Control system and method of automatic welding machine for splicing digital display cabinet body

Abstract

The invention relates to the technical field of welding automation control, and discloses a control system and method of an automatic welding machine for splicing a digital display cabinet body, and the control system of the automatic welding machine for splicing the digital display cabinet body comprises a model analysis module which is used for importing a three-dimensional structure model and extracting a splicing interface boundary point set; the path modeling module is used for constructing an initial welding path function based on a cubic B-spline curve; the path optimization module is used for constructing a multi-objective optimization functional comprising a speed square term, a curvature square term and a heat influence function term, and solving to obtain an optimized path function; and the thermal physical verification module is used for calculating temperature distribution, judging whether constraint is met or not, if not, returning to iteration, and if yes, transmitting the result to the instruction generation module. According to the method, the welding path, the welding gun posture and the welding speed are accurately controlled through combination of path modeling, thermal physical verification, posture sensing and an optimization algorithm, and the splicing welding precision, stability and automation level of the display cabinet are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of welding automation control, in particular to a control system and method of an automatic welding machine for splicing a digital display cabinet body. Background Art

[0002] In the current display cabinet manufacturing industry, custom-shaped, spliced cabinets are gradually replacing traditional, integrated welded solutions, becoming the mainstream due to their diverse structures and aesthetically pleasing aesthetics. This is particularly true in high-end showrooms, customized smart homes, and modular assembly, where cabinets require automated splicing and welding with high precision, multi-angle joints, and irregular spaces.

[0003] Some existing welding path generation systems already support extracting boundary lines from CAD models, generating point-series paths, and can perform basic fitting based on cubic interpolation functions, making the overall trajectory during the welding process smooth and controllable. Some equipment also integrates fixed-posture motion solutions at the end of the welding gun, using variable frequency control or following algorithms to achieve simple posture coordination. In addition, there are technologies that assign speeds to each path point through a PLC or industrial bus to achieve basic uniform speed control or variable speed interpolation. These technologies are highly adaptable to regular welding tasks such as straight welds and curved edge tracking, and can complete basic-level trajectory output and operation matching.

[0004] However, there are still some shortcomings in the existing technology. The above-mentioned system is not good at processing complex joint structures, especially the splicing tasks of display cabinets with multiple curvatures and variable inclinations. First, the existing methods often separate path modeling from heat input and posture changes. Although the path shape is reasonable, heat accumulation is serious during actual welding, and the process consistency cannot be controlled from the source. Secondly, the posture angle is usually set as a fixed value, and there is a lack of dynamic feedback on the real-time changing relationship between the path tangent and the workpiece normal. The posture angle changes frequently and the welding gun direction is easily offset, affecting the forming. Thirdly, although the path information can be discretized into a series of points, the control information between points is loose, and the welding speed cannot be coordinated with the local structure. Especially in sharp bends, there is no speed sensing mechanism, and welding overshoot often occurs. More importantly, there are manual sorting steps in the instruction generation process, and the control information and geometric data are managed separately, which makes the subsequent maintenance complicated, the system stability poor, and is not conducive to data integration with other digital manufacturing links. Summary of the Invention

[0005] In response to the deficiencies of the prior art, the present invention provides a control system and method for an automatic welding machine for splicing digital display cabinets, which solves the problems in the prior art of disconnection between welding path and posture control, unbalanced heat input, and separate instruction generation.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a control system for an automatic welding machine for splicing a digital display cabinet, comprising:

[0007] A model parsing module is used to import a three-dimensional structural model of the display cabinet and extract a boundary point set of the splicing interface therefrom, and transmit the boundary point set to the path modeling module;

[0008] a path modeling module, configured to receive the boundary point set, construct an initial welding path function of the splicing interface based on a cubic B-spline curve, and output the path function to a path optimization module;

[0009] a path optimization module, configured to receive the initial path function, construct a multi-objective optimization functional including a path velocity square term, a curvature square term, and a thermal influence function term, solve the functional using the Euler-Lagrange equation to obtain an optimized path function, and transmit the optimized path function to the thermal physics verification module;

[0010] Thermophysics verification module is used to calculate the temperature distribution at the path points based on the Goldak double ellipsoid heat source model and determine whether the temperature upper limit and gradient constraints are met. If not, the module feeds back to the path optimization module for re-iteration. If satisfied, the module outputs the optimized path function to the instruction generation module.

[0011] The instruction generation module is used to receive the path function verified by thermophysics, discretize it into a control point sequence and generate a welding control instruction queue containing position and posture information for execution by the welding machine.

[0012] Preferably, the model parsing module includes:

[0013] Boundary recognition submodule, used to identify the edge curve of the seam based on the topological association algorithm;

[0014] A point set sampling submodule is used to uniformly generate point set data along the boundary curve, and the output format is an ordered three-dimensional vector group;

[0015] The normal generation submodule is used to calculate the normal vector of each path endpoint using the local tangent direction.

[0016] Preferably, the path modeling module includes:

[0017] The control point setting submodule is used to receive the boundary point set and determine the fitting control points of the B-spline;

[0018] The node vector generation submodule is used to calculate the quasi-equidistant weight L between nodes based on the Euclidean distance between control point sequences. i =||P i+1 -P i || 5, After normalization, construct node parameters {u i}, and repeat the endpoint values three times at the beginning and end to generate a cubic B-spline node vector U of length n + 4 = {u0, u0, u0, u1, …, u n-1 , u n , u n , u n};

[0019] Curve construction sub-module, used to calculate the cubic B-spline curve function C(u) using the deBoor recurrence formula, where:

[0020]

[0021] where: P i is the control point; N i,3 (u) is the third-order B-spline basis function; u is the parameter variable of the curve; C(u) is the position vector of the curve at parameter u, representing the continuous points on the welding path.

