Accurate control method and system for multi-axis linkage of welding equipment

Abstract

The invention provides an accurate control method and system for multi-axis linkage of welding equipment, and the method comprises the steps: S1, obtaining position feedback signals from each axis through a sensor, integrating the position feedback signals into a unified data set, and obtaining a real-time position information set; s2, applying a unified clock reference to each signal in the real-time position information set to obtain a feedback data set after time synchronization; s5, extracting coordinate instruction parameters of each axis from the optimized transport stream, distributing the coordinate instruction parameters to a controller, and determining unified motion track planning; s6, adjusting the speed and position of each axis in real time according to the unified motion track planning, and obtaining coordinated motion output; and S8, extracting efficiency index data from the final welding control sequence, comparing the efficiency index data with historical records, and determining system optimization and adjustment parameters. The system performance can be dynamically optimized, and the precision, stability and production efficiency of the welding process are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent welding, in particular to a precise control method and system for multi-axis linkage of welding equipment. BACKGROUND

[0002] As the core equipment of modern manufacturing industry, welding equipment undertakes the key connection task in the automobile, aerospace, energy and other industries. The precise control of its multi-axis linkage system directly determines the welding quality, production efficiency and the reliability of complex workpiece processing. With the increasing demand for high precision and high flexibility welding in industrial production, how to realize the coordinated motion and real-time control of the multi-axis linkage system has become an important issue to promote the upgrading of intelligent manufacturing. Especially in complex curved surface workpieces or multi-dimensional welding scenes, precise control of the coordinated motion of each axis not only concerns product quality, but also affects the overall efficiency and safety of the production line.

[0003] Currently, the control method of multi-axis linkage welding equipment has exposed significant deficiencies in actual application. Many systems rely on a decentralized control architecture, with each axis controller running independently, lacking an efficient coordination mechanism. This approach can still be handled in simple scenarios, but in complex workpieces or high-speed welding tasks, it is difficult to ensure real-time synchronization of the motion of each axis. For example, when welding large aerospace components, the motion trajectories of each axis need to strictly match the geometry of the workpiece, but decentralized control often leads to command transmission delays, which in turn causes trajectory deviations or vibrations, affecting welding precision.

[0004] In terms of technology, the core difficulty of the multi-axis linkage system of welding equipment lies in how to realize the unified processing and real-time communication of position feedback signals of each axis. The collection of position feedback signals relies on encoders and position sensors, but these signals often have a time synchronization problem between different axes. Time synchronization can cause deviations in the distribution of motion commands for each axis, especially at high speeds or in complex trajectories, where this deviation can be amplified. For example, when welding a complex three-dimensional curve, if the feedback signal of a certain axis is delayed, even by milliseconds, it can cause the weld to deviate from the expected trajectory, resulting in quality defects.

[0005] Further, the root cause of signal desynchronization lies in the lack of an efficient data communication mechanism. The transmission speed and bandwidth of traditional data buses in multi-axis systems are limited, and cannot meet the real-time interaction needs of high-frequency and large amounts of data. This makes it difficult for the controller to integrate feedback information from all axes and generate unified motion commands in a short time. For example, when a five-axis welding robot performs complex curved surface welding, the main controller needs to quickly process real-time position data from five axes, but due to the limitations of low-speed buses, the delay in command distribution can cause the motion of each axis to be inconsistent, ultimately affecting the continuity and consistency of the weld.

[0006] Therefore, how to realize efficient unified processing of position feedback signals in a multi-axis linkage system and ensure real-time communication between axis controllers through a high-speed data bus becomes a key problem for improving the precise control capability of a welding device. Solving this problem not only needs to overcome the time synchronization problem of signal acquisition, but also needs to break through the bandwidth bottleneck of data transmission to ensure the high coordination of the movement of each axis in a complex welding scenario.

