Automatic welding method, system and storage medium for steel structures

Abstract

The embodiment of the application provides an automatic welding method and system of a steel structure and a storage medium, the method comprising: obtaining a preset three-dimensional model of a target steel structure; generating a current three-dimensional model after scanning a current steel structure; determining a welding atlas according to the preset three-dimensional model, the current three-dimensional model and a structure piece to be welded; obtaining a target welding scheme after iteratively optimizing a welding scheme corresponding to the welding atlas according to preset decision variables and optimization objectives; and obtaining the target steel structure after welding the structure piece to be welded according to the target welding scheme. The welding atlas is determined by comparing the current three-dimensional model with the preset three-dimensional model, and the target welding scheme is obtained by optimizing the decision variables and the optimization objectives, so that the steel structure welding robot can accurately weld the structure piece to be welded, the welding quality is good, and the welding efficiency is high.

Description

Technical Field

[0001] This invention relates to, but is not limited to, the field of data processing technology, and particularly to an automatic welding method, system, and storage medium for steel structures. Background Technology

[0002] Steel structures are structures mainly composed of steel materials and are one of the main types of building structures. The structure is mainly composed of steel beams, steel columns, steel frames and other components made of steel sections and steel plates. The components or parts are usually connected by welds, bolts or rivets. Due to its light weight and simple construction, it is widely used in large factories, stadiums, bridges, super high-rise buildings and other fields.

[0003] Traditional steel structures are welded manually, which easily leads to unstable welding quality and low welding efficiency. Summary of the Invention

[0004] The following is an overview of the subject matter described in detail herein. This overview is not intended to limit the scope of the claims.

[0005] The main objective of this invention is to provide an automatic welding method, system, and storage medium for steel structures, which can improve the welding quality and efficiency of steel structures.

[0006] In a first aspect, embodiments of the present invention provide an automatic welding method for steel structures, applied to a steel structure welding robot, the method comprising:

[0007] Obtain a preset three-dimensional model of the target steel structure, which is used to characterize the complete structure of the steel structure;

[0008] The current 3D model is generated after scanning the current steel structure;

[0009] Welding patterns are determined based on the preset 3D model, the current 3D model, and the structural component to be welded. The welding patterns are used to characterize the welding areas and welding thicknesses of the structural component to be welded. The structural component to be welded represents the structural component to be welded in the current 3D model.

[0010] The target welding scheme is obtained by iteratively optimizing the welding scheme corresponding to the welding pattern according to the preset decision variables and optimization objectives. The optimization objectives include minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency. The decision variables include welding torch path variables, welding torch process variables, and welding sequence variables.

[0011] The target steel structure is obtained by welding the structural components to be welded according to the target welding scheme.

[0012] In some optional embodiments, determining the welding pattern based on the preset three-dimensional model, the current three-dimensional model, and the structural component to be welded includes:

[0013] After mapping the current three-dimensional model to the preset three-dimensional model, the target structural component model of the structural component to be welded in the preset three-dimensional model is determined;

[0014] Based on the positional relationship between the target structural component model and other structural component models in the preset three-dimensional model, the material information of each structural component, the shape information of each structural component, and the load information, the stress spectrum corresponding to the target structural component model is determined, and the load information represents the maximum load capacity of the target steel structure.

[0015] Obtain the weld information of the structural component to be welded, wherein the weld information indicates the weld type, weld size and weld location;

[0016] The welding pattern is determined based on the stress pattern and the weld information.

[0017] In some optional embodiments, determining the force spectrum corresponding to the target structural component model based on the positional relationship between the target structural component model and other structural component models in the preset three-dimensional model, the material information of each structural component, the shape information of each structural component, and the load information of each structural component includes:

[0018] Based on the positional relationship, the connection constraints between the target structural component model and the adjacent structural component model are determined, as well as the target position of the target structural component model in the target steel structure is determined. The connection constraints include rigid connection constraints, hinged connection constraints, and simply supported constraints. The connection constraints are used to indicate the degree of freedom of movement of the target structural component model.

[0019] The static load distribution map corresponding to the target structural component model is determined based on the target location, the connection constraints, the material information of each structural component, and the shape information of each structural component.

[0020] The dynamic load distribution pattern corresponding to the target structural component model is determined based on the load information, the target location, and the connection constraints.

[0021] The stress diagram is obtained by superimposing the static load distribution diagram and the dynamic load distribution diagram.

[0022] In some optional embodiments, determining the static load distribution map corresponding to the target structural component model based on the target location, the connection constraints, the material information of each structural component, and the shape information of each structural component includes:

[0023] Based on the target location, determine all associated structural component models in the preset three-dimensional model that contribute to the gravitational load of the target structural component model;

[0024] The gravity load distribution map of the target structural component model is determined based on the material information of each structural component, the shape information of each structural component, and the associated structural component model.

[0025] The static load distribution map is obtained by correcting the gravity load distribution map according to the connection constraints.

[0026] In some optional embodiments, determining the dynamic load distribution pattern corresponding to the target structural component model based on the load information, the target location, and the connection constraints includes:

[0027] The initial load distribution pattern of the target structural component model is determined based on the load information and the target location;

[0028] The dynamic load distribution pattern that changes over time is determined based on the connection constraints and the initial load distribution pattern.

[0029] In some optional embodiments, determining the welding pattern based on the stress pattern and the weld information includes:

[0030] Determine the stress coefficients of each region on the target structural component model based on the stress spectrum;

[0031] The stress coefficient corresponding to the weld on the target structural component model is determined based on the stress coefficient of each region and the location of the weld.

[0032] The welding thickness and welding process of the weld are determined based on the stress coefficient and the weld type;

[0033] The welding area of ​​the weld is determined based on the stress coefficient and the weld size.

[0034] In some optional embodiments, the step of iteratively optimizing the welding scheme corresponding to the welding pattern based on preset decision variables and optimization objectives to obtain the target welding scheme includes:

[0035] Construct a welding neural network model;

[0036] The welding pattern, the current 3D model, and the structural component to be welded are input into the welding neural network model;

[0037] The optimization objective of the welding neural network model is configured as minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency.

[0038] Construct the objective function corresponding to the optimization objective;

[0039] The design variables of the welding neural network model are configured as the decision variables. The welding torch path variables include the welding torch trajectory, welding torch travel speed and welding torch pose. The welding torch process variables include welding current, welding voltage, wire feed speed and interpass temperature. The welding sequence variables include the welding sequence of each weld, the welding layers of multiple weld layers and the segment length of segmented skip welding.

[0040] The constraints of the welding neural network model are configured as structural blocking constraints, welding torch operating parameter constraints, and thermal deformation constraints.

[0041] The optimal solution set is obtained by iterating the design variables based on the constraints, the objective function, and the optimization objective.

[0042] The target welding scheme is obtained by selecting the target solution from the set of optimal solutions based on the overall benefit requirements.