[0022] Preferably, the path optimization module includes:

[0023] Functional construction sub-module, used to construct a path cost functional in the following form:

[0024]

[0025] where: C(u) represents the path curve function, and its value is the position vector in three-dimensional space; C′(u) is the first derivative of the path curve, representing the position change rate per unit parameter length, and its physical meaning is the path velocity vector; ||C′(u)|| 2 is the path velocity square term, used to reflect the path smoothness; κ(u) is the path curvature function, representing the degree of bending of the path at a certain point; Q(C(u)) is the heat influence function, representing the degree of heat load influence at the path point; α1, α2, α3 are the weighting coefficients of each cost term, respectively controlling the optimization importance of velocity smoothness, path curvature, and heat influence; J[C(u)] is the path optimization objective functional, representing the path cost function to be optimized; du is the differential element of the integration variable, corresponding to the infinitesimal length change in the parameter space, used for integral solution of the overall cost.

[0026] Extreme value solution operator module, used to solve according to the Euler-Lagrange equation:

[0027]

[0028] where: u is the curve parameter variable; L = L(C(u), C′(u), u) is the Lagrangian function, which is the target density function in path optimization; To express the full derivative with respect to the path parameter u, the rule of composite derivatives of functions is covered; Represents the partial derivative of the path position vector, reflecting the impact of the position change at the path point on the optimization target; It represents the partial derivative of the first-order derivative of the path, reflecting the influence of the path direction and speed on the target; It represents the total derivative of the term with respect to the parameter u, which is the core derivative form in the Euler-Lagrange variational solution.

[0029] The above functional is numerically iterated to obtain the optimal solution function C * (u).

[0030] Preferably, the path curvature function κ(u) is calculated by the following curvature formula:

[0031]

[0032] Where: C′(u) is the first-order derivative of the path curve, that is, the tangent vector; C″(u) is the second-order derivative of the path curve, that is, the acceleration vector; ||·|| represents the Euclidean norm of the vector; κ(u) represents the geometric curvature at the curve parameter u, which is used to measure the path change rate.

[0033] Preferably, the thermophysical verification module includes:

[0034] The temperature calculation submodule is used to calculate the heat input value of the path point based on the Goldak double ellipsoid heat source model:

[0035]

[0036] Where: q(x, y, z) is the heat flux density at the spatial point (x, y, z); f is the heat distribution proportional coefficient of the front or rear area; Q is the total heat input power; a, b, c are the semi-axis lengths of the double ellipsoid in the front-back direction, the side direction, and the vertical direction, respectively; exp[·] is the natural exponential function; x, y, z are the position coordinates of the current heat source in the reference coordinate system; π is the pi constant;

[0037] Thermal impact function calculation submodule, used to construct the path thermal impact function:

[0038]

[0039] Where: Q(C(u)) is the thermal impact evaluation function at the path point in the optimization path function; T(C(u)) is the temperature value at the path point, obtained by heat source model simulation; T0 is the reference temperature threshold used to set the thermal impact sensitive center; σ T It is a temperature change sensitive factor, which controls the influence of temperature change on the optimization target;

[0040] A constraint determination sub-module, used to determine whether the following conditions are met:

[0041]

[0042] Where: T max is the upper limit of the allowable path temperature; is the temperature gradient vector at the path point; δ T is the temperature gradient threshold; The inequality is used to constrain the heat conduction condition to prevent overheating or heat stress concentration.

[0043] Preferably, the instruction generation module includes:

[0044] A path discretization sub-module, used to perform interpolation sampling on the path function C * (u) based on Chebyshev nodes;

[0045] A welding torch attitude calculation sub-module, used to calculate the welding attitude angle θ according to the path tangent direction and the workpiece normal direction i ;

[0046] An instruction encapsulation sub-module, used to encapsulate the path point set (x i , y i , z i , θ i ) into recognizable numerical control instructions.

[0047] Preferably, the instruction generation module further includes:

[0048]

[0049] Where: T(u i ) is the path tangent vector; N(u i ) is the normal of the welding surface; θ i is the welding head attitude angle, used for the angle control of the welding torch actuator; arccos(·) is the inverse cosine function, used to calculate the included angle; ||·|| is the Euclidean norm of the vector.

[0050] The present invention also provides a control method for an automatic welding machine for splicing the cabinet body of a digital display cabinet, including the following steps:

[0051] S1. Import the three-dimensional structure model of the display cabinet, extract the boundary point set of the splicing interface, and use it as the data source for path modeling;

[0052] S2. Use the cubic B-spline interpolation method to construct the initial welding path function to ensure the continuity and differentiability of the path;

[0053] S3. Construct a path cost functional, which consists of a path velocity squared term, a path curvature squared term, and a heat influence function term, and solve the functional by the Euler-Lagrange variational principle;

[0054] S4. Use the Goldak double ellipsoid heat source model to calculate the temperature value at the path point, and judge whether the preset temperature upper limit and gradient limit conditions are satisfied. If not, feedback to step S3 to reconstruct the path;

[0055] S5. Discretize the path function verified by the thermophysical constraints into a control point sequence, calculate the path attitude angle, and generate a welding control instruction to drive the welding machine to execute.

[0056] The present invention provides a control system and method for an automatic welding machine for splicing the cabinet body of a digital display cabinet.

