.

[0104] The present application discloses an arc additive manufacturing weld bead forming interlayer control system, which is used to implement the additive manufacturing weld bead forming interlayer control method described in this application.

[0105] The control system includes a scanning module, a neural network prediction module, an accumulation module, and an inter-layer control module.

[0106] The scanning module is used to acquire point cloud data of the current weld layer surface;

[0107] The neural network prediction module is used to output the feature data of the target weld layer based on the input target weld layer process parameters;

[0108] The interlayer control module is used to calculate the deviation between the target model and the predicted layer point cloud data. If the deviation exceeds the preset threshold, the process parameters of the target cladding layer are adjusted until the deviation meets the preset threshold requirement. The adjusted process parameters of the target cladding layer are then input into the arc additive manufacturing equipment to perform the deposition of the target cladding layer.

[0109] Example

[0110] like Figure 1-7 As shown, in this embodiment, the number of slices for each weld pass is H=9, and the target weld layer is J=8 passes. Therefore, the dataset contains the welding gun projection point data for a total of 72 slices in the additive manufacturing path. Read and save. Simultaneously, obtain the base contour line where the welding torch projection point is located through 3D scanning, which will be used to subsequently correct the feature points of the weld cross-section;

[0111] Select the X-coordinate distances on the weld cross-section relative to the welding torch projection point P as -5, -2.5, 0, 2.5, 5 (i.e., d = 10), and uniformly set the current welding speed to v = 20 cm / min. Therefore, the initial velocity sequence of the first weld cross-section of the first pass is defined as (V... -2 V -1 V0, V2, V1) = (0, 0, 20, 20, 20); the target weld layer height is L = 5 mm.

[0112] The above specific embodiments are only for illustrating the technical concept and structural features of the present invention, and are intended to enable those skilled in the art to implement them. However, the above content does not limit the scope of protection of the present invention. Any equivalent changes or modifications made in accordance with the spirit and essence of the present invention should fall within the scope of protection of the present invention.

Claims

1. A method for controlling the interlayer formation of weld beads in arc additive manufacturing, characterized in that, The specific steps are as follows: Step S1: Collect point cloud data of the current weld layer; Step S2: Design and train the neural network prediction module, input the target weld layer process parameters into the neural network prediction module, and output the target weld layer feature data; add the target weld layer feature data with the current weld layer point cloud data to obtain the prediction layer point cloud data; The specific steps in step S2 are as follows: Step S21: Train the artificial neural network model to obtain the prediction module; Step S22: Input the target weld layer process parameters, i.e. the velocity sequence of the current weld cross section, into the weld cross section prediction unit in the neural network prediction module, and output the target weld layer feature data; The specific steps of step S22 are as follows: Step S221: Initialize parameters j=1, h=1, define j as the weld bead number, j∈{1,2,.j..,J}, J as the total number of weld beads, h as the slice number of each weld bead, h∈{1,2,.h..,H}, H as the total number of slices for each weld bead; Step S222: Input the velocity sequence of the current weld cross section into the trained artificial neural network model, and output the height axis increment of the corresponding positions of these five feature points relative to the current layer substrate; The speed sequence includes: the welding speed parameters corresponding to the h-th piece of the current j-th weld pass. Predict the welding speed parameters at the locations of the first two slices of the weld cross-section. and Predict the welding speed parameters at the locations of the two slices following the weld cross-section. and The height axis increment is defined as follows: ; Step S223: Select the current base contour feature point, add the horizontal coordinate value of the base contour feature point to the horizontal coordinate value of the current welding gun projection point to obtain the horizontal coordinate value of the predicted point; add the height coordinate value of the base contour feature point to the height axis increment to obtain the height coordinate value of the predicted point, and then obtain the coordinates of the predicted point and save the data. Step S224: Repeat steps S222-S223 until... ; Step S225: Repeat steps S222-S224 until... Thus, the predicted feature points of the target weld layer J*H slices of the target weld layer are obtained; Step S23: Correct the weld overlap; Step S24: Connect the feature points at corresponding positions of different weld cross sections according to the predicted feature points and correction results to obtain the three-dimensional model of the weld, and obtain the target weld layer feature data according to the three-dimensional model of the weld. Step S3: Based on the deviation between the target model and the predicted layer point cloud data, adjust and optimize the process parameters of the target cladding layer, and transmit the optimized process parameters to the arc additive manufacturing equipment to perform the target layer deposition.

