Friction welding process control method and system for tail lamp bracket

Abstract

The invention provides a friction welding process control method and system for a tail lamp support, and relates to the technical field of friction welding, and the friction welding process control method comprises the steps that a multi-mode sensing unit is integrated at the tail end of a welding mechanical arm, comprises a six-dimensional force sensor and a thermal infrared imager, and is connected to a control middle table through an industrial real-time Ethernet protocol; in the beat synchronous reciprocating welding machining process, multi-physics field coupling data low-delay uploading is carried out; welding control offset prediction is carried out, and a six-degree-of-freedom pose offset vector is output; solving the joint angle compensation amount, and correcting the joint space of the welding mechanical arm; performing time sequence accumulation analysis, and outputting a look-ahead intervention compensation parameter; and performing joint space dynamic intervention compensation. The technical problems that in the prior art, a real-time feedback and dynamic compensation mechanism is lacked, welding deviation cannot be effectively corrected to a certain degree, the welding process cannot be efficiently completed, and the production cycle and the production cost are affected are solved.

Description

Technical Field

[0001] This invention relates to the field of friction welding technology, and more specifically to a friction welding process control method and system for taillight brackets. Background Technology

[0002] Friction welding is a solid-state joining technology that uses mechanical force and frictional heat generated by relative motion to cause plastic deformation of the contact surfaces of workpieces and form a connection. In automobile manufacturing, taillight brackets are usually manufactured using friction welding technology. The welding of taillight brackets requires high precision and stability, especially in the welding process of complex shapes and different materials. Any slight deviation may affect the welding quality, and thus affect the strength, durability and appearance of the taillight bracket.

[0003] Traditional friction welding technology typically relies on preset trajectories and fixed welding conditions for welding control, and mainly depends on the hardware control of robotic arms for operation. It lacks real-time feedback mechanisms and intelligent adjustments. During the welding process, with changes in force and dynamic changes in the temperature field, the movement trajectory of the robotic arm may deviate, resulting in uneven weld joints. Due to the lack of real-time feedback and dynamic compensation mechanisms, welding deviations cannot be effectively corrected to a certain extent, or can only be manually corrected after welding. This delayed correction makes the welding process inefficient and prone to joint defects or unqualified workpieces, affecting production cycles and production costs. Summary of the Invention

[0004] This application provides a method and system for controlling the friction welding process of taillight brackets, aiming to solve the technical problems of existing technologies lacking real-time feedback and dynamic compensation mechanisms, which make it impossible to effectively correct welding deviations to a certain extent, or can only be manually corrected after welding, resulting in the welding process not being completed efficiently and affecting the production cycle and production cost.

[0005] The first aspect disclosed in this application provides a method for controlling the friction welding process of a taillight bracket. The method includes: integrating a multimodal sensing unit at the end effector of a welding robotic arm, wherein the multimodal sensing unit includes a six-dimensional force sensor and an infrared thermal imager, and is connected to a control platform via an industrial real-time Ethernet protocol; during the cycle-synchronous reciprocating welding process driven by the welding robotic arm based on a preset welding trajectory, low-latency uploading of multi-physics coupling data is performed through the multimodal sensing unit; the control platform receives and predicts welding control offset based on the multi-physics coupling data, outputting a six-degree-of-freedom pose offset vector; calculating joint angle compensation based on the six-degree-of-freedom pose offset vector to correct the joint space of the welding robotic arm; the control platform accumulates the six-degree-of-freedom pose offset vector and performs time-series cumulative analysis, outputting look-ahead intervention compensation parameters; and using the look-ahead intervention compensation parameters to perform dynamic intervention compensation for the joint space of the welding robotic arm.

[0006] The second aspect of this application discloses a friction welding process control system for a taillight bracket. The system is used in the aforementioned friction welding process control method for a taillight bracket. The system includes: a sensing unit integration module for integrating a multimodal sensing unit at the end of a welding robotic arm, wherein the multimodal sensing unit includes a six-dimensional force sensor and an infrared thermal imager, and is connected to a control platform via an industrial real-time Ethernet protocol; and a data upload module for performing multi-physics coupling data through the multimodal sensing unit during the synchronous reciprocating welding process driven by the welding robotic arm based on a preset welding trajectory. The system includes a low-latency upload module; an offset prediction module for receiving and predicting welding control offsets based on the multiphysics coupling data, and outputting a six-degree-of-freedom pose offset vector; a joint space correction module for calculating joint angle compensation based on the six-degree-of-freedom pose offset vector, and correcting the joint space of the welding robot; an accumulation analysis module for accumulating the six-degree-of-freedom pose offset vector for time-series accumulation analysis, and outputting look-ahead intervention compensation parameters; and an intervention compensation module for using the look-ahead intervention compensation parameters to perform dynamic intervention compensation for the joint space of the welding robot.

[0007] One or more technical solutions provided in this application have at least the following beneficial effects: By integrating a six-dimensional force sensor and an infrared thermal imager at the end effector of the robotic arm, multi-physics parameters such as force and heat can be monitored in real time during the welding process. This data acquisition method can accurately track possible deviations during welding, ensuring that the movement of the robotic arm and the welding quality remain in an ideal state. During welding, based on multi-physics coupled data, pose shifts generated during welding can be analyzed and predicted in real time through low-latency upload. This instant feedback and prediction capability can effectively avoid welding quality fluctuations caused by minor errors of the robotic arm or external disturbances, ensuring that each step in the welding process can be precisely controlled and avoiding welding defects. Based on the six-degree-of-freedom pose shift vector, the joint space of the robotic arm can be dynamically corrected by solving the joint angle compensation, so that the welding robotic arm maintains consistency with the preset trajectory. This dynamic correction can compensate for minor deviations that occur in the robotic arm during welding, thereby improving welding accuracy. Accumulating the six-degree-of-freedom pose shift vector during welding and performing time-series cumulative analysis, outputting look-ahead intervention compensation parameters, and then dynamically intervening to compensate for the joint space of the robotic arm. This technology not only reduces the accumulation of errors during welding but also compensates and adjusts in advance by predicting future deviations, thereby further optimizing the welding trajectory and improving welding accuracy.