[0057] It has the following beneficial effects:

[0058] 1. The present invention adopts a path modeling and variational optimization fusion scheme to construct a spatial trajectory expression model based on curvature perception, effectively realizing the smooth continuity control of the welding path. The path is not only geometrically continuous, but also suppresses the speed jump of the high-speed bending section through the optimization process. Compared with the traditional CAD path discrete point-line connection method, it avoids the problem of large amplitude jitter of the welding torch attitude during weld splicing, and significantly improves the welding forming stability.

[0059] 2. The present invention introduces a thermophysical verification module nested with the Goldak heat source model, enabling the path design to first have the verification ability consistent with the heat input intensity, ensuring the controllability of the heat field. This structure estimates the heat input based on the spatial relative relationship between the path point and the welding heat source, and combines the double constraints of temperature and gradient, avoiding the overburning and insufficient penetration problems caused by "suitable path but unbalanced heat field" in the traditional system, and controlling welding process defects from the source.

[0060] 3. The present invention designs an attitude perception type instruction generation mechanism, constructs an attitude angle using the path unit tangent vector and the workpiece normal, and generates a structured instruction quadruple that can be directly applied to the industrial robot control system. This mechanism makes the path information no longer a pure spatial point position expression, but has directionality, process intention, and speed regulation ability. Compared with the existing G-code control method that cannot finely express the multi-axis attitude problem, it significantly enhances the flexible adaptability of digital welding tasks.

[0061] 4. The present invention adopts a modular data flow processing architecture, which couples five key links, namely model parsing, path modeling, optimization solving, thermal verification, and instruction encapsulation, into a closed-loop control chain according to their functions, realizing the automatic mapping from a 3D model to control code. This encapsulation method is different from the existing path input process that relies on manual step-by-step processing, greatly reducing the intermediate interaction requirements of engineering personnel, solving problems such as the dependence on human experience and the fragmentation of multiple software in the digital implementation of welding processes, and improving the engineering delivery efficiency and control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a system module diagram of the present invention;

[0063] Figure 2 It is a method flow diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the specification of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0065] Please refer to the attached Figure 1 , the embodiment of the present invention provides a control system for an automatic welding machine used for splicing the cabinet body of a digital display cabinet, including:

[0066] A model parsing module, configured to import a 3D structure model of the display cabinet, extract a set of boundary points of the splicing interface therefrom, and transfer the set of boundary points to the path modeling module;

[0067] In the control system involved in the present invention, as the first step of the data chain, the model parsing module must ensure that the spatial topological structure, geometric features, and local normal information of the weld profile are accurately extracted and meet the requirements of the path modeling module for the orderliness, continuity, and parameter differentiability of the point set.

[0068] For this reason, the model parsing module not only completes boundary extraction and normal calculation but also needs to generate a point sequence structure with certain parameter attributes to support the subsequent construction of a path function based on cubic B-spline and the calculation of an attitude control strategy.

[0069] Generally, the model parsing module receives a 3D part model of the display cabinet, usually a CAD model containing surface topological information. The model source can be output from an industrial design platform or an automatic modeling platform, and the specific software environment is not limited.

[0070] In a possible implementation, the model parsing module filters out the contour information of the area to be welded from the model through a boundary line extraction algorithm. This algorithm can be jointly set based on a two-way angle determination rule and a curvature mutation threshold. When the normal angle between surfaces is greater than the set threshold, it can be recognized as the weld boundary.

[0071] Specifically, in this embodiment, the generation of the boundary point set is carried out in three steps:

[0072] First, construct an adjacency graph structure between the model patches;

[0073] Then, mark the high-turning regions based on angle analysis;

[0074] Finally, extract and form an ordered boundary path.

[0075] In this embodiment, the boundary points are resampled at uniform intervals. The sampling interval is adjustable, with a default of 2 mm. The sampling points are marked as P i =(x i ,y i ,z i ), where i = 0, 1, …, N, in millimeters.

[0076] To implement cubic B-spline path interpolation, it is also necessary to construct path parameter variables corresponding to this point set. The present invention uses a method based on cumulative arc length to estimate the initial parameter nodes for subsequent knot vector generation. In this embodiment, the path parameters {u i} are constructed as follows:

[0077] u0 = 0, u i =u i-1 +||P i -P i-1 ||, for i = 1, …, N;

[0078] where: u0 is the initial path parameter, usually zero, representing the parameter value of the path starting point; u i is the parameter value of the path at the i-th sampling point, used to describe the progress of each point on the path; it is the cumulative length between consecutive sampling points on the path; P i-1 is the spatial coordinate of the (i - 1)-th sampling point on the path, in millimeters (mm); N is the total number of sampling points on the path, representing the number of points for path discretization; u i represents the path parameter value corresponding to the i-th path point; ||P i -P i-1 || is the Euclidean distance between adjacent points, in millimeters; P i is the three-dimensional spatial coordinate of the i-th boundary sampling point;

[0079] The obtained parameter sequence {u i} For use as reference nodes for B-spline fitting.

[0080] This path parameter estimation method is characterized by simplicity, stability, and independence from the model type, and is particularly suitable for the scenario of constructing boundary paths on complex three-dimensional surfaces.

[0081] In some embodiments, to further enhance the ability to perceive the node distribution in the path bending region, a weighting function can be introduced into the distance calculation term, for example:

[0082] u i = u i-1 + w i · ||P i - P i-1 ||;

[0083] Where: w i is a weight factor derived based on local curvature.

[0084] This method can enhance the response ability of parameter mapping to the path bending region and improve the accuracy of the fitting curve in the high-curvature section.