[0043] In some optional embodiments, obtaining the optimal solution set by iterating the design variables based on the constraints, the objective function, and the optimization objective includes:

[0044] A first number of first decision variable combinations are randomly generated, and the first decision variable combinations satisfy the constraints.

[0045] Substitute the first combination of decision variables into the objective function to calculate the objective function value;

[0046] The second number of combinations of the first decision variables that minimize the objective function value is retained, where the first number is greater than the second number.

[0047] The first combination of decision variables is subjected to parameter cross-validation and random parameter mutation to generate the second combination of decision variables, and the iteration number is incremented by one.

[0048] If the number of iterations is less than or equal to a preset number and the objective function value has not converged, the second decision variable combination is configured as the first decision variable combination and then iterative calculation is performed. The iterative calculation represents the above-mentioned step of generating the second decision variable combination.

[0049] If the number of iterations is greater than a preset number, or if the objective function value converges, the second combination of decision variables is configured as the optimal solution set.

[0050] In a second aspect, embodiments of the present invention provide an automatic welding system for steel structures, including a controller. The controller includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the automatic welding method for steel structures described in the first aspect.

[0051] Thirdly, embodiments of the present invention provide a computer storage medium storing computer-executable instructions, which are used to execute the automatic welding method for steel structures described in the first aspect.

[0052] The beneficial effects of this invention include: acquiring a preset three-dimensional model of the target steel structure, the preset three-dimensional model being used to characterize the complete structure of the steel structure; generating a current three-dimensional model after scanning the current steel structure; determining a welding pattern based on the preset three-dimensional model, the current three-dimensional model, and the structural component to be welded, the welding pattern being used to characterize the welding area and welding thickness of each area of ​​the structural component to be welded, the structural component to be welded representing the structural component to be welded in the current three-dimensional model; obtaining a target welding scheme by iteratively optimizing the welding scheme corresponding to the welding pattern based on preset decision variables and optimization objectives, the optimization objectives including minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency, the decision variables including welding torch path variables, welding torch process variables, and welding sequence variables; and obtaining the target steel structure by welding the structural component to be welded according to the target welding scheme. During welding, the steel structure welding robot compares the scanned current 3D model with the preset 3D model to determine the welding pattern of the structural component to be welded. It then optimizes the target welding scheme for the welding area on the welding pattern by using decision variables and optimization objectives. This enables the steel structure welding robot to accurately weld the structural component to be welded, resulting in good welding quality and high welding efficiency.

[0053] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the steps of an automatic welding method for steel structures provided in an embodiment of the present invention.

[0055] Figure 2 This is a schematic diagram of a controller provided in one embodiment of the present invention.

[0056] Reference numerals: Controller 1000, Processor 1100, Memory 1200. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0058] It should be noted that although functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, or the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0059] This application provides an automatic welding method, system, and storage medium for steel structures, which will be described in detail in the following embodiments.

[0060] like Figure 1 As shown, this embodiment of the invention provides an automatic welding method for steel structures, including steps S100, S200, S300, S400, and S500:

[0061] Step S100: Obtain a preset three-dimensional model of the target steel structure, wherein the preset three-dimensional model is used to characterize the complete structure of the steel structure.

[0062] Specifically, the preset 3D model in this application can be model data that has been acquired in advance and stored on a backend server or in a corresponding storage space, or it can be obtained by accessing a corresponding data storage system or platform. The specific acquisition method is not limited here.

[0063] A preset 3D model is a pre-set 3D model that can display the specific structure of a steel structure after construction (such as a 3D model of a bridge). The preset 3D model includes: the complete geometric features of the steel structure, such as member dimensions, node types, bevel shapes (V-shaped, X-shaped, U-shaped), blunt edge height, assembly gaps, etc.; and the design attributes of the welds, such as weld location and welding type.

[0064] Step S200: After scanning the current steel structure, generate the current 3D model.

[0065] Specifically, the steel structure is scanned using scanning equipment such as a scanner or scanning arm to generate a corresponding three-dimensional model. The current steel structure is the one under construction or being welded.

[0066] The original discrete point cloud data obtained from the scan is denoised (using statistical filtering to remove irrelevant noise such as scaffolding and fixtures), registered (using the ICP iterative nearest point algorithm to stitch together point clouds from multiple regions into a complete model), and segmented and simplified (separating the point clouds of the structural components to be welded from those of the welded / attached components, and downsampling non-critical areas to reduce the amount of data). The processed point cloud data is then transformed into a visualization model: specifically, a triangular mesh model is generated based on the point cloud data, and algorithms such as Poisson reconstruction are used to fill mesh gaps and smooth surfaces to intuitively reflect the true shape of the components; the triangular mesh model is then transformed into a parametric solid model to obtain the current 3D model.

[0067] Step S300: Determine the welding pattern based on the preset three-dimensional model, the current three-dimensional model, and the structural component to be welded. The welding pattern is used to characterize the welding area and the welding thickness of each area of ​​the structural component to be welded. The structural component to be welded characterizes the structural component to be welded in the current three-dimensional model.

[0068] Specifically, the current 3D model is aligned with a preset 3D model in terms of coordinates. Feature matching is used to find the target structural component in the preset model corresponding to the structural component to be welded in the current model. The stress map of the target structural component in the preset 3D model is then calculated. Based on the stress distribution at various locations of the target structural component, the area to be welded and the weld thickness at the weld joint are determined, ensuring that the structural component to be welded has the corresponding load-bearing capacity after construction. In other words, the welding position and thickness are determined based on the stress distribution map of the structural component to be welded at different weld joints, ensuring welding stability. The specific stress map is analyzed by substituting the structural component to be welded into the overall model (preset 3D model) after construction. When welding a specific structural component, the pre-stress map after substituting the structural component into the overall model is automatically selected, thereby choosing the welding position and weld thickness at each location to ensure that the welding meets future load-bearing requirements.

[0069] In some optional embodiments, determining the welding pattern based on the preset three-dimensional model, the current three-dimensional model, and the structural component to be welded includes:

[0070] S310. After mapping the current three-dimensional model to the preset three-dimensional model, determine the target structural component model of the structural component to be welded in the preset three-dimensional model;

[0071] Specifically, a coordinate registration algorithm (such as the iterative nearest-point algorithm) is used to spatially align the current 3D model obtained from the scan with a preset 3D model. Using the coordinate system of the preset 3D model as a reference, common feature points of the two models (such as bolt hole centers, component end face edges, bevel outlines, etc.) are extracted. The spatial distance between corresponding points is iteratively calculated and reduced until the registration error is controlled within the allowable range for engineering (e.g., ≤0.1mm for bridge steel structures, ≤0.5mm for conventional steel structures). Then, the structural component to be welded (i.e., the part that has not been welded and needs to be welded) is extracted from the current 3D model. Through geometric feature comparison (such as cross-sectional shape, size, and connection port type), a target structural component model that perfectly matches the structural component to be welded is found in the preset 3D model. Simultaneously, the geometric deviations between the structural component to be welded and the target structural component model are calculated, including assembly gap deviation, bevel angle deviation, and component position offset.