2. The method for controlling interlayer formation of weld beads in arc additive manufacturing as described in claim 1, characterized in that, In step S223, the welding torch projection point The absolute coordinates of the projection of the intersection point of the additive path of the target weld layer and the cross-sectional slice of the weld bead of the target weld layer onto the current weld layer are defined as follows: The coordinates of the current welding torch projection point are defined as follows: Extract the projection point of the current weld cross section relative to the welding torch. The horizontal coordinate distance is , 0 , The five points are used as the current base contour feature points. , , , and The height direction coordinates of the base contour feature points are defined as [ , , , The coordinates of the five base contour feature points are as follows: ( , , , , Where d is the additive path spacing distance; That is, the first Dao Di The coordinates of the predicted feature points of the layer are , , , , .

3. The method for controlling interlayer formation of weld beads in arc additive manufacturing as described in claim 2, characterized in that, In step S23, the specific steps for correcting the weld overlap are as follows: Step S231: Initialize j=1, h=1. Step S232: For weld and The first weld The height coordinates of the overlapping portion of the slice section are corrected, as follows: ； in, For the first Welding The height coordinates of the fifth weld bead profile feature point on the weld bead cross-section. For the first weld bead The height coordinates, The overlap between the j-th and j+1-th layers of the h-th slice in the current weld layer. The coordinate value in the height direction.

4. The method for controlling interlayer formation of weld beads in arc additive manufacturing as described in claim 1, characterized in that, Step S3 includes: Step S31: Given the target cladding layer height as L, calculate the deviation between the target model and the predicted layer point cloud data. Compare with a preset threshold; if the deviation is... Less than the preset threshold If there is a deviation, proceed to step S33; Greater than or equal to the preset threshold Then the deviation of each weld cross section Perform calculations; Determine the deviation of each weld cross section. Is it greater than the adjustment threshold? ; Regarding deviation Less than or equal to the adjustment threshold The weld cross-section maintains the original welding parameters; deviations in each weld cross-section are addressed. Weld cross sections exceeding the adjustment threshold are categorized according to deviation. Welding speed in descending order Make corrections; Step S32: Based on the adjusted welding speed, repeat step S2 to obtain the predicted layer point cloud data again, input it into step S31, until the deviation between the target model and the predicted layer point cloud data meets the preset threshold requirement. Step S33: Input the adjusted target cladding layer process parameters into the arc additive manufacturing equipment to perform target cladding layer deposition.

5. The method for controlling interlayer formation of weld beads in arc additive manufacturing as described in claim 4, characterized in that, Deviation of each weld section in step S31 The calculation formula is as follows: ； in, Indicates the first Dao Di Section of weld bead The height coordinates of each feature point. This represents the coordinate value in the height direction of the target weld layer. Indicates the first Dao Di Deviation between the cross section of the weld bead and the target weld layer.

6. The method for controlling interlayer formation of weld beads in arc additive manufacturing as described in claim 5, characterized in that, welding speed The formula for correction is: ； in, Indicates the adjusted number Dao Di Welding speed of sheet weld cross section Indicates the originally scheduled number Dao Di Welding speed of sheet weld cross section This represents the correction factor.

7. A neural network-based interlayer control system for arc additive manufacturing weld bead formation, used in the interlayer control method for additive manufacturing weld bead formation as described in any one of claims 1-6, characterized in that, include: The scanning module acquires point cloud data of the current weld layer surface; The neural network prediction module outputs target weld layer feature data based on the input target weld layer process parameters. The accumulation module adds the target weld layer feature data to the current weld layer point cloud data to obtain the prediction layer point cloud data; The interlayer control module calculates the deviation between the target model and the predicted layer point cloud data. If the deviation exceeds the preset threshold, the target cladding layer process parameters are adjusted until the deviation meets the preset threshold requirements. The adjusted target cladding layer process parameters are then input into the arc additive manufacturing equipment to perform the target cladding layer deposition.

8. A robotic arc additive manufacturing apparatus, equipped with the additive manufacturing weld bead forming interlayer control system of claim 7, for implementing the additive manufacturing weld bead forming interlayer control method of any one of claims 1-6, characterized in that, It includes a welding robot, an argon arc welding torch fixed at the end of the welding robot, a 3D scanner, an industrial computer, an electric welding machine, and a robot controller. The 3D scanner collects the current point cloud data of the arc additive manufacturing structural parts, and the industrial computer receives and processes the data.

Images