[0085] In addition to coordinate information, this module also needs to calculate the normal vector n i for each boundary point. This normal information is used for subsequent welding attitude angle solving and heat source direction construction. In this embodiment, the normal vector is obtained by least squares fitting of the local plane, and the specific formula is as follows:

[0086]

[0087] Where: n i is the unit normal vector at path point i, representing the normal direction of the workpiece surface at this point; a is the coefficient in the x-direction of the plane equation, representing the component of the plane normal vector on the x-axis; b is the coefficient in the y-direction of the plane equation, representing the component of the plane normal vector on the y-axis; is the Euclidean norm (magnitude) of the plane normal vector, used to normalize the normal vector into a unit vector;

[0088] The normalization process ensures that the vector magnitude is 1.

[0089] Generally, the number of neighborhood points is set to 7 to 9, depending on the distribution density of boundary points and the complexity of the surface. As an engineering implementation method, the neighborhood size can also be adaptively adjusted according to the average distance between points.

[0090] As an improved scheme, Gaussian weights can be introduced during fitting to assign different influence weights to different neighborhood points. The weight is defined as follows:

[0091]

[0092] Where: w j is the fitting weight of the jth neighborhood point; σ is the weight decay parameter, which can generally be taken as 1.5 times the average neighborhood distance; ||P j -P i || 2 is the Euclidean distance between the jth neighbor and the center point.

[0093] In summary, the data output by the model parsing module includes:

[0094] Point Set Each point contains three-dimensional coordinates;

[0095] Normal vector set One-to-one correspondence to boundary points;

[0096] Parameter node sequence Used for path function construction.

[0097] These outputs are directly fed into the path modeling module as input to the control point and node sequence. The accuracy and stability of this module significantly impact the subsequent path fitting and posture control, so its robustness and adaptability have been enhanced as much as possible in the design.

[0098] A path modeling module is used to receive a boundary point set, construct an initial welding path function of the splicing interface based on a cubic B-spline curve, and output the path function to a path optimization module;

[0099] In the system of the present invention, the path modeling module takes over the welding boundary point set and normal vector set output by the model analysis module, and mainly completes the task of constructing a three-dimensional path. The path is a path defined in the parameter interval [u0,u n ] is used to describe the desired position sequence of the welding gun motion trajectory. The fitting curve must have continuous derivatives and finite curvature to meet the requirements of smoothness and posture control during the welding process.

[0100] Generally, to ensure high-order geometric continuity in the modeling results, the system selects cubic B-spline curves as the construction form. This type of curve supports segmented definition, has continuous curvature, and has good local controllability. It is suitable for fitting industrial boundary paths with a certain degree of free curvature variation.

[0101] Specifically, in this embodiment, the path function is defined as:

[0102]

[0103] Where: C(u) is the path function, which represents the desired position of the welding gun tip at parameter u in three-dimensional Euclidean space; n is the total number of control points minus 1; N i,3(u) is the i-th cubic B-spline basis function, whose value is related to the path parameter u and the knot vector U; u is the path parameter, a dimensionless variable with monotonic increasing property; [u0, u n are the start and end points of the path parameter interval, corresponding to the first and last endpoints of the boundary point set.

[0104] In a possible implementation, the control point set P i is constructed following the boundary sampling order, maintaining the topological consistency of the point set arrangement without reordering. The system ensures geometric continuity between the start and end of the path through external boundary recognition logic, avoiding self-intersection and redundancy.

[0105] The generation of the knot vector is based on a fifth-order quasi-uniform strategy:

[0106] L i = ||P i+1 - P i || 5 ;

[0107]

[0108] U = {u0, u0, u0, u1, …, u n-1 , u n , u n , u n};

[0109] Where: L i is the quasi-uniform weight between control points, and using the fifth-order power can enhance the resolution ability for curvature mutation regions; ||P i+1 - P i || 5 is the Euclidean distance between adjacent control points, with the unit of millimeter (mm); is the total weight of the knots; u i is the normalized knot parameter value, representing the relative position of the control point P i in the overall path; U is the B-spline knot vector, with a length of n + 4, and its start and end each contain 3 repeated endpoints to ensure the integrity of the construction of the cubic spline boundary segment.

[0110] Inside the path modeling module, the value of each basis function N i,3 (u) is calculated by a recursive formula:

[0111]

[0112] Where: N i,0 (u) represents the basis function in the i-th interval. When the parameter u falls within the interval [u i , u i+1 , the value is 1, otherwise it is 0; u iThe \(i\)-th node of the path parameter, representing a certain parameter position on the discretized path; \(u\) i+1 The \((i + 1)\)-th node of the path parameter, representing the next parameter position on the discretized path; \(N\) i,k \(N_{i,k}(u)\) represents the \(k\)-th order basis function on the \(i\)-th interval, used for higher-order interpolation; \(k\) represents the order of the basis function. When the order is 0, it represents a simple piecewise constant basis function (i.e., \(N_{i,0}(u)\)), while higher-order basis functions are used for more complex interpolation processes; \(N\) i,0 (u)) and higher-order basis functions are used for more complex interpolation processes; \(N\) i,k-1 \(N_{i,k - 1}(u)\) represents the \((k - 1)\)-th order basis function on the \(i\)-th interval, which is part of the recursive formula and is used to calculate the \(k\)-th order basis function; \(u\) i+k represents the \((i + k)\)-th node of the path parameter, participating in the interpolation in the recursive calculation; \(u\) i+k+1 represents the \((i + k + 1)\)-th node of the path parameter, participating in the interpolation in the recursive calculation.