[0072] S320. Determine the stress spectrum corresponding to the target structural component model based on the positional relationship between the target structural component model and other structural component models in the preset three-dimensional model, the material information of each structural component, the shape information of each structural component, and the load information, wherein the load information represents the maximum load capacity of the target steel structure.

[0073] Specifically, the positional relationship refers to the connection method (rigid connection, hinged connection, simply supported) between the target structural component model and other structural component models in the preset 3D model, the stiffness difference between adjacent components, and the position of the target structural component in the overall steel structure (such as the mid-span of a bridge, the support, and other critical stress areas); the material information refers to the steel grade, density, elastic modulus, yield strength, and other parameters of the target structural component and the steel structural components that contribute to the gravity of the target structural component; the shape information refers to the shape and size information of the target structural component and the steel structural components that contribute to the gravity of the target structural component; and the load information refers to the maximum rated load capacity, load type (static load, dynamic load), and distribution form of the target steel structure.

[0074] The positional relationships, material information, shape information, and load information of each structural component are input into a pre-defined finite element model for stress simulation calculation. Boundary constraints are set according to the positional relationships (rigid connections are fixed, hinged connections are restricted in translation and released in rotation). Material properties are assigned based on the material information. Mesh generation is performed based on the shape information (mesh is refined in stress concentration areas such as nodes and bevels). Corresponding static loads (self-weight, auxiliary dead loads) and dynamic loads (vehicle loads, vibration loads) are applied according to the load information, and load combinations are performed according to industry standards. Finally, the stress spectrum of the target structural component model is generated through the calculation, and this spectrum visually presents the stress distribution of the structural component in the form of a cloud map.

[0075] S330. Obtain the weld information of the structural component to be welded, wherein the weld information indicates the weld type, weld size and weld location;

[0076] Specifically, weld information can be extracted from the target structural component model of the preset 3D model or from the structural component to be welded in the current 3D model. By identifying geometric mutations in point cloud data, parameters such as the actual weld location, actual bevel size, and assembly gap can be determined, thereby determining the weld type, weld size, and weld location.

[0077] S340. Determine the welding pattern based on the stress pattern and the weld information.

[0078] Specifically, the location coordinates of each weld are mapped onto a stress diagram to determine the stress zone of each weld, thereby classifying the stress level of the weld. Welds located in high-stress zones or stress concentration areas are defined as primary stress welds; welds located in medium-stress zones are defined as secondary stress welds; and welds located in low-stress zones are defined as non-stress welds. Then, welding process parameters are matched according to the stress level: for primary stress welds, the welding thickness required to meet penetration requirements (matching the bevel filling amount, appropriately increasing if the gap exceeds tolerance), the number of weld layers (multiple layers and multiple passes are required for thick plates), and the welding method (such as submerged arc welding) need to be determined; for secondary stress welds, the welding thickness is determined according to the designed weld size, and a more efficient welding method is adopted; for non-stress welds, meeting the basic connection strength is sufficient, the welding thickness is appropriately reduced, and the welding process is simplified.

[0079] Finally, a standardized welding atlas is generated: the atlas consists of two parts: a visual atlas and a data-driven atlas. The visual atlas uses the current 3D model as the base map, marking the main, secondary, and non-stressed welds with different colors, and indicating the welding thickness, number of weld layers, and welding sequence for each weld. The data-driven atlas is presented in tabular form, including information such as weld ID, location coordinates, stress level, welding thickness, process parameters (current, voltage, speed), and quality requirements, and can be directly imported into the welding robot's offline programming software.

[0080] In some optional embodiments, determining the force spectrum corresponding to the target structural component model based on the positional relationship between the target structural component model and other structural component models in the preset three-dimensional model, the material information of each structural component, the shape information of each structural component, and the load information of each structural component includes:

[0081] S321. Determine the connection constraints between the target structural component model and the adjacent structural component model based on the positional relationship, and determine the target position of the target structural component model in the target steel structure. The connection constraints include rigid connection constraints, hinged connection constraints, and simply supported constraints. The connection constraints are used to indicate the degree of freedom of movement of the target structural component model.

[0082] Specifically, the positional relationship between the target structural component model and adjacent structural component models in the preset 3D model is analyzed to obtain the connection method between the target structural component and adjacent structural components. That is, whether the target structural component and adjacent components are rigidly welded (corresponding to rigid connection constraint), pin-shaft connected (corresponding to hinge constraint), or fixed at one end and sliding at the other end (corresponding to simply supported constraint). Different connection methods correspond to different connection constraints. Each constraint clearly restricts the movement degree of freedom of the target structural component. Rigid connection constraint restricts 6 degrees of freedom (no translation, no rotation), hinge constraint restricts 3 translational degrees of freedom and releases 3 rotational degrees of freedom, and simply supported constraint restricts some translational degrees of freedom (such as restricting vertical displacement and releasing axial displacement to adapt to thermal expansion and contraction).

[0083] The target position of the target structural component model in the target steel structure is the specific position of the target structural component in the entire target steel structure (such as the mid-span position of the steel box girder of a bridge, the middle section of the load-bearing main beam of the equipment frame, the edge position of the auxiliary components, etc.).

[0084] S322. Determine the static load distribution map corresponding to the target structural component model based on the target location, the connection constraints, the material information of each structural component, and the shape information of each structural component.

[0085] Specifically, the static load borne by the target structural component mainly includes two parts: first, the self-weight load, which is obtained by multiplying the volume calculated based on the shape information by the steel density, and is the basic static load for all structural components; second, the auxiliary dead load, which is determined according to the target location (such as the weight of the guardrail of a bridge component or the weight of the equipment component itself), and is applied uniformly as a surface load or line load. Connection constraints are set as boundary conditions, material information is assigned as material properties, and the static load is applied to the target structural component model. Through static finite element analysis, the static load distribution map is finally generated.

[0086] S323. Determine the dynamic load distribution map corresponding to the target structural component model based on the load information, the target location, and the connection constraints;

[0087] Specifically, the load information includes the maximum rated load capacity of the target steel structure and the load type (moving load, such as vehicle load on a bridge; vibration load, such as equipment operating vibration; impact load, such as crane lifting impact). The target location determines the area of ​​action of the dynamic load (e.g., moving loads only act on the bridge's driveway components); the dynamic stiffness characteristics of the connection constraints must be considered (e.g., rigid constraints have high dynamic stiffness and fast vibration attenuation; hinged constraints have low dynamic stiffness and more pronounced vibration response). The load information is transformed into a finite element-recognizable dynamic load model: moving loads are transformed into surface loads moving along the target location, vibration loads are transformed into sinusoidal alternating loads, and impact loads are transformed into instantaneous pulse loads. Boundary conditions are set based on the dynamic stiffness characteristics of the connection constraints, and the dynamic loads are applied to the target structural component model. The dynamic load distribution map is then calculated through transient dynamics or modal analysis. The dynamic load distribution map is presented in the form of dynamic stress time history curves and fatigue stress amplitude cloud maps, which intuitively show the stress change law of the target structural component under dynamic loads, and mark the peak value, valley value and cycle number of stress cycles to reflect the "fatigue stress state" of the structural component and determine the fatigue resistance design requirements of the weld.