[0008] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0009] Figure 1 This is a schematic flowchart of a friction welding process control method for a taillight bracket provided in an embodiment of this application.

[0010] Figure 2 This is a schematic diagram of the friction welding process control system for a taillight bracket provided in an embodiment of this application.

[0011] Figure labeling: Sensing unit integration module 10, data upload module 20, offset prediction module 30, joint space correction module 40, cumulative analysis module 50, intervention compensation module 60. Detailed Implementation

[0012] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0013] Example 1, as Figure 1As shown in the embodiment of this application, a method for controlling the friction welding process of a taillight bracket is provided, the method comprising: A multimodal sensing unit is integrated at the end of a welding robotic arm. The multimodal sensing unit includes a six-dimensional force sensor and an infrared thermal imager, and is connected to the control console via an industrial real-time Ethernet protocol.

[0014] The six-dimensional force sensor is a combination of a three-dimensional force sensor and a three-dimensional torque sensor. It is used to measure the force and torque experienced by the welding robot arm in real time during the welding process. The six-dimensional force sensor can acquire force and torque data of the object in the x, y, and z axes, i.e., the force application and rotation direction. The infrared thermal imager is used to detect temperature changes in the welding area, obtain thermal images, and help determine the temperature distribution and heat-affected zone during the welding process. The industrial real-time Ethernet protocol is used to upload the measurement data of the multimodal sensing units to the control platform in real time. Through real-time data transmission, feedback and control during the welding process are ensured to have the lowest possible latency.

[0015] During the synchronous reciprocating welding process driven by the welding robot arm based on the preset welding trajectory, the multi-physical field coupled data is uploaded with low latency through the multi-modal sensing unit.

[0016] The welding robotic arm performs welding according to a preset welding trajectory, which is a fixed path pre-set based on experience. After completing one welding action within a specified cycle time, the robotic arm precisely moves the workpiece to the next workstation, such as an inspection or assembly station. This step ensures seamless connection of the production line, meaning there is no lag between operations at each workstation. Cycle time synchronization refers to the robotic arm executing welding tasks in time synchronization, ensuring that each welding action is completed within a fixed time to facilitate timely workpiece transfer and guarantee production line efficiency. During the welding process, the multimodal sensing unit collects multiphysics data, including force and temperature data, through a six-dimensional force sensor and an infrared thermal imager. This data forms multiphysics coupling data, which is uploaded to the control center in real time to ensure data timeliness and prepare for the next control step.

[0017] The control platform receives and predicts welding control offset based on the multiphysics coupling data, and outputs a six-degree-of-freedom pose offset vector.

[0018] The control platform receives multi-physics coupled data, which not only provides real-time mechanical changes and thermal field distribution during the welding process, but also incorporates the influence of temperature changes on workpiece deformation. Through comprehensive analysis of this data, the pose offset of the welding robot arm is predicted, generating a six-degree-of-freedom pose offset vector. This six-degree-of-freedom pose offset vector describes the changes in the robot arm's spatial displacement in x, y, and z axes, as well as its rotation around the x, y, and z axes. It can reflect the displacement or rotation of the robot arm due to external disturbances, such as force fluctuations and temperature field changes, providing a basis for subsequent compensation and correction.

[0019] The joint angle compensation is calculated based on the six-degree-of-freedom pose offset vector, and the joint space of the welding robot is corrected.

[0020] The six-degree-of-freedom pose offset vector represents the displacement and rotational offset of the robotic arm in space. To ensure the accuracy of the welding process, the kinematics of the robotic arm needs to be corrected; this correction is called joint space correction. First, the pre-stored kinematic model parameters of the welding robotic arm are called, including information such as the length of the robotic arm and the joint range. Using these parameters, the forward kinematic equations are constructed, that is, the position and orientation of the robotic arm end effector are solved based on the joint variables. Based on the six-degree-of-freedom pose offset vector, this offset is superimposed on the original welding trajectory target pose to obtain the corrected target pose. This corrected target pose is the new target position of the robotic arm, which is the precise position after considering the offset. Next, using the inverse mapping relationship of the forward kinematic equations, the joint angle compensation corresponding to the corrected target pose is calculated through the inverse kinematics algorithm. Inverse kinematics refers to: given the target pose of the end effector, calculating the required angles of each joint of the robotic arm to achieve that pose.

[0021] The control platform accumulates the six-degree-of-freedom pose offset vectors and performs time-series cumulative analysis to output forward-looking intervention compensation parameters.

[0022] During welding, the robotic arm may experience continuous small-range displacements due to various factors such as force fluctuations and temperature changes. To prevent these displacements from affecting subsequent welding quality, proactive compensation is necessary—this is the core of proactive intervention compensation. Specifically, the control platform accumulates and analyzes the six-degree-of-freedom pose displacement vectors at multiple moments, forming a pose displacement time-series database. This database records the changes in displacement over time. Through sliding window analysis, key features are extracted from the time-series data, including the slope of the linear regression of displacement deviation, the standard deviation of attitude angle deviation, and the autocorrelation coefficient of displacement. Simultaneously, temperature change data related to the welding process is analyzed to extract thermodynamic features such as the rate of change of temperature gradient and the rate of maximum temperature rise.

[0023] Based on these extracted welding offset trend features and time-varying temperature trend features, a dynamic model of thermal deformation is established to predict the friction welding offset caused by thermal expansion or temperature changes during the welding process. Finally, the predicted friction welding offset is converted into a feedforward compensation amount in the joint space and output as a look-ahead intervention compensation parameter.

[0024] The joint space dynamic intervention compensation of the welding robot arm is performed using the aforementioned forward-looking intervention compensation parameters.

