A welding seam scanning-free adaptive polishing method and system based on direct driving of welding data

Abstract

The present application provides a kind of based on welding data direct drive's welding seam scanning-free adaptive polishing method and system, it is related to robot automation processing and intelligent manufacturing technical field, including obtaining welding process parameter set and welding robot terminal trajectory;Process parameter set is input into the welding seam appearance prediction model pre-established, deduce welding seam maximum excess height and welding seam width;Based on welding trajectory and welding seam geometric parameter planning horizontal polishing pass and longitudinal polishing layering, generate adaptive polishing path;Control the polishing robot along the path of six-dimensional force sensor's polishing, according to welding line energy pre-judgment welding seam hardness partition and set different target normal contact force, simultaneously based on sensor feedback force signal and torque signal, through closed loop control dynamic adjustment;The present application does not need independent 3D scanning process, realizes welding data direct drive polishing control, adapts to welding seam appearance and hardness change, improves production efficiency and polishing quality consistency.

Description

Technical Field

[0001] This invention relates to the field of robotic automated processing and intelligent manufacturing technology, and in particular to a method and system for adaptive grinding of weld seams without scanning, driven directly by welding data. Background Technology

[0002] In the welding and manufacturing process of metal structural components, to meet requirements such as fatigue strength, corrosion resistance, appearance, and assembly accuracy, the weld reinforcement usually needs to be automatically ground. In existing technologies, automated weld grinding mainly employs the following two modes: One approach is the "scan first, then grind" robotic operation. This method uses 3D vision sensors or laser scanners to independently scan the weld seam, acquiring its 3D point cloud data. Based on this data, a grinding path is planned and the grinding operation is performed. However, this method requires an additional independent scanning process and corresponding scanning equipment, significantly increasing production cycle time and equipment costs. Furthermore, scanning accuracy is easily affected by environmental factors such as arc light, smoke, and reflections, leading to path planning deviations. In addition, the scanning data and welding data are independent of each other, meaning that the process parameters and trajectory information recorded during welding are not effectively utilized, resulting in data redundancy and information waste.

[0003] Another mode is fixed-path grinding based on a pre-programmed procedure. This method generates a fixed grinding trajectory in advance or through offline programming for batches of workpieces, and then repeats the grinding operation. Its drawbacks are: it lacks the ability to adapt to fluctuations in the actual weld morphology (such as changes in reinforcement height and weld width), and when the weld size does not match the taught trajectory, it is easy to under-grind or over-grind; at the same time, it lacks the ability to perceive and adapt to differences in material hardness in the weld area (such as the hardness distribution between the fusion zone and the heat-affected zone in the weld center), and cannot adjust the grinding parameters differently, resulting in inconsistent grinding quality; and it also cannot utilize real-time process data generated during welding, creating an information silo between the welding and grinding processes.

[0004] It is evident that existing technologies generally suffer from the following common technical problems: the rich process parameters and robot trajectory data generated in the welding process, such as welding current, arc voltage, welding speed, and welding torch posture, are not effectively inherited and utilized in the grinding process, resulting in information interruption between the welding and grinding processes; at the same time, existing grinding methods lack comprehensive adaptive capabilities to the geometry and hardness distribution of the weld, making it difficult to achieve high-efficiency, high-quality, and intelligent integrated welding and grinding operations.

[0005] Therefore, how to eliminate redundant independent scanning steps, directly utilize the data generated during the welding process to drive the grinding operation, and intelligently adapt to weld morphology fluctuations and hardness inhomogeneity has become a technical problem that urgently needs to be solved in this field. Therefore, this invention proposes a weld scanning-free adaptive grinding method and system based on welding data to solve the problems existing in the prior art. Summary of the Invention

[0006] To address the aforementioned problems, the present invention aims to propose a weld seam scanning-free adaptive grinding method and system based on direct welding data driving, thereby solving the problems existing in the prior art.