[0088] S324. The static load distribution map and the dynamic load distribution map are superimposed to obtain the force map.

[0089] Specifically, static and dynamic loads are multiplied by load partial factors and then superimposed; the static load partial factor is 1.2, and the dynamic load partial factor is 1.4, to simulate the most unfavorable stress condition of the structural component in actual service. Then, element-by-element superposition calculation is performed: the coordinate systems and mesh elements of the static and dynamic load distribution maps are fully aligned, and the comprehensive stress value is calculated for each finite element of the target structural component model according to the following formula: ,in, This is the combined stress value. This is the static stress value. This represents the peak dynamic stress. The final stress spectrum is generated based on the comprehensive stress values.

[0090] In some optional embodiments, determining the static load distribution map corresponding to the target structural component model based on the target location, the connection constraints, the material information of each structural component, and the shape information of each structural component includes:

[0091] S3221. Based on the target location, determine all associated structural component models in the preset three-dimensional model that contribute to the gravitational load of the target structural component model;

[0092] Specifically, the target location allows us to determine the spatial position of the target structural component within a pre-defined 3D model (e.g., the web of a bridge steel box girder, the main beam of an equipment frame), as well as its direct / indirect connection relationships with surrounding structural components. Based on these spatial connections, we can filter related structural component models: the gravity of related structural component models is directly transferred to the target structural component model through the connection surfaces. This includes directly related components and indirectly related components. Directly related components are those that are in direct contact with and rigidly connected to the target structural component model (e.g., top plates, bottom plates, and diaphragms directly welded to the box girder web); indirectly related components are auxiliary structural components fixed to directly related components, whose gravity needs to be transferred through these components (e.g., pavement layers and guardrail supports fixed to the top plate of the box girder).

[0093] S3222. Determine the gravity load distribution map of the target structural component model based on the material information of each structural component, the shape information of each structural component, and the associated structural component model.

[0094] Specifically, based on the shape information, parameters such as cross-sectional dimensions, length, and height of the target structural component model and related structural component models are extracted to calculate the volume; then, the weight of the component itself is calculated by combining the steel density from the material information. For each related structural component model, the weight transferred to the target structural component model is determined according to the load transfer ratio (e.g., the weight of the top plate is distributed to the two web plates according to the contact area ratio).

[0095] The gravity load is transformed into a load form that can be identified by finite element analysis: the self-weight of the target structural component is a body load, which acts uniformly on the entire structural component; the external gravity transmitted by the associated structural components is a surface load / line load, which acts on the connection contact surface between the two.

[0096] Constructing a finite element analysis model: The elastic modulus and Poisson's ratio from the material information are entered into the model. The target structural component and related structural components must match their parameters according to their respective material grades. The mesh is optimized based on the shape information. The mesh is refined in stress-sensitive areas such as nodes, bevels, and variable cross-sections of the target structural component, while large-size meshes are used in conventional areas to balance accuracy and efficiency. The calculated self-gravity and transmitted gravity are precisely applied to the corresponding locations on the model as volume loads and surface / line loads. Finally, the static finite element analysis model is run without boundary constraints (considering only free gravitational action), calculating the stress and strain distribution of the target structural component model and generating a gravity load distribution map.

[0097] S3223. The static load distribution map is obtained by correcting the gravity load distribution map according to the connection constraints.

[0098] Specifically, the connection constraints are transformed into finite element boundary conditions and applied to the connection points between the target structural component model and adjacent structural components:

[0099] If it is a rigid constraint (such as the target structural component being welded to an adjacent component to form a whole), it is set as a fixed constraint, which restricts the translation in the three directions of x, y, and z and the rotation about the three axes, for a total of 6 degrees of freedom;

[0100] If it is a hinged constraint (such as the target structural member being connected to an adjacent member by a pin), setting it as a hinged constraint restricts only 3 translational degrees of freedom and releases 3 rotational degrees of freedom;

[0101] If it is a simply supported constraint (such as the target structural member being fixed at one end and sliding at the other end), set it as a simply supported constraint, restricting 3 translational degrees of freedom at one end and restricting only vertical translation at the other end, releasing axial translational and rotational degrees of freedom.

[0102] The finite element analysis model with boundary constraints was recalculated: constraints prevent free deformation of structural components under gravity, causing stress changes in the constrained areas (e.g., increased stress at fixed supports), and simultaneously correcting the overall stress distribution. By comparing the gravity load distribution maps before and after correction, unreal stress regions under unconstrained conditions were eliminated, and the true static stress distribution under constraints was retained to generate the final static load distribution map.

[0103] In some optional embodiments, determining the dynamic load distribution pattern corresponding to the target structural component model based on the load information, the target location, and the connection constraints includes:

[0104] S3231. Determine the initial load distribution pattern of the target structural component model based on the load information and the target location;

[0105] Specifically, analyzing the load information yields: the maximum rated load capacity of the target steel structure; the load type, categorized as moving load (such as vehicle loads on bridges, crane lifting loads), vibration loads (such as alternating vibration loads during construction machinery operation), and impact loads (such as vehicle impacts on bridge components, and instantaneous impact loads during crane lifting); and the load application form, i.e., whether the load is a surface load (such as the pressure of a vehicle on the bridge deck), a line load (such as the force exerted by equipment vibration on the main beam), or a concentrated load (such as the point of application of an impact load).

[0106] The target location is used to filter the load's effective area: the target location determines whether a dynamic load acts on the structural member and the extent of its effect. For example, the top plate of a bridge steel box girder is located directly below the roadway and is the direct area affected by moving vehicle loads; while the web of the box girder is the indirect load transfer area, bearing only the dynamic component transmitted from the top plate; and bridge auxiliary components do not bear moving loads.

[0107] The load information is transformed into a finite element load model: for moving loads, the rated load is converted into a standard surface load value, and the boundary of the load's action area is defined according to the target location; for vibration loads, the rated load is combined with the equipment's vibration frequency to convert it into the amplitude of a sinusoidal alternating load (the load magnitude changes periodically with time according to a sine curve); for impact loads, the rated load is multiplied by the impact coefficient to convert it into the peak value of an instantaneous pulse load.

[0108] The transformed finite element load model is applied to the target structural component model without any connection constraints (simulating a free stress state). The spatial distribution of the load is solved statically to generate an initial load distribution map. This map is presented as a contour plot, visually displaying the static spatial stress distribution of the dynamic load on the target structural component.

[0109] S3232. Determine the dynamic load distribution pattern that changes over time based on the connection constraints and the initial load distribution pattern.

