Modular welding assembly method and system for steel structure beam column

Abstract

This invention discloses a modular welding assembly method and system for steel structure beams and columns, relating to the field of intelligent building industrial technology. The method includes: dividing modular units based on original design data; constructing an actual geometric digital twin model containing manufacturing errors and embedding digital anchor point identifiers; establishing a virtual pre-assembly environment; calculating node gaps, connection plate misalignment, and bevel deviations to generate virtual pre-assembly deviation data; executing a hierarchical error compensation algorithm based on the deviation data and initial thermal deformation coefficient to calculate manufacturing geometric errors, assembly posture errors, and welding thermal deformation prediction compensation matrices, generating welding trajectory correction amounts; establishing a coordinate system mapping on-site through digital anchor points; controlling welding based on the correction amounts and collecting interlayer temperatures; triggering thermal deformation compensation when temperatures exceed limits; collecting post-weld data to correct the thermal deformation coefficient and transmitting it back for optimization. This invention achieves closed-loop control between virtual pre-simulation in the factory and precise execution on-site, solving the industry problem of controlling manufacturing errors and welding deformation.

Description

Technical Field

[0001] This invention relates to the field of intelligent building industrial technology, specifically to a modular welding assembly method and system for steel structure beams and columns. Background Technology

[0002] Modular steel structure construction technology significantly improves construction efficiency through factory prefabrication and on-site assembly. Beam-column joints, as critical force-transmitting components, directly determine the overall structural safety through their welding and assembly precision. Currently, the industry mainly employs two methods: one is BIM-based virtual construction technology, which performs component collision checks and hoisting path planning in software; the other is using automated welding equipment, executing standardized weld operations through teach-through programming or offline programming. However, existing technical solutions have significant limitations. BIM virtual pre-assembly is mostly based on ideal design models and does not fully consider the actual manufacturing errors generated during component cutting and assembly, resulting in a "disconnect between the virtual environment and the physical entity." During on-site assembly, the cumulative errors of previous processes cause node gaps and misalignments of connecting plates to frequently exceed allowable deviations. Traditional rigid tooling is difficult to adaptively adjust and can only rely on manual repair or rework. In the welding process, the prediction of thermal deformation for multi-layer, multi-pass welding of thick plates still relies on fixed empirical parameters and cannot dynamically adjust the process according to real-time interlayer temperatures. Furthermore, the high-precision machining benchmarks in the factory are difficult to reproduce in high-altitude operations on-site, causing the welding trajectory to deviate from the preset path. More importantly, existing technologies lack a data closed loop from virtual pre-simulation, on-site execution to quality feedback. The thermal deformation coefficient cannot be self-learned and optimized based on actual welding results, leading to the recurrence of the same deviations in similar components. Therefore, there is an urgent need for a modular welding assembly method for steel structure beams and columns that can integrate actual manufacturing errors, achieve accurate virtual-on-site coordinate mapping, and have the ability to self-evolve parameters, in order to solve the industry problem of error accumulation and welding deformation control of non-standard components. Summary of the Invention

[0003] To address the aforementioned technical problems, this paper provides a modular welding assembly method and system for steel structure beams and columns. This technical solution resolves the contradictions between design idealization and manufacturing reality, the disconnect between virtual simulation and physical execution, and the problem of fixed and non-evolvable welding process parameters in the modular construction of steel structures.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The modular welding assembly method for steel structure beams and columns includes the following steps: S1. Based on the original design data, the steel structure frame is divided into modular beam and column units that can be independently hoisted. After prefabrication in the factory, each unit is subjected to three-dimensional laser scanning to construct an actual geometric digital twin model that includes manufacturing errors, and digital anchor point markers are embedded and initial thermal deformation coefficients are set. S2. Based on the digital twin model, establish a virtual pre-assembly environment, perform multi-body assembly interference detection, calculate the node gap, connecting plate misalignment and bevel geometric parameter deviation caused by the accumulation of manufacturing errors, and generate virtual pre-assembly deviation data. S3. Based on the deviation data and the initial thermal deformation coefficient, execute the hierarchical error compensation algorithm to calculate the manufacturing geometric error compensation matrix, the assembly posture error compensation matrix and the welding thermal deformation prediction compensation matrix, generate the welding deformation prediction value and the comprehensive welding trajectory correction amount, and adjust the welding process parameters based on the correction amount. S4. Transport the modular unit to the site, establish the mapping relationship between the site coordinate system and the factory virtual coordinate system through the implanted digital anchor point markers, control the welding equipment to perform welding based on the welding trajectory correction amount and the adjusted welding process parameters, collect the interpass temperature in real time, and trigger the thermal deformation compensation mode to dynamically adjust the welding speed when the temperature exceeds the limit. S5. Based on the field coordinate system, collect the post-weld geometric dimensions and inspection data, calculate the deviation between the actual welding deformation and the predicted welding deformation value, and transmit the corrected thermal deformation coefficient back to the digital twin model to optimize the virtual pre-assembly parameters of subsequent similar units.

[0005] Preferably, in step S1, the step of constructing the actual geometric digital twin model including manufacturing errors is as follows: A laser tracker or photogrammetry system is used to perform a full-domain 3D scan of the modular unit to obtain the original point cloud data; The original point cloud data is processed for denoising, simplification, and coordinate unification. The processed measured point cloud data is registered and compared with the theoretical geometric model corresponding to the original design data using the ICP algorithm, and the manufacturing deviation of each feature point is calculated. Based on the measured point cloud, the actual surface geometry is reconstructed to generate an actual geometric digital twin model carrying the manufacturing deviation as a geometric attribute field.