[0007] To achieve the objectives of this invention, the invention is implemented through the following technical solution: a weld seam scanning-free adaptive grinding method based on direct-driven welding data, comprising the following steps: Step 1: Welding Data Acquisition: Acquire the welding process parameter set and welding robot end trajectory recorded in real time or exported offline during the welding process. The welding process parameter set includes welding current, arc voltage, welding speed, welding torch tilt angle, welding torch nozzle height, and welding wire extension length. Step 2: Derivation of weld morphology parameters: Input the obtained welding process parameter set into the pre-established weld morphology prediction model to calculate the key geometric parameters of the weld cross section. The key geometric parameters of the cross section include the maximum weld reinforcement height and the weld weld width. The weld morphology prediction model is a mathematical model established by fitting the historical welding experimental data based on a specific combination of base material and welding material using a multivariate nonlinear regression method. Step 3, Adaptive Grinding Path Planning: Based on the welding robot trajectory obtained in Step 1 and the maximum weld reinforcement height and weld width calculated in Step 2, the number of transverse grinding passes and the longitudinal grinding layers are planned to generate a two-dimensional mesh adaptive grinding path. When planning the number of transverse grinding passes, the number of transverse grinding passes and the offset position of each pass are determined based on the comparison between the weld width and the effective working width of the grinding tool. When planning the longitudinal grinding layers, the number of longitudinal layers is determined based on the maximum weld reinforcement height, the target reinforcement height, and the geometric features of the weld cross-sectional profile. Step 4, Force-Controlled Adaptive Grinding Execution: The grinding robot equipped with a six-dimensional force sensor is controlled to perform grinding operations along the adaptive grinding path generated in Step 3. During the grinding process, the welding line energy is calculated based on the welding process parameters obtained in Step 1. Then, based on the magnitude of the welding line energy, the weld area is pre-judged as a high-hardness weld center fusion zone and a relatively low-hardness heat-affected zone. Different target normal contact forces are set for different areas. At the same time, based on the force and torque signals fed back in real time by the six-dimensional force sensor, the normal displacement, feed speed and grinding spindle speed of the grinding robot are dynamically adjusted through closed-loop control to adapt to the actual hardness changes of the weld area.

[0008] The further improvement lies in the following: In step two, the weld morphology prediction model is characterized in the following way: the maximum weld height is a linear combination of the first-order terms of each parameter in the welding process parameter set, the cross terms between each parameter, and the model error term; the weld width is also an expression of the first-order terms of each parameter in the welding process parameter set, the cross terms between each parameter, and the model error term; and the coefficients of each term in the above combination are determined by multivariate nonlinear regression fitting of historical welding experimental data.

[0009] A further improvement is made in the following way: In step three, the specific method for determining the number of transverse grinding passes and the offset position of each pass based on the comparison result between the weld width and the effective working width of the grinding tool includes: if the weld width is not greater than the effective working width of the grinding tool, then the number of transverse grinding passes is one, and the path of this pass coincides with or is approximately coincident with the trajectory of the welding robot; if the weld width is greater than the effective working width of the grinding tool, then the number of transverse grinding passes is multiple, and a corresponding number of offset paths parallel to the trajectory of the welding robot are generated.

[0010] A further improvement is made in that: when the weld width is greater than the effective working width of the grinding tool, the number of transverse grinding passes is calculated based on the weld width, the effective working width of the grinding tool, and the preset overlap rate between passes, wherein the overlap rate between passes is a preset value that is greater than zero and less than one.

[0011] A further improvement is that the calculation method for the number of transverse grinding passes is as follows: divide the weld width by the product of the difference between the effective working width of the grinding tool and the preset overlap rate, and then round the resulting quotient up.