[0110] Specifically, connection constraints (rigid, hinged, simply supported) not only restrict the degrees of freedom of structural members, but also determine their dynamic stiffness. Dynamic stiffness is the ability of a structural member to resist deformation under dynamic loads, and the dynamic stiffness varies significantly among different constraints.

[0111] Rigid constraint: high dynamic stiffness, small deformation of structural components under dynamic loads, fast vibration decay rate, and short duration of stress peak.

[0112] Hinged constraint: low dynamic stiffness, structural components are allowed to rotate around the hinge point, large deformation under dynamic load, more obvious vibration response, and more stress cycle times;

[0113] Simply supported constraint: The dynamic stiffness is between the two, which only restricts displacement in some directions, and the dynamic response is manifested as local vibration.

[0114] The connection constraints are transformed into dynamic boundary conditions and applied to the connection points of the target structural component model. For example, rigid connection constraints correspond to fixed dynamic boundaries (restricting dynamic displacement in all degrees of freedom), and hinged connection constraints correspond to hinged dynamic boundaries (allowing dynamic displacement in the rotational direction). Then, a time dimension is introduced to simulate the change process of dynamic loads.

[0115] Moving load: Set the load to move along a preset path along the target location (e.g., from one end of the bridge to the other), divide the time into multiple time steps, each time step corresponds to a location of the load, and calculate the stress distribution at that location;

[0116] Vibration load: Set the alternation period of the load (e.g., the vibration frequency of the equipment is 50Hz), and calculate the stress amplitude at different times within each period according to the time step;

[0117] Impact load: Set the duration of the load (e.g., an instantaneous impact lasts only 0.1 seconds) and calculate the stress changes throughout the entire process of impact occurrence, peak value, and attenuation.

[0118] Finally, the dynamic boundary conditions, the time-varying load model, and the initial load distribution map are combined, and the dynamic stress response of the structure at each time step is simulated through transient dynamics or modal analysis, ultimately generating a time-varying dynamic load distribution map. The dynamic load distribution map can visually display the distribution location and peak value changes of stress on the structure at different times.

[0119] In some optional embodiments, determining the welding pattern based on the stress pattern and the weld information includes:

[0120] S341. Determine the stress coefficients of each region on the target structural component model based on the stress spectrum;

[0121] Specifically, the combined stress value (stress value resulting from the superposition of static and dynamic loads) of each region on the target structural component model, and the yield strength of the steel used in the target structural component. Then, the stress coefficient of each region is obtained by dividing the combined stress value by the yield strength.

[0122] S342. Determine the stress coefficient corresponding to the weld on the target structural component model based on the stress coefficient of each region and the location of the weld;

[0123] Specifically, the weld position coordinates are extracted from the weld information, and the centerline of each weld is mapped onto the region division map of the target structural component model to determine the stress coefficient region where each weld is located.

[0124] If the weld is entirely within a certain stress coefficient region, then the stress coefficient of the weld is equal to the stress coefficient of that region. If the weld crosses multiple stress coefficient regions (such as a weld passing through both a high-stress region and a medium-stress region), then the maximum value of the stress coefficient of each region is taken as the stress coefficient of the weld. If the weld is located at stress concentration points such as node intersections or bevel edges, then a correction factor is multiplied by the stress coefficient of the corresponding region to compensate for the additional stress risk caused by stress concentration.

[0125] S343. Determine the welding thickness and welding process of the weld based on the stress coefficient and the weld type;

[0126] Specifically, based on the weld type, such as butt welds (used for axial connections of components, such as segmental butt joints of bridge steel box girders), fillet welds (used for vertical connections of components, such as beam-column connections of frames), and penetration welds (used for high-strength connections of main load-bearing parts), a matching rule is established between stress coefficient, weld type, weld thickness, and welding process to obtain the specific weld thickness and welding process.

[0127] S344. Determine the welding area of ​​the weld based on the stress coefficient and the weld size.

[0128] Specifically, the weld size is extracted from the weld information, including the designed weld length, width, and bevel range. Then, based on the weld stress coefficient, the final weld area is adjusted and determined according to the following rules:

[0129] For welds with high stress coefficients: If the designed range of weld size does not completely cover the stress concentration area, the welding area needs to be extended outward by 5-10mm along the stress diffusion direction; for butt welds, the welding area needs to cover the entire bevel width, and an additional 10-15mm arc initiation / termination zone should be added at both ends to avoid incomplete penetration at the ends.

[0130] For welds with medium stress coefficient: the welding area is designed according to the weld size and no additional expansion is required; only a 5-10mm arc initiation / termination zone needs to be added at both ends of the weld to ensure the continuity of the weld formation.

[0131] Low stress coefficient welds: The welding area can be appropriately reduced according to the design range of the weld size (e.g., when intermittent welding is used, the length of the welding area is 70%-80% of the design length) to save welding materials and time.

[0132] The identified welding areas are marked on the structural components to be welded in the current 3D model, and the welding boundary and arc initiation / termination positions of each weld are clearly defined, forming a visual welding area division diagram.

[0133] Step S400: After iteratively optimizing the welding scheme corresponding to the welding pattern according to the preset decision variables and optimization objectives, the target welding scheme is obtained. The optimization objectives include minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency. The decision variables include welding torch path variables, welding torch process variables, and welding sequence variables.

[0134] Specifically, the decision variables include welding torch path variables (trajectory, travel speed, and posture), welding torch process variables (current, voltage, wire feed speed, and interpass temperature), and welding sequence variables (welding sequence of each weld, number of weld layers, and length of segmented skip welding). The four optimization objectives—minimizing welding path length, minimizing thermal stress deformation, minimizing welding torch collision risk, and maximizing welding efficiency—are transformed into quantifiable objective functions. Constraints such as structural obstruction, welding torch parameter range, and thermal deformation threshold are also set. The decision variables are iteratively adjusted through an intelligent optimization algorithm, continuously calculating the objective function values ​​corresponding to each variable combination, and selecting the optimal solution set that satisfies all constraints. Finally, considering the comprehensive engineering benefit requirements (e.g., bridges prioritize deformation control, while engineering machinery prioritizes efficiency improvement), the most suitable solution is selected from the optimal solution set, which is the target welding scheme.

[0135] In some optional embodiments, the step of iteratively optimizing the welding scheme corresponding to the welding pattern based on preset decision variables and optimization objectives to obtain the target welding scheme includes:

[0136] S410. Construct a welding neural network model;

[0137] Specifically, a welding-specific neural network model is built that integrates geometric feature extraction, process parameter mapping, and multi-objective prediction. The core function of this model is to quickly fit the complex relationship between decision variables and four optimization objectives, thereby significantly improving the efficiency of iterative optimization.