[0025] The calculated look-ahead intervention compensation parameters are essentially compensation amounts that proactively adjust the welding robot arm's offset at a future point in time. These parameters can effectively compensate for welding deviations caused by mechanical changes, temperature variations, or other external factors. The control console writes the look-ahead intervention compensation parameters into the robot arm's feedforward compensation register to establish a dynamic compensation baseline, which is used for future compensation operations. During the welding interval of the welding robot arm, i.e., the pause between two welding actions, the robot arm uses the look-ahead intervention compensation parameters to dynamically intervene and compensate for joint space. This compensation dynamically adjusts the joint angles of the robot arm to ensure the accuracy of the robot arm's position.

[0026] Furthermore, the control platform accumulates the six-degree-of-freedom pose offset vectors and performs time-series cumulative analysis to output look-ahead intervention compensation parameters, including: By incrementally accumulating the six-degree-of-freedom pose offset vectors, a pose offset time-series database is obtained, wherein the pose offset time-series database is associated with a welding temperature field time-series slice library. A sliding window analysis is performed on the pose offset time-series database to extract the linear regression slope of displacement deviation, the standard deviation of attitude angle deviation, and the autocorrelation coefficient of offset, as welding offset trend features. A sliding window analysis is also performed on the welding temperature field time-series slice library to extract the temperature gradient change rate and the maximum temperature rise rate, as temperature time-varying trend features. Based on the welding offset trend features and the temperature time-varying trend features, dynamic thermal deformation modeling is performed to output the predicted friction welding offset. The predicted friction welding offset is converted into a joint space feedforward compensation amount, and the look-ahead intervention compensation parameters are output.

[0027] The changes in the six-DOF pose offset vectors during the welding process are collected and recorded over time. The pose offset of the robotic arm at each moment is stored to form a temporal database of pose offsets. This data reflects the dynamic behavior of the robotic arm during the welding process, such as offsets caused by mechanical factors and temperature changes. In addition to pose offsets, the temperature field during the welding process, especially the distribution and changes in temperature in the welding zone, also affects the posture and position of the robotic arm. Therefore, temperature field data is also collected synchronously to form a welding temperature field temporal slice library, which contains temperature data at different time points during the welding process. With each welding operation, new pose offset data and temperature field data are incrementally accumulated in the database to ensure that the data is continuously updated over time for subsequent analysis and prediction.

[0028] Sliding window analysis is a time series analysis method that analyzes the trend of data changes within a fixed-size time window. As time progresses, the window slides forward, gradually covering new data points. Sliding window analysis on a pose offset time series database extracted three main welding offset trend features: the slope of the linear regression of displacement deviation, which indicates the trend of displacement deviation over time and reflects whether the displacement deviation has a continuous increasing or decreasing trend; the standard deviation of attitude angle deviation, which measures the stability of the robotic arm's attitude; a larger standard deviation means greater fluctuations in attitude angle deviation and a higher potential for attitude instability during welding; and the autocorrelation coefficient of offset, which measures the correlation between the current offset and historical data. A high autocorrelation coefficient indicates that the offsets occurring during welding have a certain periodicity or regularity.

[0029] During welding, the temperature field has a significant impact on welding offset. As welding progresses, the temperature in the welding zone changes, which leads to thermal deformation. To understand the impact of temperature changes on welding offset, a sliding window analysis was performed on the welding temperature field time-series slice library, extracting two main time-varying temperature trend features: the temperature gradient change rate, which indicates the speed of temperature change in different areas during welding; areas with large temperature gradients may lead to local thermal stress, thereby causing pose offset; and the maximum temperature rise rate, which refers to the rate at which the temperature rises during welding; a high temperature rise rate may lead to uneven thermal expansion, thereby causing the robotic arm to offset.

[0030] By combining the welding offset trend characteristics and the time-varying temperature trend characteristics, a dynamic model of thermal deformation is established. This model simulates the pose offset caused by the combined action of thermal stress and mechanical factors during the welding process, and outputs the predicted friction welding offset, reflecting the change in the pose of the robotic arm.

[0031] The predicted friction welding offset is converted into a joint space feedforward compensation value. This value, based on the prediction of the welding process, compensates for the robotic arm in advance by adjusting the joint angles to eliminate or reduce the predicted offset. Finally, the obtained joint space feedforward compensation value is output as a look-ahead intervention compensation parameter. This parameter is used for subsequent dynamic compensation, meaning the system can make adjustments in advance to ensure the robotic arm maintains a precise position and orientation throughout the welding process.

[0032] Furthermore, based on the welding offset trend characteristics and temperature time-varying trend characteristics, dynamic modeling of thermal deformation is performed, and the predicted friction welding offset is output, including: By integrating the welding offset trend features and the temperature time-varying trend features, a thermo-coupling feature vector is constructed; a temperature deformation transfer function is constructed based on the thermal expansion characteristics of the taillight bracket material; the thermo-coupling feature vector is processed using the temperature deformation transfer function to output the thermal deformation pose offset component; after interactively obtaining the elastic deformation compensation amount of the robotic arm, it is superimposed on the thermal deformation pose offset component to output the predicted friction welding offset amount.

[0033] By combining the welding offset trend characteristics and the temperature time-varying trend characteristics, a composite feature set is formed, called the thermo-coupling feature vector. This vector is a comprehensive representation of the influence on the welding process and can more comprehensively describe the possible causes of welding offset.

[0034] Every material undergoes thermal expansion when heated, and this expansion follows a regular pattern with temperature changes. The taillight bracket, as a specific welding object, is made of materials such as metals or alloys, and exhibits certain thermal expansion characteristics upon heating. Based on the thermal expansion characteristics of the taillight bracket material, a temperature deformation transfer function is constructed. This function converts temperature changes into deformation, i.e., the material's expansion behavior at different temperatures. The temperature deformation transfer function describes the relationship between the material's dimensional changes and temperature changes based on the material's coefficient of thermal expansion, mapping the temperature component of the thermo-mechanical coupling eigenvector to predict the impact of temperature changes on the deformation of the taillight bracket during the welding process.