[0012] A further improvement lies in the following: In step four, the closed-loop control dynamic adjustment includes a force servo control sub-step and a process parameter adaptive adjustment sub-step, wherein: The force servo control sub-step takes the preset target normal contact force as the expected value, the normal force measured in real time by the six-dimensional force sensor as the feedback value, and calculates the normal displacement adjustment of the robot end through the proportional-integral-derivative control algorithm to achieve tracking control of the normal contact force. The adaptive adjustment sub-step of process parameters is to calculate and monitor the rate of change of the estimated value of instantaneous specific grinding energy in real time. The estimated value of instantaneous specific grinding energy is directly proportional to the product of the tangential resultant force and the speed of the grinding spindle, and inversely proportional to the material removal rate. When the rate of change exceeds the preset positive threshold, it is determined that the grinding tool has entered a higher hardness region, triggering the control logic. Under the premise of maintaining force servo control, the grinding feed speed of the grinding robot is reduced in a coordinated manner and the grinding spindle speed is adaptively adjusted.

[0013] A further improvement is made in step four, where the welding line energy is calculated by dividing the product of the welding current, arc voltage, and thermal efficiency coefficient by the welding speed. Then, based on the calculated welding line energy, the weld area is pre-judged as a high-hardness weld center fusion zone and a relatively low-hardness heat-affected zone. A higher target normal contact force is set for the high-hardness weld center fusion zone, and a lower target normal contact force is set for the low-hardness heat-affected zone.

[0014] A weld seam scanning-free adaptive grinding system directly driven by welding data includes: Welding data interface module: including welding robot and welding machine, used to perform welding operations and acquire or output welding process parameter set and welding robot end trajectory in real time; Intelligent analysis and planning module: including an industrial computer, which has a pre-stored weld morphology prediction model and path planning algorithm. It is used to receive the welding process parameter set and trajectory output by the welding data interface module, call the weld morphology prediction model to calculate the maximum weld reinforcement height and weld width, and call the path planning algorithm to generate a two-dimensional meshed adaptive grinding path. The six-dimensional force control execution module includes a grinding robot, a six-dimensional force sensor, a force-controlled grinding device, and grinding tools. The grinding robot receives an adaptive grinding path generated by the intelligent analysis and planning module and moves along the path. The six-dimensional force sensor collects force and torque signals during the grinding process in real time and feeds them back to the industrial computer. The force-controlled grinding device dynamically adjusts the grinding parameters according to the control instructions of the industrial computer. Central control module: including host computer, which communicates with welding data interface module, intelligent analysis and planning module and six-dimensional force control execution module through industrial communication network, and is used to coordinate the operation status and data flow transmission of each module.

[0015] The beneficial effects of this invention are as follows: (1) This invention derives the weld morphology directly from the welding process parameter set through a weld morphology prediction model, eliminating the need for a separate 3D scanning process and achieving seamless data integration between welding and grinding processes. Compared to the traditional "scan first, then grind" mode, it can significantly shorten the production cycle, reduce equipment costs, and improve production efficiency.

[0016] (2) This invention combines prior prediction of weld morphology, hardness feedforward prediction based on welding line energy, and real-time hardness identification based on force and torque signals fed back by a six-dimensional force sensor to form a fully closed-loop intelligent control for grinding. Through force servo control and adaptive adjustment of process parameters, it can accurately adapt to changes in weld geometry and the hardness difference between the fusion zone and heat-affected zone in the weld center, effectively avoiding over- or under-grinding and ensuring grinding quality and consistency.

[0017] (3) The weld morphology prediction model used in this invention is based on multivariate nonlinear regression fitting of historical welding experimental data. The force-controlled adaptive control has clear logic and algorithm, and the system modules are clearly divided. The entire technical solution is specific and reliable, and easy to deploy and implement in digital welding production lines. It realizes the direct driving of welding data to grinding control, which is a key technical feature that distinguishes it from the traditional "scan-driven grinding". Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the weld seam scanning-free adaptive grinding method based on direct welding data driven by the present invention. Figure 2 This is a schematic diagram of the system structure of the single-station weld seam scanning-free adaptive grinding method directly driven by welding data according to the present invention. Figure 3 This is a flowchart illustrating the weld seam scanning-free adaptive grinding method based on direct-driven welding data according to the present invention. Figure 4 This is a schematic diagram of the transverse grinding pass planning of the present invention, where the weld width is not greater than the effective working width of the grinding tool; Figure 5 This is a schematic diagram of the transverse grinding pass planning of the present invention, in which the weld width is greater than the effective working width of the grinding tool; Figure 6 This is a schematic diagram of the longitudinal grinding and layering plan for the weld reinforcement of the present invention.