[0113] In this embodiment, the constructed path function \(C(u)\) is not directly used for generating path discrete points, but is used as the input of the subsequent path optimization module and attitude calculation module.

[0114] In some embodiments, to verify the smoothness of the curve construction result, first-order derivative continuity and curvature change analysis can be used for auxiliary determination. The first-order derivative of the curve is as follows:

[0115]

[0116] where: \(C'(u)\) is the first-order derivative of the path function, representing the tangent vector direction at the parameter \(u\); \(N'_{i,k}(u)\) i,3 is the derivative of the cubic B-spline basis function with respect to \(u\), representing the slope response of the basis function.

[0117] The finally generated path function exists in the form of an analytical expression, which can be called by the path optimization module when establishing the energy functional, and can also be directly used for trajectory simulation.

[0118] The accuracy of this module depends on the number and distribution of control points. Therefore, the present invention supports automatically refining the control point density in a specific area by interpolation. However, the adjustment of control points and the interpolation strategy are both controlled by external logic and do not belong to the inherent responsibilities of the path modeling module.

[0119] The path optimization module is used to receive the initial path function, construct a multi-objective optimization functional including the path velocity squared term, curvature squared term, and thermal influence function term, and solve this functional through the Euler-Lagrange equation to obtain the optimized path function, and transmit it to the thermophysical verification module;

[0120] Although the path modeling module provides a spatially continuous path constructed by cubic B-spline, in the actual welding process, if the path changes too violently or the heat input is uneven, process problems such as warping, burn-through, or insufficient soldering of the workpiece will occur. Therefore, the path optimization module reprocesses the original path function C(u), and realizes the joint constraint of multiple physical quantities by constructing a multi-objective functional, ensuring that the path has structural safety, thermal stability, and process consistency.

[0121] Generally, the path optimization module uses an integral functional form with the path function as the independent variable for modeling. This functional consists of three parts: a velocity term, a curvature term, and a heat load term. By setting different weight coefficients, the adjustability and flexibility of the optimization objectives are realized.

[0122] In this embodiment, the objective functional constructed by the path optimization module is as follows:

[0123]

[0124] Where: C(u) represents the path curve function, and its value is the position vector in three-dimensional space; C′(u) is the first derivative of the path curve, representing the position change rate per unit parameter length, and its physical meaning is the path velocity vector; ||C′(u)|| 2 is the path velocity squared term, which is used to reflect the path smoothness; κ(u) is the path curvature function, which represents the degree of bending of the path at a certain point; Q(C(u)) is the heat influence function, which represents the degree of heat load influence at the path point; α1, α2, α3 are the weighting coefficients of each cost term, which respectively control the optimization importance of velocity smoothness, path curvature, and heat influence; J[C(u)] is the path optimization objective functional, which represents the path cost function to be optimized; du is the differential element of the integration variable, corresponding to the tiny length change in the parameter space, and is used for integral calculation of the overall cost.

[0125] Q(C(u)) is the heat penalty term at the path point, which reflects the feedback of the welding heat input on the path shape, and is defined as follows:

[0126]

[0127] Where: Q(C(u)) is the heat influence evaluation function at the path point in the optimized path function; T(C(u)) is the temperature value at the path point, which is obtained by simulation of the heat source model; T0 is the reference temperature threshold, which is used to set the heat influence sensitive center; σ T is the temperature change sensitivity factor, which controls the influence degree of temperature change on the optimization objective.

[0128] As a possible implementation method, the system uses the Euler-Lagrange variational method to solve the above functional. The basic form is as follows:

[0129]

[0130] where: u is the curve parameter variable; L = L(C(u), C′(u), u) is the Lagrangian function, which is the objective density function in path optimization; denotes the total derivative with respect to the path parameter u, covering the function composite derivative rule; denotes the partial derivative with respect to the path position vector, reflecting the influence of the position change at the path point on the optimization objective; denotes the partial derivative with respect to the first-order derivative of the path, reflecting the influence of the path direction and speed on the objective; denotes the total derivative of this term with respect to the parameter u, which is the core derivative form in the Euler - Lagrange variational solution;

[0131] This equation is the Euler - Lagrange equation set for three-dimensional variational problems, used to obtain the minimum path of the functional.

[0132] In this embodiment, the equation set is numerically discretized using the finite difference method. The parameter interval [u0, u n is equally divided into M segments. First-order and second-order difference approximations are performed on the path function at each node, and then about 3M linear equations on the entire path are constructed and iterated by the sparse matrix solver.

[0133] In some embodiments, to limit the excessive deviation between the optimized path and the original modeled path, a path deviation constraint term can be introduced:

[0134] Δ(u) = λ d ·||C(u) - C0(u)|| 2 ;

[0135] where: Δ(u) is the conformal constraint term, with the unit of mm 2 ; λ d is the deviation penalty coefficient, generally a positive real number; C0(u) is the initial cubic B-spline path function;

[0136] This constraint is added in the form of an additional term when solving the minimum value of the functional, which helps the path to "converge without deviation".

[0137] Finally, the system will output the optimized path function C * (u), which meets the engineering requirements in terms of physical path length, smoothness, bending distribution, and thermal uniformity, and is qualified to be directly sent to the subsequent thermophysical simulation and attitude angle generation modules.