[0035] The constructed temperature deformation transfer function is applied to the temperature component of the thermo-coupling eigenvector. Through the calculation of the function, the temperature change is directly converted into the pose offset, reflecting the change in the robot arm pose caused by the thermal expansion of the material during the welding process. Specifically, the expansion and contraction caused by temperature will cause the geometry of the taillight bracket to change, which in turn leads to the offset of the robot arm pose. Through the action of the temperature deformation transfer function, the thermal deformation pose offset component can be obtained, that is, the pose offset caused by the deformation of the taillight bracket due to temperature change.

[0036] Besides deformation caused by thermal expansion, the robotic arm itself may also be subjected to forces during the welding process, leading to elastic deformation. This elastic deformation also affects the precision of the welding process. Therefore, it is necessary to consider the elastic deformation of the robotic arm itself and compensate for it accordingly. Based on real-time data from the robotic arm, the elastic deformation compensation amount is calculated, which is the deformation caused by mechanical forces during the robotic arm's movement. After obtaining the thermal deformation pose offset component and the elastic deformation compensation amount, these two quantities are superimposed, that is, the offset caused by thermal expansion and the offset caused by the elastic deformation of the robotic arm are combined to obtain a comprehensive predicted friction welding offset.

[0037] Furthermore, the joint space dynamic intervention compensation of the welding robot arm using the aforementioned forward-looking intervention compensation parameters includes: The forward-looking intervention compensation parameters are written into the feedforward compensation register of the welding robot to establish a dynamic compensation baseline; during the welding interval of the welding robot, the forward-looking intervention compensation parameters are used to perform spatial dynamic intervention compensation; the forward-looking intervention compensation parameters are iteratively updated based on the pose residual after compensation.

[0038] The feedforward compensation register is a storage space specifically used to store calculated compensation parameters. These parameters are dynamically adjusted based on previous welding trajectories, predicted welding offsets, and other relevant data. After writing the obtained look-forward intervention compensation parameters into the feedforward compensation register, the robotic arm establishes a dynamic compensation baseline. This dynamic compensation baseline represents an ideal path or target position after compensation, serving as a reference line to guide the robotic arm in correcting its motion path. Through this dynamic compensation baseline, the robotic arm can precisely adjust its behavior according to the compensation parameters to compensate for deviations caused by various factors during the welding process.

[0039] The welding process is divided into multiple stages, including a welding interval—a short break after each welding action. During this interval, the robotic arm has not yet performed a new welding task, allowing for dynamic intervention and compensation to ensure accuracy adjustments for subsequent welding tasks. During this interval, based on calculated look-ahead intervention and compensation parameters, the position and posture of the welding robotic arm are dynamically adjusted. This means proactively adjusting the robotic arm's position to ensure it accurately completes its tasks along a pre-set trajectory in the following work cycles.

[0040] After intervention and compensation, the robotic arm's pose may still have a slight difference from the target position. This difference is called the post-compensation pose residual. The post-compensation pose residual originates from the robotic arm's dynamic response and non-ideal factors during the motion process, such as inertia and friction. The look-ahead intervention and compensation parameters are iteratively updated using the post-compensation pose residual. Specifically, the current residual is analyzed and compared with historical data to identify possible systematic error patterns or trends. Then, an adaptive algorithm is used to fine-tune the look-ahead intervention and compensation parameters, ensuring that the compensation process is continuously optimized and the residual is reduced. This iterative update is a closed-loop feedback process, ensuring that the welding robotic arm's compensation algorithm becomes increasingly accurate over time, continuously improving the precision of the welding process during long-term use.

[0041] Furthermore, the control platform receives and performs welding control offset prediction based on the multiphysics coupling data, and outputs a six-degree-of-freedom pose offset vector, including: Spatially synchronize the force sensing time-series data and temperature field time-series data in the multiphysics coupling data; acquire six-degree-of-freedom force fluctuations from the force sensing time-series data to obtain six-degree-of-freedom force fluctuation components; calculate the extreme values ​​of the temperature gradient and the temperature of the welding zone based on the temperature field time-series data; concatenate the six-degree-of-freedom force fluctuation components, the extreme values ​​of the temperature gradient, and the extreme values ​​of the temperature of the welding zone to construct a thermo-mechanical coupling feature vector; perform ultra-short-term nonlinear state estimation based on the thermo-mechanical coupling feature vector to output the six-degree-of-freedom pose offset vector.

[0042] In multiphysics coupling data, force sensing time series data and temperature field time series data are spatially separate and need to be spatially synchronized. That is, data from different sources are mapped to the same reference coordinate system so that they can be effectively compared and analyzed.

[0043] By collecting time-series force sensing data, the six-degree-of-freedom force fluctuations at different time points are obtained, reflecting the changes in force during the welding process. This reveals the changing trends and patterns of the physical quantities measured by the force sensor. The resulting six-degree-of-freedom force fluctuation components are used for subsequent analysis to help evaluate the impact of force on the robotic arm's motion during the welding process.

[0044] Temperature gradient refers to the rate of temperature change with space, especially in the welding zone. This change affects the thermal expansion, melting, and solidification processes of the material. Based on time-series temperature field data, the extreme values ​​of the temperature gradient in the welding zone are calculated, i.e., the highest and lowest temperature change rates during the welding process. These extreme values ​​can reveal the thermal behavior patterns of the welding zone, providing a basis for subsequent thermal deformation prediction. Besides the temperature gradient, the extreme temperatures in the welding zone, i.e., the highest and lowest temperatures, directly affect the material deformation, stress distribution, and the quality of the weld joint. By analyzing time-series temperature data, the extreme temperatures during the welding process are identified, providing necessary temperature data for the construction of subsequent thermal deformation models.