[0019] Figure 7 This is a schematic diagram of the system structure of the dual-station weld seam scanning-free adaptive grinding method directly driven by welding data according to the present invention. The components include: 1. Host computer; 2. Industrial computer; 3. Welding machine; 4. Welding robot; 5. Grinding robot; 6. Welded parts; 7. First worktable; 8. Second worktable. Detailed Implementation

[0020] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0021] Example 1 according to Figures 1-6 As shown, this embodiment provides a specific implementation process for a weld seam scanning-free adaptive grinding method and system based on direct driving of welding data, which adopts a single-station serial operation.

[0022] The host computer 1 is connected to the industrial computer 2 and the welding robot 4 via industrial Ethernet to monitor the system operation status, set process parameters, and manage data. The industrial computer 2 is connected to the welding robot 4 and the grinding robot 5 to perform welding data analysis, weld morphology prediction, grinding path planning, and grinding control. The welding machine 3 is connected to the welding robot 4 to provide welding current, voltage, and shielding gas. The grinding robot 5 is equipped with grinding tools and a six-dimensional force sensor at its end to perform weld grinding operations.

[0023] Step 1: Welding Data Acquisition The workpiece 6 to be processed is fixedly placed on the first worktable 7, and the position of the weld is determined according to the predetermined welding process requirements. The welding robot 4 performs automatic welding operations on the workpiece 6 under the preset welding trajectory, and the welding machine 3 provides stable energy input such as current and voltage for the welding process. During the welding process, the welding robot 4 and the welding machine 3 generate and record the welding process parameter set and the welding robot end trajectory data in real time. The data is transmitted to the host computer 1 for storage, visualization, or preprocessing, and then transmitted to the industrial computer 2 through the industrial communication interface. The welding process parameter set includes welding current, arc voltage, welding speed, welding torch tilt angle, welding torch nozzle height, and welding wire extension length.

[0024] Step 2: Derivation of weld morphology parameters After receiving the welding data, the industrial computer 2 inputs the welding process parameter set into a pre-established weld morphology prediction model. The weld morphology prediction model is a mathematical model established using a multivariate nonlinear regression method based on historical welding experimental data of a specific base material and welding material combination. In this embodiment, the model is characterized as follows: the maximum weld reinforcement is an expression for the first-order terms of each parameter in the welding process parameter set, the interaction terms between parameters, and the model error term; the weld width is also an expression for the first-order terms of each parameter in the welding process parameter set, the interaction terms between parameters, and the model error term; the coefficients in the above expressions are determined through multivariate nonlinear regression fitting of historical welding experimental data.

[0025] As an example, the functional relationship of the weld morphology prediction model can be characterized by an equation of the following form: in, , Elements for the set of welding process parameters; , , , , , Model coefficients determined by fitting historical experimental data; , This is the model error term.

[0026] The key geometric parameters of the weld, including the maximum weld reinforcement, were calculated using the above model. and weld width w This prediction process requires no additional 3D scanning equipment, thus enabling the direct conversion of welding data into grinding data.

[0027] Step 3: Adaptive Grinding Path Planning Industrial Computer 2 performs adaptive grinding path planning based on the welding robot's end-effector trajectory and predicted weld geometry parameters. Specifically, this includes: Horizontal grinding pass planning: based on weld width w With the effective working width of the grinding tool W t The comparison results determine the number of transverse grinding passes and the offset position of each pass. If the weld width is not greater than the effective working width of the grinding tool, the number of transverse grinding passes is one, and the path of this pass coincides with or approximately coincides with the welding robot trajectory. If the weld width is greater than the effective working width of the grinding tool, the number of transverse grinding passes is multiple, and a corresponding number of offset paths parallel to the welding robot trajectory are generated. When the weld width is greater than the effective working width of the grinding tool, the number of transverse grinding passes is calculated based on the weld width, the effective working width of the grinding tool, and a preset overlap rate between passes. The overlap rate between passes is a preset value that is greater than zero and less than one. Preferably, the number of transverse grinding passes is calculated by dividing the weld width by the product of the difference between the effective working width of the grinding tool and the preset overlap rate between passes, and then rounding the quotient up. The calculation formula is as follows: In the formula, This refers to the number of horizontal grinding passes. The preset inter-track overlap rate (0 < <1).