[0138] The thermophysical verification module is used to calculate the temperature distribution at the path points based on the Goldak double-ellipsoid heat source model, and determine whether the temperature upper limit and gradient constraint conditions are met. If not, it feeds back to the path optimization module for re-iteration. If satisfied, it outputs the optimized path function to the instruction generation module;

[0139] After the path optimization module outputs the optimal path function C * (u), the system cannot immediately use it for generating welding control instructions. The reason is that although this path already meets the speed and curvature requirements, it has not performed physical verification on the controllability of the heat load in the welding process. If the temperature field analysis on the path is ignored, it may cause quality defects such as thermal damage to the workpiece surface, structural deformation, and discontinuous weld seams. Therefore, the system introduces a thermophysical verification module as an intermediate control link to simulate and evaluate the heat influence behavior of the path points.

[0140] Generally, the welding heat input has the characteristics of local concentration and high-gradient change, resulting in uneven heat distribution between path points. This module calculates the steady-state temperature value and temperature gradient of each path point to determine whether it meets the preset thermal process safety interval. If there is an overlimit, it automatically feeds back to the path optimization module to adjust the weight of the thermal penalty term.

[0141] This module discretizes the path into N sampling points C i = C(u i ), where i = 1, 2,..., N, and the step size Δu can be set as equally spaced or adaptively spaced, usually taking a value of 0.01 or smaller.

[0142] To calculate the heat input of the path points, the system constructs a welding thermal field based on the Goldak double-ellipsoid heat source model. The heat input expression is as follows:

[0143]

[0144] where: q(x, y, z) is the heat flux density at the spatial point (x, y, z); f is the heat distribution proportionality coefficient of the front or rear area; Q is the total heat input power; a, b, c respectively represent the semi-axis lengths of the double-ellipsoid in the front-back direction, lateral direction, and vertical direction; exp[·] is the natural exponential function; x, y, z are the position coordinates in the current heat source reference coordinate system; π is the constant of the circumference ratio.

[0145] This local coordinate system is obtained through the three-axis transformation of the path point tangent vector and the welding torch attitude direction. In this module, it is default that the three-axis transformation matrix is provided by the path modeling module.

[0146] In some embodiments, the heat input intensity q(x i , y i , z i)The temperature T(C) of the path point can be further derived through numerical integration or Gaussian distribution accumulation. i )To simplify the calculation, this module uses a steady-state temperature rise model to estimate the temperature of the path point, and the expression is as follows:

[0147] T(C i ) = T0 + η·q(x i , y i , z i );

[0148] Where: T(C i ) is the estimated value of the steady-state temperature of path point C i , with the unit of Kelvin (K); T0 is the initial ambient temperature, generally taken as 300K; η is the thermal efficiency coefficient, with the unit of mm 3 ·K / W, reflecting the proportional relationship between heat input and temperature rise; q(x i , y i , z i ) is the heat input value of the i-th point, substituted by the above formula;

[0149] The temperature rise result is a linear estimate, suitable for path screening and does not replace the analysis of the thermal field simulation software.

[0150] The path temperature check includes two constraint judgments:

[0151] First, the absolute temperature constraint:

[0152] T(C i ) ≤ T max ;

[0153] Where: T max is the maximum allowable working temperature of the material, with the unit of K, usually 90% of the melting point;

[0154] For example, austenitic stainless steel can be set to 1700K.

[0155] Second, the temperature gradient constraint:

[0156]

[0157] Where: is the temperature gradient of the path point, with the unit of K / mm; Δs is the spatial distance between path points; δ T is the upper threshold of the temperature gradient, with the unit of K / mm, set according to the stress sensitivity of the material, generally 250K / mm.

[0158] In this embodiment, if the temperature or gradient of any path point does not meet the above constraint conditions, the system will enter the thermal check failure branch and perform the following operations:

[0159] If the proportion of unqualified path points is less than a certain threshold (e.g., 5%), the control points can be adjusted by local interpolation to refine the local path;

[0160] If the thermal mismatch is concentrated and the area persists, the system will re - call the path optimization module, increase the coefficient α3 of the thermal penalty term, and solve the variational path again;

[0161] After each adjustment, re - enter the thermophysical verification link of this module until all points meet the constraints.

[0162] Finally, this module outputs a path function C * (u) that meets the thermophysical requirements and passes it to the next - level attitude calculation and control instruction generation module for path space transcoding.

[0163] The instruction generation module is used to receive the path function verified by thermophysics, discretize it into a control point sequence, and generate a welding control instruction queue containing position and attitude information for the welding machine to execute;

[0164] After the control system in the present invention completes path optimization and thermophysical verification, it passes the optimal path function C * (u) to the instruction generation module. The task of this module is to convert the geometric information and attitude parameters of the path points into the underlying control instructions for driving the automatic welding equipment, realizing the control implementation from high - order modeling to the equipment layer.

[0165] Generally, an industrial welding robot needs to receive a set of discrete control points that clearly express position and attitude, and cooperate with accessory information such as welding speed, trajectory segment type, and process switch. Therefore, this module needs to complete multiple tasks such as three - dimensional position calculation, attitude angle solution, speed distribution, and instruction data encapsulation.

[0166] As an option, in this embodiment, the path function is discretized in the parameter space.

[0167] The system equally divides the interval [u0, u n into N nodes, generating a parameter sequence For each path node u i ;

[0168] Subsequently, the system calculates the first - order derivative of the path function to obtain the path direction:

[0169]

[0170] where: T(u i ) is the unit tangent vector of the path point u i , and its direction is consistent with the movement direction of the welding torch; C′(u i ) is the derivative of the path at u iThe velocity vector at , i.e., the first derivative, is in mm; ||C′(u i )|| is the magnitude (Euclidean norm) of this velocity vector, in mm.