[0006] Preferably, in step S1, the implantation of digital anchor point identification is performed as follows: Countersunk holes are made at the corners of the beam-column joint connection plates of the modular unit, and high-temperature resistant ceramic substrates are implanted therein; The substrate surface is etched with a machine vision-recognizable QR code pattern, and an RFID chip storing a unique number is encapsulated inside. The QR code is associated with and bound to the RFID chip number to serve as the geometric origin marker of the actual geometric digital twin model.

[0007] Preferably, in step S2, the step of generating virtual pre-assembly deviation data is as follows: Each of the actual geometric digital twin models is imported into the virtual pre-assembly environment according to the assembly sequence determined during the division. Assembly constraints between each of the modular units are set, including node plate contact constraints, weld bevel fit constraints, and bolt hole alignment constraints. Based on the assembly constraint relationship, multi-body assembly interference detection is performed to identify the geometric interference region between the digital twin models, calculate the spatial distance information of the geometric interference region, extract the node gap, connecting plate misalignment and bevel geometric parameter deviation, and generate virtual pre-assembly deviation data containing deviation type, deviation amount and deviation location.

[0008] Preferably, in step S3, the step of calculating the manufacturing geometric error compensation matrix is ​​as follows: Extract the geometric feature points of each connecting plate of the modular unit from the virtual pre-assembly deviation data, and establish the local coordinate system of the actual geometric digital twin model; The manufacturing geometric error compensation matrix is ​​determined by rigid body transformation calculation of the global coordinate system of the theoretical geometric model corresponding to the local coordinate system and the original design data. The spatial coordinates of the welding start point are corrected based on the manufacturing geometric error compensation matrix to form the position reference component of the comprehensive welding trajectory correction amount.

[0009] Preferably, in step S3, the step of calculating the assembly pose error compensation matrix is ​​as follows: Based on the virtual pre-assembly deviation data, the six-degree-of-freedom pose deviation of the modular unit during the assembly process is determined. The six-degree-of-freedom pose deviation includes translational deviation along the X, Y, and Z axes and rotational deviation around the three axes. The six-degree-of-freedom pose deviations are adjusted based on the multibody constraint solution algorithm so that the node gaps and connecting plate misalignments meet the tolerances required by the assembly process, and the assembly pose error compensation matrix is ​​determined. The welding posture angle is corrected based on the assembly posture error compensation matrix to form the posture component of the comprehensive welding trajectory correction amount.

[0010] Preferably, in step S3, the welding thermal deformation prediction and compensation matrix step is as follows: Calculate the welding heat input based on the welding current, voltage, and speed in the welding process parameters. Based on the initial thermal deformation coefficient and welding heat input, the welding temperature field and stress-strain field of the modular unit are predicted by the thermo-elastic-plastic finite element model. The welding deformation is predicted based on the welding temperature field and stress-strain field, and the welding thermal deformation prediction compensation matrix is ​​determined. The deformation offset of the welding path is pre-compensated based on the welding thermal deformation prediction compensation matrix to form the dynamic compensation component of the comprehensive welding trajectory correction.

[0011] Preferably, in step S4, establishing the mapping relationship between the on-site coordinate system and the factory virtual coordinate system through the implanted digital anchor point identifiers includes: The machine vision system identifies the QR code pattern on the surface of the digital anchor point marker to obtain the on-site spatial coordinates of the modular unit. Based on the registration between the on-site spatial coordinates and the geometric origin markers in the actual geometric digital twin model, a mapping relationship between the on-site coordinate system and the factory virtual coordinate system is established. The thermal deformation compensation mode includes: when the interpass temperature exceeds the interpass temperature control threshold in the welding process parameters, dynamically adjusting the welding current, welding speed and welding torch swing based on the difference between the interpass temperature and the interpass temperature control threshold, so as to control the heat input and maintain the weld formation quality.

[0012] Preferably, step S5 is as follows: The actual post-welding geometry of the modular unit is obtained by using three-dimensional laser scanning or total station measurement; The actual post-weld geometry is compared with the theoretical post-weld geometry corresponding to the predicted welding deformation value to calculate the actual welding deformation. Based on the difference between the actual welding deformation and the predicted welding deformation, the initial thermal deformation coefficient is corrected by the parameter identification algorithm to generate the corrected thermal deformation coefficient. The corrected thermal deformation coefficient is updated to the digital twin model and used as the initial thermal deformation coefficient for subsequent similar modular units to optimize the accuracy of thermal deformation prediction in the virtual pre-assembled environment.