[0007] By solving the above problems, the welding device can realize high-precision coordinated movement under complex working conditions. For example, in automobile body welding, the multi-axis system needs to control the movement, rotation of the welding gun and the positioning of the workpiece at the same time. If the real-time coordination of each axis cannot be guaranteed, the weld may be uneven or defective, directly affecting the strength of the vehicle body. Therefore, building a control architecture that can uniformly process feedback signals and realize high-speed communication has become a key problem for precise control of multi-axis linkage of welding devices. SUMMARY

[0008] To solve the technical problems raised in the background art, the first aspect of the present application provides a precise control method for multi-axis linkage of a welding device, which comprises:

[0009] S1, obtaining position feedback signals from each axis through a sensor and integrating them into a unified data set to obtain a set of real-time position information; S2, applying a unified clock reference to each signal in the set of real-time position information to obtain a feedback data set after time synchronization; S3, if the feedback data set after time synchronization has noise interference, performing smoothing processing to obtain refined position data after noise removal; S4, obtaining a high-speed data bus transmission channel according to the refined position data after noise removal, judging whether the transmission load exceeds a preset transmission load threshold, if it does, dynamically adjusting the data packet size to obtain an optimized transmission stream; S5, extracting each axis coordination instruction parameter from the optimized transmission stream and distributing it to the controller to determine a unified motion trajectory planning; S6, adjusting the speed and position of each axis in real time according to the unified motion trajectory planning to obtain a coordinated motion output; S7, monitoring the weld path deviation through the coordinated motion output, if the deviation exceeds a preset path deviation threshold, iteratively updating the instruction parameter until the deviation meets the preset path deviation threshold to generate a final welding control sequence; S8, extracting efficiency index data from the final welding control sequence and comparing it with historical records to determine system optimization adjustment parameters.

[0010] Optionally, the step S1 of obtaining position feedback signals from each axis through a sensor and integrating them into a unified data set to obtain a set of real-time position information comprises:

[0011] Step S11, obtaining position feedback signals from each axis sensor, using a low-pass filter for signal preprocessing to obtain standardized sensor data, and integrating the standardized sensor data into a unified data set;

[0012] Step S12, generating a real-time position information set through unified data set;

[0013] Step S13, for the real-time position information set, if time sequence continuity meets a preset first continuity threshold, performing data smoothing processing on the real-time position information set, and determining a data stream required for synchronization processing;

[0014] Step S14, real-time transmitting the data stream required for synchronization processing, and outputting to a subsequent processing module.

[0015] Optionally, the step S12 of generating the real-time position information set through the unified data set comprises: generating the real-time position information set by using Kalman filtering, first initializing a state transition matrix G, a control matrix B, an observation matrix H, a process noise covariance Q and a measurement noise covariance R, then predicting a state for each time step and updating based on a new measurement, and outputting a corrected position set, wherein the state transition matrix G represents a linear relationship between a position and a speed, the control matrix B represents an external input influence, the observation matrix H maps sensor data to a state vector, the process noise covariance Q represents environmental random disturbance, and the measurement noise covariance R represents sensor error.

[0016] Optionally, the control matrix B is set to 0, and the measurement noise covariance R is set to 0.1.

[0017] Optionally, the step S2 of applying a unified clock reference to each signal in the real-time position information set to obtain a time-synchronized feedback data set comprises:

[0018] Step S21, acquiring each signal from the real-time position information set, using a network time protocol to mark a time stamp, applying a unified clock reference, and generating a time-synchronized feedback data set;

[0019] Step S22, detecting time stamp continuity of the time-synchronized feedback data set, and if the continuity meets a preset second continuity threshold, determining an effective synchronization data set;

[0020] Step S23, extracting a time sequence segment through the effective synchronization data set to obtain a local position feature set;

[0021] Step S24, smoothing time sequence data according to the local position feature set to generate a smoothed position feature set;

[0022] Step S25, acquiring a data stream from the smoothed position feature set, and segmenting according to a time stamp to obtain a segmented synchronization data set;

[0023] Step S26, outputting the segmented synchronization data set to a subsequent processing module to generate a real-time synchronization data stream.