[0138] The model employs a three-layer hybrid architecture: geometric feature extraction, process parameter mapping, and multi-objective prediction. The first layer, the geometric feature extraction layer, integrates PointNet and CNN networks to extract key geometric information such as weld groove shape, component cross-sectional features, and spatial relationships from the 3D model. The second layer, the process parameter mapping layer, uses a fully connected neural network to transform parameters such as weld stress level and weld thickness from the welding atlas into process feature vectors recognizable by the model. The third layer, the multi-objective prediction layer, has four independent output heads, corresponding to the predicted values ​​of four optimization objectives: welding path length, thermal stress deformation, welding torch collision risk, and welding efficiency. During the model training phase, a large amount of historical welding case data (including 3D models of different steel structures, welding atlases, combinations of decision variables, and corresponding actual optimization objective results) is collected. Training is conducted with the goal of minimizing the mean square error between predicted and actual values ​​until the prediction error is controlled within an acceptable engineering range, ensuring the model possesses accurate predictive capabilities.

[0139] S420. Input the welding pattern, the current three-dimensional model, and the structural component to be welded into the welding neural network model;

[0140] Specifically, the welding pattern (including welding area, thickness, and process requirements), the current 3D model (including geometric deviations of the actual component), and the structural component to be welded (including weld location, type, and size) are standardized and preprocessed before being input into the welding neural network model as the basic data for optimization calculation.

[0141] S430. Configure the optimization objective of the welding neural network model as minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency.

[0142] Specifically, four optimization objectives are set for the welding neural network model: first, to minimize the welding path length and reduce the ineffective idle stroke of the welding torch; second, to minimize thermal stress deformation and control the deformation of the component after welding; third, to minimize the risk of welding torch collision and avoid interference between the welding torch and the component or fixture; and fourth, to maximize welding efficiency and shorten the overall welding time.

[0143] S440. Construct the objective function corresponding to the optimization objective;

[0144] Specifically, for the four interdependent optimization objectives, corresponding multi-objective functions are constructed. First, the objective values ​​are normalized to eliminate dimensional differences. Then, weight coefficients are assigned to different objectives according to engineering requirements to form a quantifiable and calculable comprehensive objective function, which is used to evaluate the optimization effect of each set of decision variables.

[0145] S450. Configure the design variables of the welding neural network model as the decision variables. The welding torch path variables include the motion trajectory of the welding torch, the walking speed of the welding torch, and the posture of the welding torch. The welding torch process variables include the welding current, the welding voltage, the wire feeding speed, and the interpass temperature. The welding sequence variables include the welding sequence of each weld, the welding layers of multiple weld layers, and the segment length of the segmented skip welding.

[0146] Specifically, three types of decision variables are set as the model's design variables (i.e., adjustable parameters): First, the welding torch path variable, which converts the motion trajectory into a discrete coordinate point sequence of the weld centerline, the walking speed into a continuously adjustable value of 0-500 mm / min, and the pose into continuous values ​​of the walking angle and working angle (0°-90°); Second, the welding torch process variable, which converts the welding current, voltage, and wire feed speed into continuously adjustable values ​​within the equipment's rated range (e.g., current 120-300A, voltage 20-35V), and the interpass temperature into a continuously controllable value ≤120℃; Third, the welding sequence variable, which converts the welding sequence of each weld into a discrete value of the permutation and combination of weld IDs, the welding layers of multiple weld layers into integer discrete values ​​matching the welding thickness (e.g., 3-8 layers), and the segment length of segmented skip welding into an optional discrete value of 50-300 mm.

[0147] S460. Configure the constraints of the welding neural network model as structural blocking constraints, welding torch working parameter constraints, and thermal deformation constraints.

[0148] Specifically, three types of rigid constraints are set for the model to ensure that the solution of the iterative optimization conforms to the actual feasibility of the engineering: First, structural blocking constraints, setting a minimum safe distance threshold (e.g., 10mm) between the welding torch and the robot arm and the components and fixtures. During the model iteration process, the welding torch trajectory corresponding to all combinations of decision variables must satisfy "minimum distance ≥ safe threshold", otherwise the solution will be directly eliminated; Second, welding torch working parameter constraints, limiting the welding current, voltage, wire feeding speed and other parameters to within the rated working range of the robot equipment to avoid equipment failure or weld quality defects due to parameters exceeding the standard; Third, thermal deformation constraints, setting a maximum allowable deformation threshold for the components after welding (e.g., ≤2mm for bridge components) according to the design requirements of the target steel structure. The thermal deformation predicted by the model must not exceed this threshold.

[0149] S470. Based on the constraints, the objective function, and the optimization objective, the optimal solution set is obtained by iterating the design variables.

[0150] Specifically, a hybrid strategy of "neural network prediction + intelligent optimization algorithm iteration" is adopted to drive iterative optimization of design variables and generate the optimal solution set.

[0151] First, multiple initial populations of design variables are randomly generated (e.g., 200 populations) to ensure that the values ​​of each variable are within the set range. Then, each group of variables is input into the welding neural network model to quickly predict the four corresponding optimization objective values, and these values ​​are substituted into the objective function to calculate the comprehensive score. Next, the population is screened according to the constraints, and schemes that violate structural obstruction, parameter range, and thermal deformation threshold are eliminated. Then, a genetic algorithm or particle swarm optimization algorithm is used to select, crossover, and mutate the selected feasible schemes to generate a new generation of design variable populations. The "prediction-screening-iteration" process is repeated until the objective function value does not decrease significantly for several generations, reaching a convergence state.

[0152] After the iteration is completed, a set of Pareto optimal solutions is output. Each solution in the set is a "non-dominated solution", which means that it is impossible to improve the performance of one optimization objective without sacrificing the other. Each solution represents a different objective balancing strategy.

[0153] S480. The target welding scheme is obtained by selecting the target solution from the set of optimal solutions based on the comprehensive benefit requirements.

[0154] Specifically, considering the comprehensive benefit requirements of the engineering scenario, the final target solution is selected from the Pareto optimal solution set and transformed into an executable target welding scheme.

[0155] In some optional embodiments, obtaining the optimal solution set by iterating the design variables based on the constraints, the objective function, and the optimization objective includes:

[0156] S471. Randomly generate a first number of first decision variable combinations, wherein the first decision variable combinations satisfy the constraints.

[0157] Specifically, the first quantity refers to the initial population size, such as 200 groups, and the constraints include structural blocking constraints, welding torch operating parameter constraints, and thermal deformation constraints.

[0158] Within the range of values ​​for the three types of decision variables (the trajectory coordinates, speed, and pose range of the welding torch path variable; the current and voltage range of the welding torch process variable; and the sequence combination and segment length range of the welding sequence variable), a first number of decision variable combinations are randomly generated.

[0159] During the generation process, constraint verification must be performed simultaneously: for each randomly generated combination, check whether it meets the structural blocking constraint (minimum distance between welding torch and component ≥ safety threshold), welding torch working parameter constraint (current, voltage, etc. are within the rated range of equipment), and thermal deformation constraint (predicted thermal deformation ≤ design threshold). Eliminate all combinations that violate the constraints, and finally obtain the first number of first decision variable combinations that fully satisfy the constraints, ensuring that the initial population is a feasible solution.