[0028] Longitudinal grinding layer planning: Based on the maximum excess height of the weld, the target excess height, and the geometric characteristics of the weld cross-section, calculate the total removal volume, and determine the number of longitudinal layers in combination with the maximum grinding depth of a single layer.

[0029] Based on the above planning, a two-dimensional mesh-based adaptive grinding path is generated.

[0030] Step 4: Force-controlled adaptive grinding execution After path planning is completed, industrial computer 2 sends the grinding path and control parameters to grinding robot 5. Grinding robot 5 is equipped with grinding tools and a six-dimensional force sensor at its end, and grinds the weld along the planned path during execution.

[0031] During the polishing execution phase, the system adopts a force control adaptive control strategy based on a combination of prediction and feedback: Hardness prediction and differentiated target force setting: The welding line energy is calculated based on the welding process parameters obtained in step one. In this embodiment, the welding line energy is calculated by dividing the product of the welding current, arc voltage, and thermal efficiency coefficient by the welding speed. Based on the calculated welding line energy, the weld area is predicted to be a high-hardness weld center fusion zone and a relatively low-hardness heat-affected zone. A higher target normal contact force is set for the high-hardness weld center fusion zone, and a lower target normal contact force is set for the low-hardness heat-affected zone.

[0032] Force servo control: Using the preset target normal contact force as the desired value and the normal force measured in real time by the six-dimensional force sensor as the feedback value, the normal displacement adjustment of the robot end effector is calculated through the proportional-integral-derivative (PID) control algorithm to achieve tracking control of the normal contact force.

[0033] Adaptive adjustment of process parameters: Real-time calculation and monitoring of the rate of change of the estimated instantaneous specific grinding energy. The estimated instantaneous specific grinding energy is directly proportional to the product of the tangential resultant force and the grinding spindle speed, and inversely proportional to the material removal rate. When the rate of change exceeds a preset positive threshold, it is determined that the grinding tool has entered a higher hardness region, triggering the control logic. While maintaining force servo control, the grinding feed speed of the grinding robot is reduced in a coordinated manner, and the grinding spindle speed is adaptively adjusted.

[0034] The six-dimensional force sensor collects force and torque signals in real time during the grinding process and feeds them back to the industrial computer 2. Through the closed-loop control mentioned above, the normal displacement, feed speed and spindle speed of the grinding robot 5 are dynamically adjusted to adapt to the changes in material properties in different areas of the weld.

[0035] Through the above implementation method, this embodiment realizes the direct transmission of welding process data to grinding control parameters, eliminating the traditional three-dimensional scanning step before grinding, and forming a closed-loop control process of "welding-prediction-planning-execution-feedback". While improving processing efficiency, it effectively ensures the quality and consistency of weld grinding.

[0036] Example 2 according to Figure 7 As shown, this embodiment improves upon Embodiment 1 by adopting a dual-station structure with separate welding and grinding stations. The welding robot and grinding robot are positioned in different station areas, and the welding and grinding processes are connected through workpiece transfer. This system can also be expanded into a multi-station production line structure.