[0013] A modular welding assembly system for steel structure beams and columns, applied to the aforementioned modular welding assembly method for steel structure beams and columns, includes the following modules: The digital twin construction module is used to divide the steel structure frame into independently hoistable beam and column modular units based on the original design data, and to perform three-dimensional laser scanning on each unit after factory prefabrication to construct an actual geometric digital twin model including manufacturing errors and initial thermal deformation coefficients. The modular units are embedded with digital anchor point identifiers. The virtual pre-assembly module is used to establish a virtual pre-assembly environment based on the digital twin model, perform multi-body assembly interference detection, calculate the node gap, connecting plate misalignment and bevel geometric parameter deviation caused by the accumulation of manufacturing errors, and generate virtual pre-assembly deviation data. The error compensation decision module is used to execute a hierarchical error compensation algorithm based on the deviation data and the initial thermal deformation coefficient, calculate the manufacturing geometric error compensation matrix, the assembly posture error compensation matrix and the welding thermal deformation prediction compensation matrix, generate the welding deformation prediction value and the comprehensive welding trajectory correction amount, and adjust the welding process parameters based on the correction amount. The on-site welding execution module is used to transport the modular unit to the site, establish a mapping relationship between the site coordinate system and the factory virtual coordinate system through the digital anchor point identifier, control the welding equipment to perform welding based on the welding trajectory correction amount and the adjusted welding process parameters, and collect the interpass temperature in real time and trigger the thermal deformation compensation mode to dynamically adjust the welding speed when the temperature exceeds the limit. The quality feedback optimization module is used to collect post-weld geometric dimensions and inspection data based on the field coordinate system, calculate the deviation between the actual welding deformation and the predicted welding deformation value, and send the corrected thermal deformation coefficient back to the digital twin construction module to optimize the virtual pre-assembly parameters of subsequent similar units.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes to construct a real geometric digital twin model containing manufacturing errors based on measured point clouds. By driving virtual pre-assembly with real geometric data, it achieves accurate mapping between factory processing status and virtual simulation environment, eliminating systematic deviations caused by design idealization. It also proposes to perform multi-body assembly interference detection in the virtual environment and generate a hierarchical error compensation matrix. Through pre-decoupling calculation of manufacturing geometric errors, assembly pose errors and welding thermal deformation, it achieves pre-compensation of welding trajectory, transforming traditional "on-site repair" into "factory pre-rehearsal" and significantly reducing rework rate. Furthermore, it proposes a thermal deformation compensation mode based on real-time interlayer temperature acquisition. Through dynamic coordinated adjustment of current, speed and amplitude, it achieves adaptive control of heat input during welding process, effectively suppressing welding residual stress and deformation accumulation. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system framework diagram of the present invention. Detailed Implementation

[0016] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0017] Reference Figure 1 As shown, the modular welding assembly method for steel structure beams and columns includes the following steps: S1. Based on the original design data, the steel structure frame is divided into modular beam and column units that can be independently hoisted. After prefabrication in the factory, each unit is subjected to three-dimensional laser scanning to construct an actual geometric digital twin model that includes manufacturing errors, and digital anchor point markers are embedded and initial thermal deformation coefficients are set. The steps for constructing a real geometric digital twin model that includes manufacturing errors are as follows: A laser tracker or photogrammetry system is used to perform full-area 3D scanning of the modular unit to obtain the original point cloud data. In this embodiment, the Leica AT960 laser tracker is used as the core measuring device, in conjunction with a high-precision photogrammetry system, to ensure a spatial resolution of 0.1 mm within a 10-meter measurement range. To achieve full-area scanning of the component without blind spots, multiple scanning stations need to be arranged around the component. Through multi-angle data stitching technology, the occlusion blind spots caused by a single viewpoint are eliminated, ensuring the integrity of the point cloud data.

[0018] The original point cloud data is subjected to denoising, simplification, and coordinate unification. The denoising process uses a statistical filtering algorithm to remove outliers far from the surface of the component by setting reasonable neighborhood radii and standard deviation thresholds. The simplification process uses a voxel grid downsampling strategy to reduce data density and improve computational efficiency while maintaining geometric features. Coordinate unification is achieved by placing standard target spheres on the component, establishing a local coordinate system using the center coordinates of the target spheres, and normalizing the multi-station data to a unified world coordinate system through a coordinate transformation matrix.

[0019] The processed measured point cloud data is registered and compared with the theoretical geometric model corresponding to the original design data using the ICP algorithm to calculate the manufacturing deviation of each feature point. Among them, the accurate alignment between the measured point cloud and the theoretical model is the key step in extracting manufacturing errors. The implementation process is divided into two stages: the initial registration stage uses artificial markers such as target balls to perform coarse alignment and establish a preliminary correspondence between the two coordinate systems; the precise registration stage uses the Iterative Closest Point (ICP) algorithm to continuously optimize the rotation matrix and translation vector by minimizing the distance between points or points on opposite sides until the registration error converges to a preset threshold (such as within 0.05 mm), ensuring the spatial consistency between the measured data and the theoretical model.

[0020] Based on the measured point cloud, the actual surface geometry is reconstructed to generate an actual geometric digital twin model carrying the manufacturing deviation as a geometric attribute field.

[0021] The steps for implanting digital anchor points are as follows: Countersunk holes are made at the corners of the beam-column joint connection plate of the modular unit to insert a high-temperature resistant ceramic substrate. The material selection and structural design of the high-temperature resistant ceramic substrate are the basis for ensuring its long-term stable operation in the high-temperature welding environment. In this embodiment, zirconia ceramic or silicon nitride ceramic is selected as the substrate material. Both have excellent high-temperature resistance (continuous temperature resistance ≥800℃) and thermal shock resistance, and can withstand the instantaneous high-temperature impact during the welding process. The substrate adopts a disc-shaped structure design with a diameter of 20 mm and a thickness of 5 mm. This size takes into account both mechanical strength and ease of installation. Countersunk holes are made at the corners of the beam-column joint connection plate. The hole diameter matches the size of the substrate. The hole position accuracy is ensured by CNC precision machining or water jet cutting process, so that the substrate is flush with the surface of the connection plate after insertion, avoiding interference or thermal stress concentration caused by protrusions during the welding process. The substrate surface is etched with a machine vision-recognizable QR code pattern, and an RFID chip storing a unique number is encapsulated inside. The QR code is associated with and bound to the RFID chip number to serve as the geometric origin marker of the actual geometric digital twin model.