[0024] Optionally, in the step S4, the high-speed data bus transmission channel is acquired according to the noise-removed refined position data, it is judged whether the transmission load exceeds the preset transmission load threshold value, if exceeding, the data packet size is dynamically adjusted, and the optimized transmission stream is obtained, including:

[0025] In step S41, the high-speed data bus transmission channel is acquired from the refined position data, the channel allocation information is extracted from the data stream, and the identification and state of the first transmission channel are obtained.

[0026] In step S42, the real-time transmission load data is acquired according to the identification and state of the first transmission channel, the current load value is calculated, and the load level is determined.

[0027] In step S43, if the load level exceeds the preset transmission load threshold value, the transmission load data size is adjusted, and the adjusted data packet sequence is obtained.

[0028] In step S44, the channel resource is re-allocated to the adjusted data packet sequence, and the optimized transmission stream configuration is obtained.

[0029] In step S45, the bandwidth utilization rate data is extracted from the optimized transmission stream configuration, and the bandwidth utilization rate evaluation result is obtained.

[0030] In step S46, if the bandwidth utilization rate evaluation result does not reach the preset target value, the real-time state of the second transmission channel is acquired, if the bandwidth utilization rate evaluation result of the second transmission channel also does not reach the preset target value, the channel allocation is adjusted using the PID controller, and the updated channel allocation scheme is obtained.

[0031] In step S47, the transmission performance data is extracted from the updated channel allocation scheme and the transmission efficiency is calculated, and the final transmission stream optimization result is obtained.

[0032] Optionally, in the step S42, the real-time transmission load data is acquired according to the identification and state of the first transmission channel, the current load value is calculated, and the load level is determined, and the current load value is equal to the total number of bytes of the current transmission load divided by the current transmission time.

[0033] Optionally, in the step S47, the transmission performance data is extracted from the updated channel allocation scheme and the transmission efficiency is calculated, and the final transmission stream optimization result is obtained, and the transmission efficiency is equal to the number of bytes of the optimized transmission stream divided by the transmission time.

[0034] Optionally, in the step S7, the weld path deviation is monitored by coordinating the motion, if the deviation exceeds the preset path deviation threshold value, the instruction parameters are iteratively updated until the deviation meets the preset path deviation threshold value, and the final welding control sequence is generated, including:

[0035] Step S71, real-time acquisition of motion output data through sensors;

[0036] Step S72, calculation of the Euclidean distance between the actual coordinates and the theoretical coordinates to obtain a path deviation value;

[0037] Step S73, if the path deviation value exceeds the preset path deviation threshold, the control parameters are optimized to obtain updated control parameters;

[0038] Step S74, driving the welding equipment through the updated control parameters, reacquiring the motion output data, and generating new weld path data;

[0039] Step S75, by comparing the new weld path data with the theoretical path, calculating a new deviation value, and determining whether the accuracy requirement is met;

[0040] Step S76, if the new deviation value still exceeds the preset path deviation threshold, extracting features based on the new weld path data, fitting the path features and parameter relationship, and obtaining fine-tuned control parameters;

[0041] Step S77, generating a final welding control sequence through the fine-tuned control parameters, and driving the welding equipment to perform precise path motion;

[0042] Step S78, real-time monitoring of the weld path deviation according to the final welding control sequence, and determining whether the accuracy requirement is continuously met.