[0160] S472. Substitute the first combination of decision variables into the objective function to calculate the objective function value;

[0161] Specifically, each group of first decision variables is substituted into the constructed multi-objective weighted objective function, and the corresponding objective function value is calculated. The magnitude of the objective function value directly reflects the overall optimization effect of the combination. The smaller the function value, the better the overall performance of the combination in the four objectives of "minimizing path length, minimizing thermal deformation, minimizing collision risk, and maximizing efficiency".

[0162] S473. Retain the second number of combinations of the first decision variables that have the smallest objective function value, wherein the first number is greater than the second number;

[0163] Specifically, based on the calculated objective function value, the first number of combinations of the first decision variables are sorted, and the elite combinations of the second number of groups with the smallest objective function value are selected (the second number is the elite retention size, such as 50 groups).

[0164] S474. After performing parameter cross-validation and random parameter mutation on the retained first decision variable combination, a second decision variable combination is generated, and the iteration number is incremented by one.

[0165] Specifically, for the second number of elite combinations retained, parameter crossover and random parameter mutation operations are performed to generate new combinations of decision variables (i.e., the second combination of decision variables):

[0166] Parameter crossover: Select two or more groups from the elite combinations as the parent generation and partially exchange their decision variables. For example, exchange the welding torch walking speed parameters, welding sequence parameters, or some coordinate points of the welding torch trajectory between the two combinations to generate offspring combinations that combine the advantages of the parent generation;

[0167] Random parameter variation: Some parameters of the offspring combinations after crossover are slightly and randomly adjusted. For example, the welding torch current is finely adjusted by ±5A within the rated range, the segmented skip welding length is finely adjusted by ±10mm, or the welding torch pose angle is finely adjusted by ±2°. After generating the second decision variable combination, constraint verification is performed again, and combinations that violate the constraints are eliminated; at the same time, the iteration count is incremented by one, and the current iteration progress is recorded.

[0168] S475. When the number of iterations is less than or equal to a preset number and the objective function value has not converged, the second decision variable combination is configured as the first decision variable combination and then iterative calculation is performed. The iterative calculation represents the above-mentioned step of generating the second decision variable combination.

[0169] Specifically, two termination conditions are checked: first, whether the number of iterations is less than or equal to a preset number (the preset number is the upper limit of iterations, such as 50 generations); second, whether the objective function value has converged (the convergence criterion is: the change in the minimum value of the objective function over multiple consecutive generations is less than or equal to a preset threshold, such as 1%). If both termination conditions are met simultaneously (the number of iterations has not exceeded the upper limit and the objective function value has not converged), the generated combination of second decision variables is redefined as a new combination of first decision variables, and then the iterative calculation steps of "calculating the objective function value → retaining the elite combination → cross-mutation to generate a new combination" are repeated to continuously optimize the decision variables.

[0170] S476. If the number of iterations is greater than the preset number, or if the objective function value converges, the second decision variable combination is configured as the optimal solution set.

[0171] Specifically, if the number of iterations exceeds the preset number, or the objective function value reaches convergence, the iteration stops. At this point, all the final generated combinations of the second decision variables are collected. These combinations are all non-dominated solutions that satisfy all constraints and have relatively good objective function values ​​(i.e., cannot improve one objective without sacrificing another), and together they constitute the optimal solution set.

[0172] Step S500: The target steel structure is obtained by welding the structural component to be welded according to the target welding scheme.

[0173] Specifically, the steel structure welding robot is configured with various parameters according to the target welding plan: the motion trajectory coordinates of the welding torch, its walking speed and posture parameters; the welding current, voltage, wire feed speed, and interpass temperature for each weld segment; and the welding sequence, multi-layer weld division, and segmented skip weld lengths. For complex node welds, a manual pull-and-teach interface is reserved for local trajectory correction. The steel structure welding robot automatically generates the corresponding target welding program based on the configured parameters.

[0174] The steel structure welding robot performs welding according to the target welding procedure. During the process, the deviation between the welding torch and the weld seam is monitored in real time by the weld seam tracking sensor, and the trajectory is dynamically adjusted to ensure the fit. When necessary, it switches to force control mode and corrects the welding path in complex areas by manually pulling.

[0175] After welding is completed, the weld is visually inspected (forming, excess height, undercut, etc.) and non-destructive tested. At the same time, the dimensional accuracy of the welded component is compared with the preset three-dimensional model to confirm that all indicators meet the design requirements, and finally a qualified target steel structure is obtained.

[0176] The beneficial effects of this invention include: acquiring a preset three-dimensional model of the target steel structure, the preset three-dimensional model being used to characterize the complete structure of the steel structure; generating a current three-dimensional model after scanning the current steel structure; determining a welding pattern based on the preset three-dimensional model, the current three-dimensional model, and the structural component to be welded, the welding pattern being used to characterize the welding area and welding thickness of each area of ​​the structural component to be welded, the structural component to be welded representing the structural component to be welded in the current three-dimensional model; obtaining a target welding scheme by iteratively optimizing the welding scheme corresponding to the welding pattern based on preset decision variables and optimization objectives, the optimization objectives including minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency, the decision variables including welding torch path variables, welding torch process variables, and welding sequence variables; and obtaining the target steel structure by welding the structural component to be welded according to the target welding scheme. During welding, the steel structure welding robot compares the scanned current 3D model with the preset 3D model to determine the welding pattern of the structural component to be welded. It then optimizes the target welding scheme for the welding area on the welding pattern by using decision variables and optimization objectives. This enables the steel structure welding robot to accurately weld the structural component to be welded, resulting in good welding quality and high welding efficiency.

[0177] like Figure 2 As shown, Figure 2 A structural block diagram of a controller 1000 according to an embodiment of this application is shown. The components of the controller 1000 include, but are not limited to, a memory 1200 and a processor 1100. The processor 1100 is connected to the memory 1200 via a bus, and the memory 1200 is used to store data.

[0178] The controller 1000 also includes an access device that enables the controller 1000 to communicate via one or more networks. Examples of such networks include a Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the Internet. The access device may include one or more of any type of wired or wireless network interface (e.g., a Network Interface Card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Global System for Microwave Access (GSM) interface, or a Wi-Fi interface. MAX interface, Ethernet interface, Universal Serial Bus (USB) interface, Cellular network interface, Bluetooth interface, Near Field Communication (NFC) interface, etc.

[0179] The controller 1000 can be any type of stationary or mobile electronic device, including mobile computers or mobile electronic devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable electronic devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary electronic devices such as desktop computers or PCs. The controller 1000 can also be a mobile or stationary server.

[0180] The processor 1100 is used to execute computer-executable instructions for an automated welding method for steel structures.

[0181] The above is a schematic representation of a controller according to this embodiment. It should be noted that the technical solution of this controller belongs to the same concept as the technical solution of the aforementioned automatic welding method for steel structures. Details not described in detail in the controller's technical solution can be found in the description of the aforementioned automatic welding method for steel structures.