[0022] Among them, machine-readable QR code and radio frequency identification (RFID) constitute a dual-mode sensing system of digital anchors. The QR code uses laser etching technology to prepare DataMatrix two-dimensional barcodes on the surface of ceramic substrate, and selects a medium error correction level (ECC200, error correction capability M level) to ensure that it can still be accurately read by machine vision when there is slight contamination or local wear. The RFID chip uses industrial-grade high-temperature resistant tags (such as the HID IronTag series), which can withstand continuous temperature of 250℃ and instantaneous temperature of up to 300℃, meeting the working environment requirements near the heat-affected zone of welding. The chip is implanted in a pre-made cavity inside the substrate and sealed and fixed with high-temperature silicone or ceramic glue, which protects the chip from high temperature damage and ensures that the QR code surface is flat and does not hinder visual recognition.

[0023] The rigid mapping between physical anchor points and digital twin models is achieved through the factory information system. Before the components leave the factory, machine vision is used to read the visual identification ID of the QR code, and at the same time, the EPC code of the chip is obtained through RFID reader. The two establish a one-to-one binding relationship in the database of the Manufacturing Execution System (MES). This binding relationship associates the unique identity of the physical component with the geometric origin mark, manufacturing deviation data and other attributes in its digital twin model, forming a precise data link from the factory virtual environment to the physical world on site. This ensures that the theoretical coordinate parameters of the corresponding digital twin model can be quickly and accurately retrieved when the on-site coordinate system is established later.

[0024] S2. Based on the digital twin model, establish a virtual pre-assembly environment, perform multi-body assembly interference detection, calculate the node gap, connecting plate misalignment and bevel geometric parameter deviation caused by the accumulation of manufacturing errors, and generate virtual pre-assembly deviation data. The steps for generating virtual pre-assembly deviation data are as follows: Each of the actual geometric digital twin models is imported into the virtual pre-assembly environment according to the assembly sequence determined during the division. Assembly constraints are set between each of the modular units, including node plate contact constraints, weld bevel fit constraints, and bolt hole alignment constraints. In the construction of the virtual pre-assembly environment, the choice of software platform and the compatibility of data formats determine the efficiency of multi-source heterogeneous data fusion. In this embodiment, Unity3D or Bentley SYNCHRO is used as the core virtual simulation engine. Both have powerful three-dimensional visualization capabilities and real-time rendering performance, and can carry out detailed modeling of large-scale steel structure nodes. The digital twin models are imported into the virtual environment in the IFC (Industry Foundation Classes) standard format. This format is a universal exchange standard for BIM data, which completely preserves the geometric information, attribute fields, and spatial relationships of the components, ensuring that the manufacturing deviation data measured in the factory can be transmitted to the virtual pre-assembly platform without loss, and realizing the accurate correspondence between physical entities and virtual models. Based on the assembly constraint relationship, multi-body assembly interference detection is performed to identify the geometric interference region between the digital twin models, calculate the spatial distance information of the geometric interference region, extract the node gap, connecting plate misalignment and bevel geometric parameter deviation, and generate virtual pre-assembly deviation data containing deviation type, deviation amount and deviation location.

[0025] Among them, the mathematical modeling of assembly constraints is the key to simulating the real physical assembly process. For the typical connection method of steel structure beam-column nodes, this embodiment defines three types of core constraints: node plate contact constraint requires that the normal distance of the contact plane of adjacent components does not exceed 0.5 mm, simulating the tight fit after the bolts are tightened; weld groove fit constraint limits the blunt edge gap to the range of 0 to 3 mm. This tolerance takes into account both the welding penetration requirement and allows for the existence of manufacturing errors; bolt hole alignment constraint uses the hole center coordinate deviation ≤ 2 mm as the threshold to ensure that high-strength bolts can be smoothly inserted. These constraints are expressed by a system of mathematical inequalities, which constitute the judgment criteria for multi-body assembly interference detection. Interference detection and deviation quantification are achieved through a hierarchical spatial collision algorithm. In the initial stage, axis-aligned bounding boxes (AABB) or oriented bounding boxes (OBB) are used for coarse detection to quickly eliminate obviously non-interfering component combinations. For potential interference areas, a precise point-to-plane distance formula is used to calculate gap values ​​and identify specific deviations such as node plate misalignment and excessive bevel gaps. After detection, the system automatically generates a structured deviation data table, storing the deviation type (gap / misalignment / angle deviation), numerical value, and spatial coordinates in CSV format. This provides directly callable quantification input for subsequent hierarchical error compensation algorithms.

[0026] S3. Based on the deviation data and the initial thermal deformation coefficient, execute the hierarchical error compensation algorithm to calculate the manufacturing geometric error compensation matrix, the assembly posture error compensation matrix and the welding thermal deformation prediction compensation matrix, generate the welding deformation prediction value and the comprehensive welding trajectory correction amount, and adjust the welding process parameters based on the correction amount. The steps for calculating the manufacturing geometric error compensation matrix are as follows: Extract the geometric feature points of each connecting plate of the modular unit from the virtual pre-assembly deviation data, and establish the local coordinate system of the actual geometric digital twin model; The manufacturing geometric error compensation matrix is ​​determined by rigid body transformation calculation of the global coordinate system of the theoretical geometric model corresponding to the local coordinate system and the original design data. The spatial coordinates of the welding start point are corrected based on the manufacturing geometric error compensation matrix to form the position reference component of the comprehensive welding trajectory correction amount.