[0037] The specific implementation process is as follows: Step 1: Welding Data Acquisition At the welding station, the workpiece 6 to be processed is fixed on the first worktable 7. The welding robot 4, in cooperation with the welding machine 3, automatically welds the workpiece 6 according to a preset welding trajectory. During the welding process, the welding robot 4 and the welding machine 3 generate welding process parameter sets and welding robot end-effector trajectory data in real time. This data is transmitted to the host computer 1 for storage, visualization, or preprocessing, and then transmitted to the industrial computer 2 via an industrial communication interface. The welding process parameter set includes welding current, arc voltage, welding speed, welding torch tilt angle, welding torch nozzle height, and welding wire extension length.

[0038] After welding is completed, the weldment is transferred from the welding station to the grinding station. The transfer method can be conveyor line transmission or automated transfer device. The grinding station is equipped with a grinding robot 5 and a corresponding second worktable 8 to support the weldment to be ground.

[0039] Step 2: Derivation of weld morphology parameters After receiving the welding data, the industrial computer 2 calls upon a pre-established weld morphology prediction model to determine the maximum weld reinforcement height. and weld width Calculations are performed. The weld morphology prediction model is the same as in Example 1, which is a mathematical model fitted by a multivariate nonlinear regression method. Its input is a set of welding process parameters, and its output is the maximum weld reinforcement and the weld width. Its functional relationship is the same as in Example 1.

[0040] Step 3: Adaptive Grinding Path Planning Industrial Computer 2 performs adaptive grinding path planning based on the welding robot's end-effector trajectory and predicted weld geometry parameters. Specifically, this includes: Horizontal grinding pass planning: based on weld width w With the effective working width of the grinding tool W t The comparison results determine the number of transverse grinding passes and the offset position of each pass. If the weld width is not greater than the effective working width of the grinding tool, the number of transverse grinding passes is one, and the path of this pass coincides with or approximately coincides with the welding robot trajectory. If the weld width is greater than the effective working width of the grinding tool, the number of transverse grinding passes is multiple, and a corresponding number of offset paths parallel to the welding robot trajectory are generated. When the weld width is greater than the effective working width of the grinding tool, the number of transverse grinding passes is calculated based on the weld width, the effective working width of the grinding tool, and a preset overlap rate between passes, where the overlap rate is a preset value greater than zero and less than one. Preferably, the number of transverse grinding passes is calculated by dividing the weld width by the product of the difference between the effective working width of the grinding tool and the preset overlap rate between passes, and then rounding the quotient up.

[0041] Longitudinal grinding layer planning: Based on the maximum excess height of the weld, the target excess height, and the geometric characteristics of the weld cross-section, calculate the total removal volume, and determine the number of longitudinal layers in combination with the maximum grinding depth of a single layer.

[0042] Based on the above planning, a two-dimensional mesh-based adaptive grinding path is generated.

[0043] Step 4: Force-controlled adaptive grinding execution During the grinding stage, the industrial computer 2 sends the generated grinding path and control parameters to the grinding robot 5. The grinding robot 5 is equipped with grinding tools and a six-dimensional force sensor at its end, and processes the weld along the planned path during the execution process.

[0044] During the polishing process, Industrial Computer 2 implements a force-adaptive control strategy based on a combination of prediction and feedback: Hardness prediction and differentiated target force setting: The welding line energy is calculated based on the welding process parameters obtained in step one. In this embodiment, the welding line energy is calculated by dividing the product of the welding current, arc voltage, and thermal efficiency coefficient by the welding speed. Based on the calculated welding line energy, the weld area is predicted to be a high-hardness weld center fusion zone and a relatively low-hardness heat-affected zone. A higher target normal contact force is set for the high-hardness weld center fusion zone, and a lower target normal contact force is set for the low-hardness heat-affected zone.

[0045] Force servo control: Using a preset target normal contact force as the desired value and the normal force measured in real time by a six-dimensional force sensor as the feedback value, the normal displacement adjustment of the robot end effector is calculated through a proportional-integral-derivative control algorithm to achieve tracking control of the normal contact force.