[0182] An automatic welding system for steel structures is also provided according to an embodiment of this application. The automatic welding system for steel structures includes a controller 1000, which automatically generates a construction plan. It should be noted that the technical solution of this automatic welding system for steel structures belongs to the same concept as the technical solution of the aforementioned automatic welding method for steel structures. Details not described in detail in the technical solution of the computing device can be found in the description of the technical solution of the aforementioned automatic welding method for steel structures.

[0183] This application embodiment also provides a storage medium, which is a computer-readable storage medium, storing a computer program that, when executed by a processor, implements the above-described automatic welding method for steel structures.

[0184] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof. The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separate, and may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0185] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as processors, such as central processing units, digital signal processors, or microprocessors executing software, or as hardware, or as integrated circuits, such as application-specific integrated circuits. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0186] The above provides a detailed description of the preferred embodiments of this application. However, this application is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. An automatic welding method for steel structures, characterized in that, Applied to steel structure welding robots, the methods include: Obtain a preset three-dimensional model of the target steel structure, which is used to characterize the complete structure of the steel structure; The current 3D model is generated after scanning the current steel structure; The welding pattern is determined based on the preset 3D model, the current 3D model, and the structural component to be welded. The welding pattern is used to characterize the welding area and the welding thickness of each area of ​​the structural component to be welded. The structural component to be welded represents the structural component to be welded in the current 3D model. The determination of the welding pattern based on the preset 3D model, the current 3D model, and the structural component to be welded includes: mapping the current 3D model to the preset 3D model, and determining the target structural component model of the structural component to be welded in the preset 3D model; determining the stress pattern corresponding to the target structural component model based on the positional relationship between the target structural component model and other structural component models in the preset 3D model, the material information of each structural component, the shape information of each structural component, and the load information, where the load information characterizes the maximum load capacity of the target steel structure; obtaining the weld information of the structural component to be welded, where the weld information indicates the weld type, weld size, and weld location; and determining the welding pattern based on the stress pattern and the weld information. The target welding scheme is obtained by iteratively optimizing the welding scheme corresponding to the welding pattern according to the preset decision variables and optimization objectives. The optimization objectives include minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency. The decision variables include welding torch path variables, welding torch process variables, and welding sequence variables. The target steel structure is obtained by welding the structural components to be welded according to the target welding scheme.

2. The automatic welding method for steel structures according to claim 1, characterized in that, The step of determining the force spectrum corresponding to the target structural component model based on the positional relationship between the target structural component model and other structural component models in the preset 3D model, the material information of each structural component, the shape information of each structural component, and the load information of each structural component includes: Based on the positional relationship, the connection constraints between the target structural component model and the adjacent structural component model are determined, as well as the target position of the target structural component model in the target steel structure is determined. The connection constraints include rigid connection constraints, hinged connection constraints, and simply supported constraints. The connection constraints are used to indicate the degree of freedom of movement of the target structural component model. The static load distribution map corresponding to the target structural component model is determined based on the target location, the connection constraints, the material information of each structural component, and the shape information of each structural component. The dynamic load distribution pattern corresponding to the target structural component model is determined based on the load information, the target location, and the connection constraints. The stress diagram is obtained by superimposing the static load distribution diagram and the dynamic load distribution diagram.

3. The automatic welding method for steel structures according to claim 2, characterized in that, The step of determining the static load distribution map corresponding to the target structural component model based on the target location, the connection constraints, the material information of each structural component, and the shape information of each structural component includes: Based on the target location, determine all associated structural component models in the preset three-dimensional model that contribute to the gravitational load of the target structural component model; The gravity load distribution map of the target structural component model is determined based on the material information of each structural component, the shape information of each structural component, and the associated structural component model. The static load distribution map is obtained by correcting the gravity load distribution map according to the connection constraints.

4. The automatic welding method for steel structures according to claim 2, characterized in that, The step of determining the dynamic load distribution map corresponding to the target structural component model based on the load information, the target location, and the connection constraints includes: The initial load distribution pattern of the target structural component model is determined based on the load information and the target location; The dynamic load distribution pattern that changes over time is determined based on the connection constraints and the initial load distribution pattern.

5. The automatic welding method for steel structures according to claim 1, characterized in that, Determining the welding pattern based on the stress pattern and the weld information includes: Determine the stress coefficients of each region on the target structural component model based on the stress spectrum; The stress coefficient corresponding to the weld on the target structural component model is determined based on the stress coefficient of each region and the location of the weld. The welding thickness and welding process of the weld are determined based on the stress coefficient and the weld type; The welding area of ​​the weld is determined based on the stress coefficient and the weld size.

6. The automatic welding method for steel structures according to claim 1, characterized in that, The step of iteratively optimizing the welding scheme corresponding to the welding pattern based on preset decision variables and optimization objectives to obtain the target welding scheme includes: Construct a welding neural network model; The welding pattern, the current 3D model, and the structural component to be welded are input into the welding neural network model; The optimization objective of the welding neural network model is configured as minimizing the welding path length, minimizing thermal stress deformation, minimizing the risk of welding torch collision, and maximizing welding efficiency. Construct the objective function corresponding to the optimization objective; The design variables of the welding neural network model are configured as the decision variables. The welding torch path variables include the welding torch trajectory, welding torch travel speed and welding torch pose. The welding torch process variables include welding current, welding voltage, wire feed speed and interpass temperature. The welding sequence variables include the welding sequence of each weld, the welding layers of multiple weld layers and the segment length of segmented skip welding. The constraints of the welding neural network model are configured as structural blocking constraints, welding torch operating parameter constraints, and thermal deformation constraints. The optimal solution set is obtained by iterating the design variables based on the constraints, the objective function, and the optimization objective. The target welding scheme is obtained by selecting the target solution from the set of optimal solutions based on the overall benefit requirements.

7. The automatic welding method for steel structures according to claim 6, characterized in that, The process of obtaining the optimal solution set by iterating the design variables based on the constraints, the objective function, and the optimization objective includes: A first number of first decision variable combinations are randomly generated, and the first decision variable combinations satisfy the constraints. Substitute the first combination of decision variables into the objective function to calculate the objective function value; The second number of combinations of the first decision variables that minimize the objective function value is retained, where the first number is greater than the second number. The first combination of decision variables is subjected to parameter cross-validation and random parameter mutation to generate the second combination of decision variables, and the iteration number is incremented by one. If the number of iterations is less than or equal to a preset number and the objective function value has not converged, the second decision variable combination is configured as the first decision variable combination and then iterative calculation is performed. The iterative calculation represents the above-mentioned step of generating the second decision variable combination. If the number of iterations is greater than a preset number, or if the objective function value converges, the second combination of decision variables is configured as the optimal solution set.

8. An automatic welding system for steel structures, characterized in that, The system includes a controller, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the automatic welding method for steel structures according to any one of claims 1-7.

9. A computer storage medium, characterized in that, The computer storage medium stores computer-executable instructions for executing the automatic welding method for steel structures according to any one of claims 1-7.

Images