[0027] Among them, the quantification and correction of manufacturing geometric errors is the primary step in ensuring the accuracy of the welding start point. For each connecting plate of the modular unit, the system first extracts the three-dimensional coordinates of its corner feature points and establishes a local coordinate system based on the principal inertial axis principle to characterize the actual spatial attitude of the connecting plate. The rigid body transformation matrix between the measured local coordinate system and the theoretical global coordinate system is solved by the singular value decomposition (SVD) algorithm. This matrix contains the rotation matrix R and the translation vector T, which directly reflects the geometric offset generated during the manufacturing process. Based on this matrix, the system calculates the XYZ direction offset (Δx, Δy, Δz) of the welding start point, which constitutes the position reference component of the comprehensive welding trajectory correction amount, ensuring that the welding torch starts from the actual physical weld seam, rather than the theoretical design position.

[0028] The steps for calculating the assembly pose error compensation matrix are as follows: Based on the virtual pre-assembly deviation data, the six-degree-of-freedom pose deviation of the modular unit during the assembly process is determined. The six-degree-of-freedom pose deviation includes translational deviation along the X, Y, and Z axes and rotational deviation around the three axes. The six-degree-of-freedom pose deviations are adjusted based on the multibody constraint solution algorithm so that the node gaps and connecting plate misalignments meet the tolerances required by the assembly process, and the assembly pose error compensation matrix is ​​determined. The welding posture angle is corrected based on the assembly posture error compensation matrix to form the posture component of the comprehensive welding trajectory correction amount.

[0029] Among them, the optimization of assembly posture deviation is the core of achieving precise assembly. The system obtains the centroid coordinates and attitude quaternions of the unit from the virtual pre-assembly module, and calculates the six-degree-of-freedom posture deviation, including translational deviations (Δtx, Δty, Δtz) along the X, Y, and Z axes and rotational deviations (Δθx, Δθy, Δθz) around the three axes. In order to minimize the posture adjustment amount under the premise of satisfying the assembly constraints, the gradient descent method is used for optimization iteration, so that the node gap and the misalignment of the connecting plate converge to the preset tolerance range. The optimized posture deviation is converted into the welding torch posture adjustment amount, which corrects the welding tilt angle and extension, and constitutes the posture component of the comprehensive welding trajectory correction amount, ensuring that the welding torch enters the bevel with the best posture.

[0030] The steps for establishing the welding thermal deformation prediction and compensation matrix are as follows: Calculate the welding heat input based on the welding current, voltage, and speed in the welding process parameters. Based on the initial thermal deformation coefficient and welding heat input, the welding temperature field and stress-strain field of the modular unit are predicted by the thermo-elastic-plastic finite element model. The welding deformation is predicted based on the welding temperature field and stress-strain field, and the welding thermal deformation prediction compensation matrix is ​​determined. The deformation offset of the welding path is pre-compensated based on the welding thermal deformation prediction compensation matrix to form the dynamic compensation component of the comprehensive welding trajectory correction.

[0031] Among them, the prediction and pre-compensation of welding thermal deformation is the key to controlling residual stress. In this embodiment, a thermo-elastic-plastic finite element model is established on the ABAQUS or ANSYS platform. The material constitutive model adopts a bilinear kinematic hardening model to accurately describe the stress-strain relationship at high temperature. The heat source model uses a Goldak double ellipsoidal heat source or a conical heat source to simulate the arc heat input. The boundary conditions are set with symmetric constraints and clamping constraints to simulate the actual clamping state. Through transient thermal analysis coupled with structural analysis, the temperature field distribution, peak temperature and residual stress and deformation field after cooling are predicted during the welding process. Based on the prediction results, the system pre-superimposes the reverse deformation offset on the welding path to form a dynamic compensation component of the comprehensive welding trajectory correction, realizing pre-compensation before welding rather than post-weld correction.

[0032] Ultimately, the three compensation components are fused together through spatial coordinate transformation and attitude interpolation algorithm to form a complete comprehensive welding trajectory correction. The position reference component determines the spatial start and end points of the welding torch, the attitude component adjusts the relative angle between the welding torch and the workpiece, and the dynamic compensation component corrects the accumulation of thermal deformation during the welding process in real time. The three components work together to ensure that the weld can still achieve the design requirements for penetration depth and forming quality under complex working conditions with manufacturing errors and thermal deformation.