[0045] By combining the six-degree-of-freedom force fluctuation components, the extreme values ​​of the temperature gradient, and the extreme values ​​of the temperature in the welding zone, a thermo-mechanical coupling feature vector is formed. This thermo-mechanical coupling feature vector contains comprehensive information on the changes in force and temperature during the welding process, and can simultaneously reflect the changes in force and temperature, which can be used to analyze and model the thermo-mechanical behavior of the welding process.

[0046] Based on the constructed thermo-coupling eigenvectors, the six-degree-of-freedom pose offset vector of the robotic arm during welding is predicted through ultra-short-term nonlinear state estimation. This means that these coupled data are used to predict the position and orientation shifts of the robotic arm in each degree of freedom during welding. Since the temperature and force changes during welding have strong nonlinear characteristics, nonlinear methods are required for modeling and prediction. Ultra-short-term means not only estimating the current welding state but also predicting pose shifts occurring in the short term, enabling rapid response and appropriate corrections and compensations during welding. The final output six-degree-of-freedom pose offset vector represents the real-time changes in the position and orientation of the robotic arm during welding.

[0047] Furthermore, based on the aforementioned thermo-coupling eigenvector, ultra-short-term nonlinear state estimation is performed, and the six-degree-of-freedom pose offset vector is output, including: Based on the time-varying offset characteristics of the force sensing time series data, a time-varying mechanical state vector sequence is constructed; the thermo-coupling feature vector is mapped to a six-degree-of-freedom virtual observation; the pose offset state is recursively deduced based on the time-varying mechanical state vector sequence, and the predicted pose offset is output; the six-degree-of-freedom virtual observation and the predicted pose offset are fused by Kalman gain weighting to output the six-degree-of-freedom pose offset vector.

[0048] Force sensor data during the welding process changes over time, reflecting fluctuations in the force between the robotic arm and the workpiece. To effectively model these fluctuations, time-varying offset features are extracted, which are the characteristics of force change trends, amplitudes, and frequencies over time. Based on the extracted time-varying offset features, a time-varying mechanical state vector sequence is constructed. This sequence describes the force's behavior at different time points, including not only the magnitude of the force but also its direction and fluctuation amplitude, reflecting the dynamic changes in force.

[0049] The thermo-coupling feature vectors are transformed into six-degree-of-freedom virtual observations through a mapping algorithm. This means that the features extracted from force and temperature data are converted into values ​​that can be used to describe the changes in position and attitude of the robotic arm in three-dimensional space, namely displacement and attitude angle. The six-degree-of-freedom virtual observations simulate the actual motion state of the robotic arm, but these states are inferred through indirect temperature and force data.

[0050] Based on the constructed time-varying mechanical state vector sequence, a recursive algorithm is used to predict the position and attitude deviation of the robotic arm during welding. Recursion refers to continuously updating and predicting future states using historical data, such as past force sensor data and state vectors. In this process, the recursive algorithm infers the pose deviation of the robotic arm based on the time-varying mechanical state vector sequence, that is, how much the robotic arm will deviate during welding within a given time period.

[0051] Kalman filtering, a highly efficient recursive estimation algorithm, is used to fuse six-DOF virtual observations and predicted pose offsets. This algorithm combines information from multiple sources to output the optimal estimate. Kalman gain is used to adjust the weights of different data sources during the fusion process based on their uncertainties, resulting in a more accurate final result. The weighted fusion of these two data points using Kalman gain yields the six-DOF pose offset vector, which is used for subsequent dynamic compensation to ensure the accuracy of the robotic arm's movements during the welding process.

[0052] Furthermore, based on the time-varying offset characteristics of the force sensing time-series data, a time-varying mechanical state vector sequence is constructed, including: The force sensing time series data is decomposed to obtain a six-degree-of-freedom time series fluctuation component; the six-degree-of-freedom time series force deviation is output by comparing the local six-degree-of-freedom reference value with the six-degree-of-freedom time series fluctuation component; the time-varying mechanical state vector sequence is constructed based on the six-degree-of-freedom time series force deviation.

[0053] To understand the patterns of force change over time, the force sensing time-series data is decomposed to extract six degrees of freedom force fluctuation components. Each fluctuation component contains the trend, amplitude, and periodic fluctuation of the force over time in that degree of freedom, which can describe the dynamic characteristics of the force in detail.

[0054] The local six-degree-of-freedom (6DOF) baseline is a reference standard obtained through prior experience, calibration, or modeling. The baseline represents the reference value for the forces in each degree of freedom under ideal conditions during the welding process. Comparing the 6DOF temporal fluctuation components with the local 6DOF baseline aims to examine the difference between the actual forces and the ideal baseline, thereby deriving the 6DOF temporal force deviation. The 6DOF temporal force deviation is a time-series deviation data point representing the degree of deviation of the force variation in each degree of freedom of the robotic arm from the baseline value at each moment.

[0055] By analyzing the time-varying force deviation of the six degrees of freedom, a time-varying mechanical state vector sequence is constructed. This vector sequence can describe the dynamic change state of the force. The force deviation information at each moment constitutes a state vector, and this state vector changes continuously with time, thus forming a time-varying sequence. This sequence can reflect the mechanical behavior, changing trend and its influence on the movement of the robotic arm throughout the welding process.

[0056] Furthermore, based on the joint angle compensation calculated from the six-degree-of-freedom pose offset vector, the joint space correction of the welding robot arm is performed, including: The pre-stored kinematic model parameters of the welding robot arm are called to construct the forward kinematic equations; after the target pose of the welding trajectory is retrieved locally, the corrected target pose is generated by superimposing the six-degree-of-freedom pose offset vector onto the target pose of the welding trajectory; based on the inverse mapping relationship of the forward kinematic equations, the joint angle compensation amount of the corrected target pose is solved by the analytical inverse kinematic algorithm.