[0046] Adaptive adjustment of process parameters: Real-time calculation and monitoring of the rate of change of the estimated instantaneous specific grinding energy. The estimated instantaneous specific grinding energy is directly proportional to the product of the tangential resultant force and the grinding spindle speed, and inversely proportional to the material removal rate. When the rate of change exceeds a preset positive threshold, it is determined that the grinding tool has entered a higher hardness region, triggering the control logic. While maintaining force servo control, the grinding feed speed of the grinding robot is reduced in a coordinated manner, and the grinding spindle speed is adaptively adjusted.

[0047] The six-dimensional force sensor collects force and torque signals in real time during the grinding process and feeds them back to the industrial computer 2. Through closed-loop control, the normal displacement, feed speed and spindle speed of the grinding robot 5 are dynamically adjusted to adapt to changes in the material removal process.

[0048] Through the above implementation method, this embodiment achieves the physical separation of welding and grinding stations, enabling welding and grinding to be performed in parallel at different stations, thereby improving production cycle time and making it suitable for automated production lines or industrial scenarios with high cycle time requirements. Simultaneously, by using welding data to drive grinding control, the matching between the grinding path and the actual weld morphology can still be ensured, improving processing quality and consistency.

[0049] 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 illustrative of the principles of the invention. Various changes and modifications can be made to the present invention without departing from its framework and scope of application, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A weld seam scanning-free adaptive grinding method based on direct-driven welding data, characterized in that: Includes the following steps: Step 1: Welding Data Acquisition: Acquire the welding process parameter set and welding robot end trajectory recorded in real time or exported offline during the welding process. The welding process parameter set includes welding current, arc voltage, welding speed, welding torch tilt angle, welding torch nozzle height, and welding wire extension length. Step 2: Derivation of weld morphology parameters: Input the obtained welding process parameter set into the pre-established weld morphology prediction model to calculate the key geometric parameters of the weld cross section. The key geometric parameters of the cross section include the maximum weld reinforcement height and the weld weld width. The weld morphology prediction model is a mathematical model established by fitting the historical welding experimental data based on a specific combination of base material and welding material using a multivariate nonlinear regression method. Step 3, Adaptive Grinding Path Planning: Based on the welding robot trajectory obtained in Step 1 and the maximum weld reinforcement height and weld width calculated in Step 2, the number of transverse grinding passes and the longitudinal grinding layers are planned to generate a two-dimensional mesh adaptive grinding path. When planning the number of transverse grinding passes, the number of transverse grinding passes and the offset position of each pass are determined based on the comparison between the weld width and the effective working width of the grinding tool. When planning the longitudinal grinding layers, the number of longitudinal layers is determined based on the maximum weld reinforcement height, the target reinforcement height, and the geometric features of the weld cross-sectional profile. Step 4, Force-Controlled Adaptive Grinding Execution: The grinding robot equipped with a six-dimensional force sensor is controlled to perform grinding operations along the adaptive grinding path generated in Step 3. During the grinding process, the welding line energy is calculated based on the welding process parameters obtained in Step 1. Then, based on the magnitude of the welding line energy, the weld area is pre-judged as a high-hardness weld center fusion zone and a relatively low-hardness heat-affected zone. Different target normal contact forces are set for different areas. At the same time, based on the force and torque signals fed back in real time by the six-dimensional force sensor, the normal displacement, feed speed and grinding spindle speed of the grinding robot are dynamically adjusted through closed-loop control to adapt to the actual hardness changes of the weld area.

2. The weld seam scanning-free adaptive grinding method based on direct-driven welding data according to claim 1, characterized in that: In step two, the weld morphology prediction model is characterized in the following way: the maximum weld height is a linear combination of the first-order terms of each parameter in the welding process parameter set, the cross terms between each parameter, and the model error term; the weld width is also an expression of the first-order terms of each parameter in the welding process parameter set, the cross terms between each parameter, and the model error term; and the coefficients of each item in the above combination are determined by multivariate nonlinear regression fitting of historical welding experimental data.