[0033] S4. Transport the modular unit to the site, establish the mapping relationship between the site coordinate system and the factory virtual coordinate system through the implanted digital anchor point markers, control the welding equipment to perform welding based on the welding trajectory correction amount and the adjusted welding process parameters, collect the interpass temperature in real time, and trigger the thermal deformation compensation mode to dynamically adjust the welding speed when the temperature exceeds the limit. The process of establishing a mapping relationship between the on-site coordinate system and the factory virtual coordinate system through implanted digital anchor point identifiers includes: The machine vision system identifies the QR code pattern on the surface of the digital anchor point marker to obtain the on-site spatial coordinates of the modular unit. Based on the registration between the on-site spatial coordinates and the geometric origin markers in the actual geometric digital twin model, a mapping relationship between the on-site coordinate system and the factory virtual coordinate system is established. In the on-site welding execution process, an intelligent total station (such as Trimble SX12) or a portable laser tracker is used as the on-site measurement reference device. The built-in high-resolution camera performs visual recognition of the digital anchor point markers of the implanted components, reads the QR code pattern to obtain the spatial coordinates (X', Y', Z') of the anchor point on the construction site. Subsequently, the system retrieves the theoretical geometric origin coordinates (X, Y, Z) of the corresponding anchor point in the factory digital twin model, and uses the least squares registration algorithm to solve for the optimal transformation matrix (rotation matrix R and translation vector T) between the on-site coordinate system and the factory virtual coordinate system, achieving a coordinate system with millimeter-level accuracy. The thermal deformation compensation mode includes: when the interpass temperature exceeds the interpass temperature control threshold in the welding process parameters, dynamically adjusting the welding current, welding speed and welding torch swing based on the difference between the interpass temperature and the interpass temperature control threshold, so as to control the heat input and maintain the weld formation quality.

[0034] It should be noted that thermal state monitoring during the welding process is the data foundation for achieving dynamic compensation. An infrared temperature sensor (such as the OPTRIS CTlaser MT type) is integrated into the tail of the welding robot's welding torch. This sensor has non-contact measurement capabilities and fast response characteristics. It collects the interpass temperature of the weld and heat-affected zone in real time at a sampling frequency of 10Hz. The collected data is transmitted to the welding controller in real time via the industrial Ethernet EtherCAT bus and compared with the preset temperature control threshold to provide a decision basis for triggering the thermal deformation compensation mode. The execution logic of the heat deformation compensation mode is based on the thermophysical properties of the material and real-time temperature feedback. The system calculates the interlayer temperature control threshold according to the carbon equivalent (CE) of the component material. In this embodiment, it is set to 200℃. When the infrared sensor detects that the interlayer temperature exceeds the threshold, the system automatically triggers the heat deformation compensation mode and uses a PID control algorithm to dynamically adjust the welding process parameters: the welding current is calculated according to the following formula.

[0035] in, For the compensated real-time welding current, The initial welding current, This is the current correction factor. This refers to the interpass temperature deviation. When the interpass temperature exceeds the safety threshold, the welding current decreases linearly with the increase of the temperature overshoot, thereby reducing heat accumulation in the molten pool and controlling welding deformation. Welding speed is calculated using the following formula:

[0036] Among them, the real-time welding speed after compensation is... For the initial welding speed, For speed correction factor, This refers to the interpass temperature deviation. When the interpass temperature exceeds the safety threshold, the welding speed increases linearly with the increase of temperature overshoot. Heat accumulation is suppressed by shortening the residence time of the molten pool. The welding torch swing amplitude is calculated using the following formula:

[0037] in, The real-time swing amplitude after compensation. This is the initial swing amplitude. This is the amplitude correction factor. This refers to the interpass temperature deviation. When the interpass temperature exceeds the limit, the weld width loss caused by the reduction in current and the increase in speed is compensated by increasing the oscillation amplitude of the welding torch, thereby maintaining the stability of the weld geometry.

[0038] S5. Based on the field coordinate system, collect the post-weld geometric dimensions and inspection data, calculate the deviation between the actual welding deformation and the predicted welding deformation value, and transmit the corrected thermal deformation coefficient back to the digital twin model to optimize the virtual pre-assembly parameters of subsequent similar units.

[0039] Step S5 is as follows: The actual post-welding geometry of the modular unit is obtained by using three-dimensional laser scanning or total station measurement; The actual post-weld geometry is compared with the theoretical post-weld geometry corresponding to the predicted welding deformation to calculate the actual welding deformation. Accurate detection of the post-weld geometry is the data foundation for verifying the compensation effect and parameter evolution. In this embodiment, a high-precision 3D laser scanner (such as the FARO Focus series) is used to perform a full-domain scan of the welded modular unit to obtain the actual geometric point cloud containing the welding deformation. Using the Iterative Closest Point (ICP) algorithm, the measured point cloud is accurately registered with the theoretical post-weld point cloud (i.e., the geometry corresponding to the predicted welding deformation) generated during the virtual pre-assembly stage. The spatial deviation between the two is calculated, and key deformation indicators such as node displacement and beam deflection are quantitatively extracted to form the actual welding deformation. .

[0040] Based on the difference between the actual welding deformation and the predicted welding deformation, the initial thermal deformation coefficient is corrected using a parameter identification algorithm to generate the corrected thermal deformation coefficient. The parameter identification algorithm is the core of the self-evolution of the thermal deformation coefficient. The system employs optimization algorithms such as least squares or Kalman filtering to calculate the actual welding deformation. Compared with the predicted value of welding deformation deviation Based on this deviation, revisions are made using the following formula:

[0041] in, The initial thermal deformation coefficient, λ is the learning rate (0 < λ < 1). This is the corrected thermal deformation coefficient; the magnitude of each update is controlled to avoid drastic fluctuations in the coefficient caused by a single abnormal data point. Corrected thermal deformation coefficient. It more closely resembles the true thermophysical properties of materials.