[0057] The kinematic model parameters of the welding robot arm are obtained through prior measurements or calibrations, including the length, rotation angle, and inertial parameters of each joint. The forward kinematic equations are used to solve for the position and orientation of the robot arm's end effector based on the angles of each joint. Using the robot arm's kinematic model, the forward kinematic equations can calculate the Cartesian space (i.e., the position and orientation of the end effector) from the joint space. This equation is achieved through matrix transformations based on the rotation and displacement parameters of each joint.

[0058] During welding, the robotic arm needs to follow a preset trajectory, and each trajectory point has a target pose. This target pose is generated through pre-planned trajectory control and represents the position and orientation the robotic arm needs to achieve during welding. During welding, due to various factors, the robotic arm may experience pose deviation, preventing it from accurately reaching the predetermined trajectory point. Therefore, the target pose is adjusted based on the six-degree-of-freedom pose offset vector. By superimposing the six-degree-of-freedom pose offset vector onto the original target pose, a corrected target pose is obtained, ensuring the robotic arm can operate according to the actual situation.

[0059] In robotic arm kinematics, forward kinematics calculates the end effector pose based on given joint angles, while inverse kinematics calculates the required joint angles based on the target pose. After generating a corrected target pose in the previous step, it's necessary to calculate how the angles of each joint of the robotic arm should change to achieve this corrected target pose—that is, the joint angle compensation. The inverse mapping relationship refers to working backward from the corrected target pose of the end effector back to the angle values ​​of each joint. In this process, the inverse kinematics algorithm solves for the joint angles corresponding to the corrected target pose. Based on the corrected target pose, the inverse kinematics equations output one or more joint angle solutions, which represent the angles to which each joint of the robotic arm should rotate to reach the target pose.

[0060] Furthermore, based on the inverse mapping relationship of the forward kinematics equations, the joint angle compensation amount of the corrected target pose is solved by an analytical inverse kinematics algorithm, including: Based on the inverse mapping relationship of the forward kinematic equation, multiple candidate solutions for the corrected target pose are solved; the target joint angle solution is obtained by filtering from the multiple candidate solutions with the minimum joint displacement as a constraint; the Jacobian matrix rank detection is performed based on the target joint angle solution to avoid singular configurations and then the joint angle compensation amount is output.

[0061] Based on the corrected target pose, the corresponding joint angles are calculated through inverse kinematics. Since inverse kinematics often has multiple solutions, i.e., the same end pose corresponds to multiple different joint angle solutions, this step will generate multiple sets of candidate joint angle solutions. These solutions may achieve the same target pose under different joint configurations.

[0062] In practical applications, it is desirable for the robotic arm to move as smoothly and efficiently as possible. Therefore, it is necessary to select the minimum joint displacement as the compensation solution. This avoids large and unnecessary adjustments, reduces the energy consumption of the robotic arm, and minimizes wear and tear on the equipment. Among multiple candidate solutions for joint angles, the optimal solution is determined by the minimum joint displacement. Specifically, the difference between each candidate solution and the current joint angle is calculated, and the solution with the smallest difference is selected. Through this constraint, the solution with the minimum adjustment under the current joint configuration is obtained as the target joint angle solution.

[0063] The Jacobian matrix is ​​a mathematical tool used to describe the relationship between changes in the position of a robot's end effector and changes in joint angles. It is a linear mapping matrix used to describe the mapping from joint space to task space. In inverse kinematics, the Jacobian matrix is ​​used to analyze the impact of small changes on pose. If the rank of the Jacobian matrix is ​​low, i.e., close to a singular value, it indicates that some degrees of freedom of the robot arm will be lost under that configuration, which may lead to invalid motion or the robot arm failing to operate smoothly.

[0064] During robot operation, certain joint configurations can lead to singular configurations, i.e., the loss of some degrees of freedom in the robotic arm. In such cases, even small joint changes can cause drastic changes in the pose of the end effector, or even prevent the control system from continuing to execute motion. Jacobian matrix rank detection can determine whether the current joint configuration exhibits singularity. If singularity exists, the robotic arm will be unable to perform effective motion control. In this case, these singular configurations are avoided by selecting a non-singular joint angle solution to prevent errors during motion. After confirming that the target joint angle solution is valid and does not exhibit singular configurations, the final joint angle compensation amount is output, indicating how much adjustment the robotic arm needs to make to achieve the corrected target pose.

[0065] Example 2, based on the same inventive concept as the friction welding process control method for taillight brackets in the foregoing examples, such as... Figure 2 As shown, this application embodiment provides a friction welding process control system for a taillight bracket, the system comprising: The sensing unit integration module 10 is used to integrate a multimodal sensing unit at the end of the welding robot arm. The multimodal sensing unit includes a six-dimensional force sensor and an infrared thermal imager, and is connected to the control platform via an industrial real-time Ethernet protocol. The data upload module 20 is used to upload multi-physics field coupled data with low latency through the multimodal sensing unit during the synchronous reciprocating welding process driven by the welding robot arm based on a preset welding trajectory. The offset prediction module 30 is used by the control platform to receive and predict welding control offset based on the multi-physics field coupled data, and output a six-degree-of-freedom pose offset vector. The joint space correction module 40 is used to solve for joint angle compensation based on the six-degree-of-freedom pose offset vector, and perform joint space correction for the welding robot arm. The cumulative analysis module 50 is used by the control platform to accumulate the six-degree-of-freedom pose offset vector for time-series cumulative analysis, and output look-ahead intervention compensation parameters. The intervention compensation module 60 is used to perform dynamic intervention compensation for the joint space of the welding robot arm using the look-ahead intervention compensation parameters.