3. The weld seam scanning-free adaptive grinding method based on direct-driven welding data according to claim 1, characterized in that: In step three, the specific method for determining the number of transverse grinding passes and the offset position of each pass based on the comparison result between the weld width and the effective working width of the grinding tool includes: if the weld width is not greater than the effective working width of the grinding tool, then the number of transverse grinding passes is one, and the path of this pass coincides with or is approximately coincident with the trajectory of the welding robot; if the weld width is greater than the effective working width of the grinding tool, then the number of transverse grinding passes is multiple, and a corresponding number of offset paths parallel to the trajectory of the welding robot are generated.

4. The weld seam scanning-free adaptive grinding method based on direct-driven welding data according to claim 3, characterized in that: When the weld width is greater than the effective working width of the grinding tool, the number of transverse grinding passes is calculated based on the weld width, the effective working width of the grinding tool, and the preset overlap rate between passes, wherein the overlap rate between passes is a preset value that is greater than zero and less than one.

5. The weld seam scanning-free adaptive grinding method based on direct-driven welding data according to claim 4, characterized in that: The method for calculating the number of transverse grinding passes is as follows: divide the weld width by the product of the difference between the effective working width of the grinding tool and the preset overlap rate, and then round the quotient up.

6. The weld seam scanning-free adaptive grinding method based on direct-driven welding data according to claim 1, characterized in that: In step four, the closed-loop control dynamic adjustment includes a force servo control sub-step and a process parameter adaptive adjustment sub-step, wherein: The force servo control sub-step takes the preset target normal contact force as the expected value, the normal force measured in real time by the six-dimensional force sensor as the feedback value, and calculates the normal displacement adjustment of the robot end through the proportional-integral-derivative control algorithm to achieve tracking control of the normal contact force. The adaptive adjustment sub-step of process parameters is to calculate and monitor the rate of change of the estimated value of instantaneous specific grinding energy in real time. The estimated value of instantaneous specific grinding energy is directly proportional to the product of the tangential resultant force and the speed of the grinding spindle, and inversely proportional to the material removal rate. When the rate of change exceeds the preset positive threshold, it is determined that the grinding tool has entered a higher hardness region, triggering the control logic. Under the premise of maintaining force servo control, the grinding feed speed of the grinding robot is reduced in a coordinated manner and the grinding spindle speed is adaptively adjusted.

7. The weld seam scanning-free adaptive grinding method based on direct-driven welding data according to claim 1, characterized in that: In step four, the welding line energy is calculated by dividing the product of welding current, arc voltage and thermal efficiency coefficient by welding speed. Then, based on the calculated welding line energy, the weld area is pre-judged as a high-hardness weld center fusion zone and a relatively low-hardness heat-affected zone. A higher target normal contact force is set for the high-hardness weld center fusion zone, and a lower target normal contact force is set for the low-hardness heat-affected zone.

8. A weld seam scanning-free adaptive grinding system based on direct-driven welding data, applied to the weld seam scanning-free adaptive grinding method based on direct-driven welding data as described in any one of claims 1-7, characterized in that: include: Welding data interface module: including welding robot and welding machine, used to perform welding operations and acquire or output welding process parameter set and welding robot end trajectory in real time; Intelligent analysis and planning module: including an industrial computer, which has a pre-stored weld morphology prediction model and path planning algorithm. It is used to receive the welding process parameter set and trajectory output by the welding data interface module, call the weld morphology prediction model to calculate the maximum weld reinforcement height and weld width, and call the path planning algorithm to generate a two-dimensional meshed adaptive grinding path. The six-dimensional force control execution module includes a grinding robot, a six-dimensional force sensor, a force-controlled grinding device, and grinding tools. The grinding robot receives an adaptive grinding path generated by the intelligent analysis and planning module and moves along the path. The six-dimensional force sensor collects force and torque signals during the grinding process in real time and feeds them back to the industrial computer. The force-controlled grinding device dynamically adjusts the grinding parameters according to the control instructions of the industrial computer. Central control module: including host computer, which communicates with welding data interface module, intelligent analysis and planning module and six-dimensional force control execution module through industrial communication network, and is used to coordinate the operation status and data flow transmission of each module.

Images