[0042] The corrected thermal deformation coefficient is updated in the digital twin model and used as the initial thermal deformation coefficient for subsequent similar modular units to optimize the thermal deformation prediction accuracy in the virtual pre-assembly environment. Specifically, the corrected thermal deformation coefficient is transmitted in real-time to the factory-side digital twin database via a 5G wireless private network or industrial Ethernet to update the thermal deformation coefficient library for the corresponding material grade and plate thickness combination. When subsequent similar modular units undergo virtual pre-assembly, the system automatically calls upon the optimized thermal deformation coefficient. By replacing the initial value, the welding thermal deformation prediction compensation matrix is ​​recalculated to achieve continuous iterative optimization of the virtual pre-assembly accuracy, forming a measurement-correction-reuse model.

[0043] Reference Figure 2 As shown, the modular welding assembly system for steel structure beams and columns includes the following modules: The digital twin construction module is used to divide the steel structure frame into independently hoistable beam and column modular units based on the original design data, and to perform three-dimensional laser scanning on each unit after factory prefabrication to construct an actual geometric digital twin model including manufacturing errors and initial thermal deformation coefficients. The modular units are embedded with digital anchor point identifiers. The virtual pre-assembly module is used to establish a virtual pre-assembly environment based on the digital twin model, perform multi-body assembly interference detection, calculate the node gap, connecting plate misalignment and bevel geometric parameter deviation caused by the accumulation of manufacturing errors, and generate virtual pre-assembly deviation data. The error compensation decision module is used to execute a hierarchical error compensation algorithm based on the deviation data and the initial thermal deformation coefficient, calculate the manufacturing geometric error compensation matrix, the assembly posture error compensation matrix and the welding thermal deformation prediction compensation matrix, generate the welding deformation prediction value and the comprehensive welding trajectory correction amount, and adjust the welding process parameters based on the correction amount. The on-site welding execution module is used to transport the modular unit to the site, establish a mapping relationship between the site coordinate system and the factory virtual coordinate system through the digital anchor point identifier, control the welding equipment to perform welding based on the welding trajectory correction amount and the adjusted welding process parameters, and collect the interpass temperature in real time and trigger the thermal deformation compensation mode to dynamically adjust the welding speed when the temperature exceeds the limit. The quality feedback optimization module is used to collect post-weld geometric dimensions and inspection data based on the field coordinate system, calculate the deviation between the actual welding deformation and the predicted welding deformation value, and send the corrected thermal deformation coefficient back to the digital twin construction module to optimize the virtual pre-assembly parameters of subsequent similar units.

[0044] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A modular welding assembly method for steel structure beams and columns, characterized in that, Includes the following steps: S1. Based on the original design data, the steel structure frame is divided into modular beam and column units that can be independently hoisted. After prefabrication in the factory, each unit is subjected to three-dimensional laser scanning to construct an actual geometric digital twin model that includes manufacturing errors, and digital anchor point markers are embedded and initial thermal deformation coefficients are set. S2. Based on the digital twin model, establish a virtual pre-assembly environment, perform multi-body assembly interference detection, calculate the node gap, connecting plate misalignment and bevel geometric parameter deviation caused by the accumulation of manufacturing errors, and generate virtual pre-assembly deviation data. S3. Based on the deviation data and the initial thermal deformation coefficient, execute the hierarchical error compensation algorithm to calculate the manufacturing geometric error compensation matrix, the assembly posture error compensation matrix and the welding thermal deformation prediction compensation matrix, generate the welding deformation prediction value and the comprehensive welding trajectory correction amount, and adjust the welding process parameters based on the correction amount. S4. Transport the modular unit to the site, establish the mapping relationship between the site coordinate system and the factory virtual coordinate system through the implanted digital anchor point markers, control the welding equipment to perform welding based on the welding trajectory correction amount and the adjusted welding process parameters, collect the interpass temperature in real time, and trigger the thermal deformation compensation mode to dynamically adjust the welding speed when the temperature exceeds the limit. S5. Based on the field coordinate system, collect the post-weld geometric dimensions and inspection data, calculate the deviation between the actual welding deformation and the predicted welding deformation value, and transmit the corrected thermal deformation coefficient back to the digital twin model to optimize the virtual pre-assembly parameters of subsequent similar units.

2. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S1, the step of constructing the actual geometric digital twin model including manufacturing errors is as follows: A laser tracker or photogrammetry system is used to perform a full-domain 3D scan of the modular unit to obtain the original point cloud data; The original point cloud data is processed for denoising, simplification, and coordinate unification. The processed measured point cloud data is registered and compared with the theoretical geometric model corresponding to the original design data using the ICP algorithm, and the manufacturing deviation of each feature point is calculated. Based on the measured point cloud, the actual surface geometry is reconstructed to generate an actual geometric digital twin model carrying the manufacturing deviation as a geometric attribute field.

3. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S1, the implantation of digital anchor identifiers is performed as follows: Countersunk holes are made at the corners of the beam-column joint connection plates of the modular unit, and high-temperature resistant ceramic substrates are implanted therein; The substrate surface is etched with a machine vision-recognizable QR code pattern, and an RFID chip storing a unique number is encapsulated inside. The QR code is associated with and bound to the RFID chip number to serve as the geometric origin marker of the actual geometric digital twin model.

4. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S2, the step of generating virtual pre-assembly deviation data is as follows: Each of the actual geometric digital twin models is imported into the virtual pre-assembly environment according to the assembly sequence determined during the division. Assembly constraints between each of the modular units are set, including node plate contact constraints, weld bevel fit constraints, and bolt hole alignment constraints. Based on the assembly constraint relationship, multi-body assembly interference detection is performed to identify the geometric interference region between the digital twin models, calculate the spatial distance information of the geometric interference region, extract the node gap, connecting plate misalignment and bevel geometric parameter deviation, and generate virtual pre-assembly deviation data containing deviation type, deviation amount and deviation location.

5. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S3, the step of calculating the manufacturing geometric error compensation matrix is ​​as follows: Extract the geometric feature points of each connecting plate of the modular unit from the virtual pre-assembly deviation data, and establish the local coordinate system of the actual geometric digital twin model; The manufacturing geometric error compensation matrix is ​​determined by rigid body transformation calculation of the global coordinate system of the theoretical geometric model corresponding to the local coordinate system and the original design data. The spatial coordinates of the welding start point are corrected based on the manufacturing geometric error compensation matrix to form the position reference component of the comprehensive welding trajectory correction amount.

6. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S3, the steps for calculating the assembly pose error compensation matrix are as follows: Based on the virtual pre-assembly deviation data, the six-degree-of-freedom pose deviation of the modular unit during the assembly process is determined. The six-degree-of-freedom pose deviation includes translational deviation along the X, Y, and Z axes and rotational deviation around the three axes. The six-degree-of-freedom pose deviations are adjusted based on the multibody constraint solution algorithm so that the node gaps and connecting plate misalignments meet the tolerances required by the assembly process, and the assembly pose error compensation matrix is ​​determined. The welding posture angle is corrected based on the assembly posture error compensation matrix to form the posture component of the comprehensive welding trajectory correction amount.

7. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S3, the steps for predicting and compensating for welding thermal deformation are as follows: Calculate the welding heat input based on the welding current, voltage, and speed in the welding process parameters. Based on the initial thermal deformation coefficient and welding heat input, the welding temperature field and stress-strain field of the modular unit are predicted by the thermo-elastic-plastic finite element model. The welding deformation is predicted based on the welding temperature field and stress-strain field, and the welding thermal deformation prediction compensation matrix is ​​determined. The deformation offset of the welding path is pre-compensated based on the welding thermal deformation prediction compensation matrix to form the dynamic compensation component of the comprehensive welding trajectory correction.

8. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, In step S4, establishing the mapping relationship between the on-site coordinate system and the factory virtual coordinate system through the implanted digital anchor point identifiers includes: The machine vision system identifies the QR code pattern on the surface of the digital anchor point marker to obtain the on-site spatial coordinates of the modular unit. Based on the registration between the on-site spatial coordinates and the geometric origin markers in the actual geometric digital twin model, a mapping relationship between the on-site coordinate system and the factory virtual coordinate system is established. The thermal deformation compensation mode includes: when the interpass temperature exceeds the interpass temperature control threshold in the welding process parameters, dynamically adjusting the welding current, welding speed and welding torch swing based on the difference between the interpass temperature and the interpass temperature control threshold, so as to control the heat input and maintain the weld formation quality.

9. The modular welding assembly method for steel structure beams and columns according to claim 1, characterized in that, Step S5 is as follows: The actual post-welding geometry of the modular unit is obtained by using three-dimensional laser scanning or total station measurement; The actual post-weld geometry is compared with the theoretical post-weld geometry corresponding to the predicted welding deformation value to calculate the actual welding deformation. Based on the difference between the actual welding deformation and the predicted welding deformation, the initial thermal deformation coefficient is corrected by the parameter identification algorithm to generate the corrected thermal deformation coefficient. The corrected thermal deformation coefficient is updated to the digital twin model and used as the initial thermal deformation coefficient for subsequent similar modular units to optimize the accuracy of thermal deformation prediction in the virtual pre-assembled environment.

10. A modular welding assembly system for steel structure beams and columns, applied to the modular welding assembly method for steel structure beams and columns as described in any one of claims 1-9, characterized in that, Includes the following modules: The digital twin construction module is used to divide the steel structure frame into independently hoistable beam and column modular units based on the original design data, and to perform three-dimensional laser scanning on each unit after factory prefabrication to construct an actual geometric digital twin model including manufacturing errors and initial thermal deformation coefficients. The modular units are embedded with digital anchor point identifiers. The virtual pre-assembly module is used to establish a virtual pre-assembly environment based on the digital twin model, perform multi-body assembly interference detection, calculate the node gap, connecting plate misalignment and bevel geometric parameter deviation caused by the accumulation of manufacturing errors, and generate virtual pre-assembly deviation data. The error compensation decision module is used to execute a hierarchical error compensation algorithm based on the deviation data and the initial thermal deformation coefficient, calculate the manufacturing geometric error compensation matrix, the assembly posture error compensation matrix and the welding thermal deformation prediction compensation matrix, generate the welding deformation prediction value and the comprehensive welding trajectory correction amount, and adjust the welding process parameters based on the correction amount. The on-site welding execution module is used to transport the modular unit to the site, establish a mapping relationship between the site coordinate system and the factory virtual coordinate system through the digital anchor point identifier, control the welding equipment to perform welding based on the welding trajectory correction amount and the adjusted welding process parameters, and collect the interpass temperature in real time and trigger the thermal deformation compensation mode to dynamically adjust the welding speed when the temperature exceeds the limit. The quality feedback optimization module is used to collect post-weld geometric dimensions and inspection data based on the field coordinate system, calculate the deviation between the actual welding deformation and the predicted welding deformation value, and send the corrected thermal deformation coefficient back to the digital twin construction module to optimize the virtual pre-assembly parameters of subsequent similar units.

Images