[0066] Furthermore, the cumulative analysis module 50 is used to perform the following operation steps: By incrementally accumulating the six-degree-of-freedom pose offset vectors, a pose offset time-series database is obtained, wherein the pose offset time-series database is associated with a welding temperature field time-series slice library. A sliding window analysis is performed on the pose offset time-series database to extract the linear regression slope of displacement deviation, the standard deviation of attitude angle deviation, and the autocorrelation coefficient of offset, as welding offset trend features. A sliding window analysis is also performed on the welding temperature field time-series slice library to extract the temperature gradient change rate and the maximum temperature rise rate, as temperature time-varying trend features. Based on the welding offset trend features and the temperature time-varying trend features, dynamic thermal deformation modeling is performed to output the predicted friction welding offset. The predicted friction welding offset is converted into a joint space feedforward compensation amount, and the look-ahead intervention compensation parameters are output.

[0067] Furthermore, the cumulative analysis module 50 is used to perform the following operation steps: By integrating the welding offset trend features and the temperature time-varying trend features, a thermo-coupling feature vector is constructed; a temperature deformation transfer function is constructed based on the thermal expansion characteristics of the taillight bracket material; the thermo-coupling feature vector is processed using the temperature deformation transfer function to output the thermal deformation pose offset component; after interactively obtaining the elastic deformation compensation amount of the robotic arm, it is superimposed on the thermal deformation pose offset component to output the predicted friction welding offset amount.

[0068] Furthermore, the intervention compensation module 60 is used to perform the following operational steps: The forward-looking intervention compensation parameters are written into the feedforward compensation register of the welding robot to establish a dynamic compensation baseline; during the welding interval of the welding robot, the forward-looking intervention compensation parameters are used to perform spatial dynamic intervention compensation; the forward-looking intervention compensation parameters are iteratively updated based on the pose residual after compensation.

[0069] Furthermore, the offset prediction module 30 is used to perform the following operation steps: Spatially synchronize the force sensing time-series data and temperature field time-series data in the multiphysics coupling data; acquire six-degree-of-freedom force fluctuations from the force sensing time-series data to obtain six-degree-of-freedom force fluctuation components; calculate the extreme values ​​of the temperature gradient and the temperature of the welding zone based on the temperature field time-series data; concatenate the six-degree-of-freedom force fluctuation components, the extreme values ​​of the temperature gradient, and the extreme values ​​of the temperature of the welding zone to construct a thermo-mechanical coupling feature vector; perform ultra-short-term nonlinear state estimation based on the thermo-mechanical coupling feature vector to output the six-degree-of-freedom pose offset vector.

[0070] Furthermore, the offset prediction module 30 is used to perform the following operation steps: Based on the time-varying offset characteristics of the force sensing time series data, a time-varying mechanical state vector sequence is constructed; the thermo-coupling feature vector is mapped to a six-degree-of-freedom virtual observation; the pose offset state is recursively deduced based on the time-varying mechanical state vector sequence, and the predicted pose offset is output; the six-degree-of-freedom virtual observation and the predicted pose offset are fused by Kalman gain weighting to output the six-degree-of-freedom pose offset vector.

[0071] Furthermore, the offset prediction module 30 is used to perform the following operation steps: The force sensing time series data is decomposed to obtain a six-degree-of-freedom time series fluctuation component; the six-degree-of-freedom time series force deviation is output by comparing the local six-degree-of-freedom reference value with the six-degree-of-freedom time series fluctuation component; the time-varying mechanical state vector sequence is constructed based on the six-degree-of-freedom time series force deviation.

[0072] Furthermore, the joint space correction module 40 is used to perform the following operation steps: The pre-stored kinematic model parameters of the welding robot arm are called to construct the forward kinematic equations; after the target pose of the welding trajectory is retrieved locally, the corrected target pose is generated by superimposing the six-degree-of-freedom pose offset vector onto the target pose of the welding trajectory; based on the inverse mapping relationship of the forward kinematic equations, the joint angle compensation amount of the corrected target pose is solved by the analytical inverse kinematic algorithm.

[0073] Furthermore, the joint space correction module 40 is used to perform the following operation steps: Based on the inverse mapping relationship of the forward kinematic equation, multiple candidate solutions for the corrected target pose are solved; the target joint angle solution is obtained by filtering from the multiple candidate solutions with the minimum joint displacement as a constraint; the Jacobian matrix rank detection is performed based on the target joint angle solution to avoid singular configurations and then the joint angle compensation amount is output.

[0074] Through the foregoing detailed description of the friction welding process control method for taillight brackets, those skilled in the art can clearly understand the friction welding process control system for taillight brackets in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0075] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method of friction welding process control for a tail lamp support characterized by, The method comprises: integrating a multi-modal sensing unit at the end of the welding robot arm, wherein the multi-modal sensing unit comprises a six-dimensional force sensor and an infrared thermal imager, and is connected to the control center through an industrial real-time Ethernet protocol; during the beat-synchronous reciprocating welding process of driving the welding robot arm based on the preset welding track, multi-physical field coupling data is uploaded in low delay through the multi-modal sensing unit; the control center receives and predicts welding control offset based on the multi-physical field coupling data, and outputs a six-degree-of-freedom pose offset vector; based on the six-degree-of-freedom pose offset vector, joint angle compensation is solved, and joint space correction of the welding robot arm is performed; the control center accumulates the six-degree-of-freedom pose offset vector for time series accumulation analysis, and outputs a forward intervention compensation parameter; the forward intervention compensation parameter is used for dynamic intervention compensation of the joint space of the welding robot arm.

2. The method of friction welding process control for a taillight support as defined in claim 1, wherein, The control center accumulates the six-degree-of-freedom pose offset vector for time series accumulation analysis, and outputs a forward intervention compensation parameter, comprising: incrementally accumulate the six-degree-of-freedom pose offset vector to obtain a pose offset time series database, wherein the pose offset time series database is associated with a welding temperature field time series slice library; performing sliding window analysis on the pose offset time series database to extract displacement deviation linear regression slope, attitude angle deviation standard deviation and offset autocorrelation coefficient as welding offset trend characteristics; performing sliding window analysis on the welding temperature field time series slice library to extract temperature gradient change rate and maximum temperature rise rate as temperature time-varying trend characteristics; based on the welding offset trend characteristics and temperature time-varying trend characteristics, a thermal deformation dynamic model is established, and a predicted friction welding offset is output; convert the predicted friction welding offset into joint space feedforward compensation, and output the forward intervention compensation parameter.