[0171] The definition of the attitude angle is used to describe the relative attitude relationship between the welding torch and the workpiece, and the specific calculation is as follows:

[0172]

[0173] where: T(u i ) is the path tangent vector; N(u i ) is the normal of the welding surface; θ i is the attitude angle of the welding head, used for the angle control of the welding torch actuator; arccos(·) is the inverse cosine function, used to calculate the included angle; ||·|| is the Euclidean norm of the vector.

[0174] In this embodiment, to achieve the adaptive adjustment of the path segment speed, the system introduces a speed adjustment strategy based on path curvature. The curvature value κ(u i ) is derived from the curvature calculation expression in the path optimization module. On this basis, the welding speed is set as:

[0175] v i = v max ·(1 - β·κ(u i )) 2 ;

[0176] where: v i is the welding speed at path point u i , in millimeters per second (mm / s); v max is the maximum welding speed allowed by the system, set by the console; β is the speed weakening coefficient, in mm 2 , used to control the sensitivity of the speed change with curvature; κ(u i ) is the curvature value at path point u i , in mm -1 ; when the path is in a sharp bend area, the speed will automatically decrease to ensure welding stability 1.

[0177] Finally, the system constructs an instruction quintuple at each path point:

[0178] I i = (x i , y i , z i , θ i , v i );

[0179] where: x i , y i , zi is the spatial position of the path point; θ i is the attitude angle, used to guide the direction of the welding torch; v i is the motion speed that the control system should maintain.

[0180] The above five-tuple will be transcoded into a specific recognizable data structure according to the target control platform. Different control platform interfaces are different, and the typical encapsulation forms are as follows:

[0181] If the target controller is an industrial robot control system (such as ABB, KUKA), the instruction can be transcoded into a structured motion block (Motion Instruction Block);

[0182] If the target device uses an open PLC platform, the data can be written into the CSV or MODBUS register format;

[0183] If used in a ROS control system, it can be packaged into a standard PoseStamped message body, and meta-information such as timestamp and frame ID can also be added.

[0184] As an optional strategy, to improve the instruction accuracy of the welding path transition section, the system can locally increase the path sampling density at the curvature mutation point, that is, at κ(u i ) > κ thresh intermediate nodes are locally added. This processing can improve the controller trajectory resolution accuracy and reduce attitude disturbance.

[0185] The control method of the automatic welding machine for digital display cabinet cabinet body splicing described below can be mutually corresponding and referred to the control system of the automatic welding machine for digital display cabinet cabinet body splicing described above.

[0186] Please refer to the appendix Figure 1 , the present invention also provides a control method for an automatic welding machine for digital display cabinet cabinet body splicing, including the following steps:

[0187] S1. Import the three-dimensional structure model of the display cabinet, extract the splicing interface boundary point set as the data source for path modeling;

[0188] S2. Use the cubic B-spline interpolation method to construct the initial welding path function to ensure the continuity and differentiability of the path;

[0189] S3. Construct a path cost functional, which is composed of the path speed square term, the path curvature square term and the heat influence function term, and solve the functional through the Euler-Lagrange variational principle;

[0190] S4. Use the Goldak double-ellipsoid heat source model to calculate the temperature value at the path point, judge whether it meets the preset temperature upper limit and gradient limit conditions, and if not, feedback to step S3 to reconstruct the path;

[0191] S5. Discretize the path function verified by the thermophysical constraint into a sequence of control points, calculate the path attitude angle, and generate a welding control instruction for driving the welding machine to execute.

[0192] The method of this embodiment can be used to implement the above system embodiment, and its principle and technical effect are similar, so they will not be elaborated here.

[0193] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. Control system of an automatic welding machine for splicing the cabinet body of a digital display cabinet, characterized in that, Including: A model parsing module, which is used to import the three-dimensional structure model of the display cabinet, extract the boundary point set of the splicing interface from it, and transfer the boundary point set to the path modeling module; A path modeling module, which is used to receive the boundary point set, construct an initial welding path function of the splicing interface based on the cubic B-spline curve, and output the path function to the path optimization module; A path optimization module, which is used to receive the initial path function, construct a multi-objective optimization functional including the path speed square term, the curvature square term and the heat influence function term, solve the functional through the Euler-Lagrange equation to obtain the optimized path function, and transmit it to the thermophysical verification module; A thermophysical verification module, which is used to calculate the temperature distribution at the path points based on the Goldak double-ellipsoid heat source model, and judge whether the temperature upper limit and gradient constraint conditions are satisfied. If not, it feeds back to the path optimization module for re-iteration. If satisfied, it outputs the optimized path function to the instruction generation module; An instruction generation module, which is used to receive the path function verified by thermophysics, discretize it into a control point sequence, and generate a welding control instruction queue containing position and attitude information for the welding machine to execute.

2. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 1, characterized in that, The model parsing module includes: A boundary recognition sub-module, which is used to recognize the seam edge curve based on the topological association algorithm; A point collection and sampling sub-module, which is used to uniformly generate point set data along the boundary curve, and the output format is an ordered three-dimensional vector group; A normal generation sub-module, which is used to calculate the normal vector of each path end point through the local tangent direction.

3. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 1, wherein, The path modeling module includes: A control point setting sub-module, which is used to receive the boundary point set and determine the fitting control points of the B-spline; A node vector generation sub-module, which is used to calculate the quasi-equidistant weight L between nodes based on the Euclidean distance between control point sequences i = ||P i+1 - P i || 5 , and after normalization, construct node parameters {u i}, and repeat the endpoint values three times at both the beginning and the end to generate a cubic B-spline node vector U = {u0, u0, u0, u1, …, u n-1 , u n , u n , u n}; A curve construction sub-module, which is used to calculate the cubic B-spline curve function C(u) using the deBoor recurrence formula, where: Where: P i is a control point; N i,3 (u) is a third-order B-spline basis function; u is a parameter variable of the curve; C(u) is the position vector of the curve at parameter u, representing the continuous points on the welding path.

4. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 1, characterized in that, The path optimization module includes: A functional construction sub-module, which is used to construct a path cost functional in the following form: where: C(u) represents the path curve function, and its value is the position vector in three-dimensional space; C′(u) is the first derivative of the path curve, representing the position change rate per unit parameter length, and its physical meaning is the path velocity vector; ||C′(u)|| 2 is the path velocity squared term, used to reflect the path smoothness; κ(u) is the path curvature function, indicating the degree of bending of the path at a certain point; Q(C(u)) is the heat influence function, representing the degree of heat load influence at the path point; α1, α2, α3 are the weighting coefficients of each cost term, respectively controlling the optimization importance of velocity smoothness, path curvature, and heat influence; J[C(u)] is the path optimization objective functional, representing the path cost function to be optimized; du is the differential element of the integration variable, corresponding to the tiny length change in the parameter space, used for integral solution of the overall cost; An extreme value solution operator module, which is used to according to the Euler-Lagrange equation: where: u is the curve parameter variable; L = L(C(u), C′(u), u) is the Lagrangian function, which is the objective density function in path optimization; denotes the total derivative with respect to the path parameter u, covering the rules of composite function derivatives; denotes the partial derivative with respect to the path position vector, reflecting the impact of position changes at path points on the optimization objective; denotes the partial derivative with respect to the first-order derivative of the path, reflecting the impact of path direction and speed on the objective; denotes the total derivative of this term with respect to the parameter u, which is the core derivative form in the Euler-Lagrange variational solution; Numerically iterate and solve the above functional to obtain the path optimal solution function C * (u).

5. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 4, characterized in that, The path curvature function κ(u) is calculated by the following curvature formula: Where: C′(u) is the first derivative of the path curve, that is, the tangent vector; C″(u) is the second derivative of the path curve, that is, the acceleration vector; ||·|| represents the Euclidean norm of the vector; κ(u) represents the geometric curvature at the curve parameter u, which is used to measure the path change rate.

6. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 1, characterized in that, The thermophysical verification module includes: A temperature calculation sub-module, which is used to calculate the heat input value at the path point based on the Goldak double-ellipsoid heat source model: Where: q(x,y,z) is the heat flux density at the spatial point (x,y,z); f is the front or rear area heat distribution proportionality coefficient; Q is the total heat input power; a, b, c respectively represent the semi-axis lengths of the double-ellipsoid in the front-back direction, lateral direction and vertical direction; exp[·] is the natural exponential function; x, y, z are the position coordinates in the current heat source reference coordinate system; π is the constant of the circumference ratio; A heat influence function calculation sub-module, which is used to construct a path heat influence function; Where: Q(C(u)) is the heat impact evaluation function at the path point in the optimized path function; T(C(u)) is the temperature value at the path point, obtained by simulating the heat source model; T0 is the reference temperature threshold for setting the heat impact sensitive center; σ T is the temperature change sensitivity factor, which controls the influence degree of temperature change on the optimization objective; A constraint determination sub-module, which is used to judge whether it satisfies: Where: T max is the upper limit of the allowable path temperature; is the temperature gradient vector at the path point; δ T is the temperature gradient threshold; The inequality is used to constrain the heat conduction condition to prevent overheating or thermal stress concentration.

7. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 1, characterized in that, The instruction generation module includes: A path discretization sub-module, which is used to perform interpolation sampling on the path function C * (u) based on Chebyshev nodes; The welding torch attitude calculation sub-module is used to calculate the welding attitude angle θ according to the path tangent direction and the workpiece normal direction i ; The instruction encapsulation sub-module is used to encapsulate the path point set (x i , y i , z i , θ i ) into recognizable numerical control instructions.

8. The control system of the automatic welding machine for splicing the digital display cabinet body according to claim 1, characterized in that, The instruction generation module further includes: Where: T(u i ) is the path tangent vector; N(u i ) is the welding surface normal; θ i is the welding head attitude angle, which is used for the angle control of the welding torch actuator; arccos(·) is the inverse cosine function, which is used to calculate the included angle; ||·|| is the Euclidean norm of the vector.

9. Control method of an automatic welding machine for splicing the cabinet body of a digital display cabinet, characterized in that, The control system of the automatic welding machine for splicing the digital display cabinet body described in any one of claims 1-8 includes the following steps: S1. Import the three-dimensional structure model of the display cabinet, extract the boundary point set of the splicing interface, and use it as the data source for path modeling; S2. Use the cubic B-spline interpolation method to construct the initial welding path function to ensure the continuity and differentiability of the path; S3. Construct the path cost functional, which is composed of the path speed square term, the path curvature square term and the heat-affected function term, and solve the functional through the Euler-Lagrange variational principle; S4. Use the Goldak double-ellipsoid heat source model to calculate the temperature value at the path point, judge whether the preset temperature upper limit and gradient limit conditions are satisfied, and if not, feedback to step S3 to reconstruct the path; S5. Discretize the path function verified by the thermophysical constraints into a control point sequence, calculate the path attitude angle, and generate a welding control instruction to drive the welding machine to execute.

Images