3. The method of friction welding process control for a taillight support of claim 2, wherein, Based on the welding offset trend characteristics and temperature time-varying trend characteristics, a thermal deformation dynamic model is established, and a predicted friction welding offset is output, comprising: fuse the welding offset trend characteristics and temperature time-varying trend characteristics to construct a thermal force coupling feature vector; based on the thermal expansion characteristics of the tail lamp support material, a temperature deformation transfer function is constructed; using the temperature deformation transfer function to process the thermal force coupling feature vector, a thermal deformation pose offset component is output; after obtaining the elastic deformation compensation of the robot arm by interaction, it is superimposed on the thermal deformation pose offset component to output the predicted friction welding offset.

4. The method of friction welding process control for a taillight support of claim 1 wherein, Using the forward intervention compensation parameter to perform dynamic intervention compensation of the joint space of the welding robot arm, comprising: write the forward intervention compensation parameter into the feedforward compensation register of the welding robot arm to establish a dynamic compensation baseline; during the welding gap period of the welding robot arm, the space dynamic intervention compensation is performed by using the forward intervention compensation parameter; iteratively update the forward intervention compensation parameter according to the residual pose after compensation.

5. The method of friction welding process control for a taillight support of claim 1 wherein, The control center receives and predicts welding control offset based on the multi-physical field coupling data, and outputs a six-degree-of-freedom pose offset vector, comprising: spatially synchronize the force sensing time series data and temperature field time series data in the multi-physical field coupling data; The six-degree-of-freedom force fluctuation component is obtained by six-degree-of-freedom force fluctuation acquisition based on the force sensing time series data; The temperature gradient extreme value and the welding zone temperature extreme value are calculated based on the temperature field time series data; The thermal force coupling feature vector is constructed by splicing the six-degree-of-freedom force fluctuation component, the temperature gradient extreme value and the welding zone temperature extreme value; The six-degree-of-freedom pose offset vector is output by super-short-term nonlinear state estimation based on the thermal force coupling feature vector.

6. The method of friction welding process control for a taillight support of claim 5 wherein, The six-degree-of-freedom pose offset vector is output by super-short-term nonlinear state estimation based on the thermal force coupling feature vector, including: A time-varying mechanical state vector sequence is constructed according to the time-varying offset characteristics of the force sensing time series data; The thermal force coupling feature vector is mapped into a six-degree-of-freedom virtual observation value; The predicted pose offset is output by pose offset state recursion based on the time-varying mechanical state vector sequence; The six-degree-of-freedom pose offset vector is output by Kalman gain weighted fusion of the six-degree-of-freedom virtual observation value and the predicted pose offset.

7. The method of friction welding process control for a taillight support of claim 6 wherein, A time-varying mechanical state vector sequence is constructed according to the time-varying offset characteristics of the force sensing time series data, including: The six-degree-of-freedom time series fluctuation component is obtained by decomposing the force sensing time series data; The six-degree-of-freedom time series force deviation is output by comparing the local six-degree-of-freedom reference value with the six-degree-of-freedom time series fluctuation component; The time-varying mechanical state vector sequence is constructed based on the six-degree-of-freedom time series force deviation.

8. The method of friction welding process control for a taillight support of claim 1 wherein, The joint angle compensation amount is solved based on the six-degree-of-freedom pose offset vector, and the joint space of the welding robot is corrected, including: The forward kinematics equation is constructed by calling the pre-stored welding robot kinematics model parameters; After the welding trajectory target pose is locally called, the corrected target pose is generated by superimposing the six-degree-of-freedom pose offset vector on the welding trajectory target pose; The joint angle compensation amount of the corrected target pose is solved by an analytical inverse kinematics algorithm based on the inverse mapping relationship of the forward kinematics equation.

9. The friction welding process control method for a taillight support bracket of claim 8, wherein, The joint angle compensation amount of the corrected target pose is solved by an analytical inverse kinematics algorithm based on the inverse mapping relationship of the forward kinematics equation, including: Based on the inverse mapping relationship of the forward kinematics equation, a plurality of joint angle candidate solutions of the corrected target pose are solved; The target joint angle solution is selected from the plurality of joint angle candidate solutions with the minimum joint displacement as the constraint; The joint angle compensation amount is output after avoiding singular configurations by performing Jacobian matrix rank detection based on the target joint angle solution.

10. A friction welding process control system for tail lamp brackets, characterized by, The system for implementing the friction welding process control method for the tail lamp support of any one of claims 1-9, the system comprising: A sensing unit integration module for integrating a multi-modal sensing unit at the end of the welding robot, wherein the multi-modal sensing unit comprises a six-dimensional force sensor and an infrared thermal imager, and is connected to the control console through an industrial real-time Ethernet protocol; A data uploading module for uploading multi-physical field coupling data with low delay through the multi-modal sensing unit during the beat-synchronous reciprocating welding process driven by the welding robot based on the preset welding trajectory. An offset prediction module is configured to receive and perform welding control offset prediction based on the multi-physical field coupling data, and output a six-degree-of-freedom pose offset vector; A joint space correction module is configured to solve a joint angle compensation amount based on the six-degree-of-freedom pose offset vector, and perform joint space correction of the welding robot arm; A cumulative analysis module is configured to perform time sequence cumulative analysis on the six-degree-of-freedom pose offset vector accumulated by the control center, and output a forward-looking intervention compensation parameter; An intervention compensation module is configured to perform dynamic intervention compensation of the joint space of the welding robot arm by using the forward-looking intervention compensation parameter.

Images