[0043] In a second aspect, the present application provides a precise control system for multi-axis linkage of welding equipment, which precisely controls the multi-axis linkage of welding equipment by using the method described above, and comprises: a signal acquisition and integration module, which acquires position feedback signals from each axis through sensors and integrates them into a unified data set to obtain a real-time position information set; a time synchronization processing module, which applies a unified clock reference to each signal in the real-time position information set to obtain a feedback data set after time synchronization to eliminate asynchronization deviation; a noise filtering processing module, which performs smoothing processing on the feedback data set after time synchronization if there is noise interference, to obtain refined position data after noise removal; a transmission optimization control module, which acquires a high-speed data bus transmission channel according to the refined position data after noise removal, judges whether the transmission load exceeds a preset transmission load threshold, and if so, dynamically adjusts the data packet size to obtain an optimized transmission stream to ensure bandwidth utilization; a trajectory planning generation module, which extracts coordinated instruction parameters of each axis from the optimized transmission stream and distributes them to the controller to determine a unified motion trajectory planning; a motion coordination control module, which adjusts the speed and position of each axis in real time according to the unified motion trajectory planning to obtain coordinated motion output; a deviation monitoring and iteration module, which monitors the weld path deviation through the coordinated motion output, and if the deviation exceeds a preset path deviation threshold, iteratively updates the instruction parameters until the deviation meets the preset path deviation threshold to generate a final welding control sequence; and a performance optimization analysis module, which extracts efficiency index data from the final welding control sequence and compares it with historical records to determine system optimization adjustment parameters. The technical solution provided by the present application has the following beneficial effects:

[0044] The present application discloses a precise control method and system for multi-axis linkage of welding equipment, which addresses the problems of poor multi-axis motion synchronization, insufficient position accuracy caused by noise interference, and limited data transmission bandwidth in complex workpiece geometry scenarios. By integrating sensor feedback signals, time stamping, Kalman filtering, dynamically adjusting data packet size, and other technical means, real-time refined position data flow is constructed. The present application uses proportional-integral-derivative control algorithm to adjust the speed and position of each axis in real time, combined with weld path deviation monitoring and instruction parameter iterative updating, to ensure the stability and high precision of coordinated motion output, and finally generate an efficient welding control sequence. By comparing with historical efficiency indicators, the present application dynamically optimizes system performance, significantly improves the precision, stability, and production efficiency of the welding process, and is particularly suitable for high-precision welding scenarios of complex workpieces. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 A flowchart of a precise control method for multi-axis linkage of welding equipment according to the present application.

[0046] Figure 2 A structure diagram of a precise control system for multi-axis linkage of welding equipment according to the present application. DETAILED DESCRIPTION

[0047] For a further understanding of the present application, reference will be made to the following description taken in conjunction with the accompanying drawings. The present application will be further described with reference to the drawings, in which embodiments of the application are shown. It is to be understood that the specific embodiments described herein are merely illustrative of the application and are not to be taken in a limiting sense. It is further noted that, because some of the components of the device are known, only those specifically relevant to the understanding of the application will be described in detail below.

[0048] As shown in Figure 1 The first aspect of the present application provides a precise control method for multi-axis linkage of welding equipment, which can specifically include:

[0049] S1, obtaining position feedback signals from each axis through a sensor and integrating them into a unified data set to obtain a real-time position information set for subsequent synchronization processing.

[0050] Optionally, this step further includes:

[0051] S11, obtaining position feedback signals from each axis sensor, using a low-pass filter for signal preprocessing to obtain standardized sensor data, and integrating the standardized sensor data into a unified data set using a weighted average method.

[0052] S12, generating a real-time position information set by using Kalman filtering through the unified data set.

[0053] Preferably, the process of generating the real-time position information set includes: initializing the state transition matrix G representing the position-velocity relationship, the control matrix B, the observation matrix H, the process noise covariance Q, and the measurement noise covariance R, then predicting the state for each time step and updating based on new measurements, and outputting the corrected position set.

[0054] S13, for the real-time position information set, if the time series continuity meets the preset first continuity threshold, using moving average filtering to perform data smoothing processing on the real-time position information set to obtain smoothed position data, and using the timestamp alignment of the Pandas library to determine the data stream required for synchronization processing.

[0055] S14, using the UDP protocol to transmit the data stream required for synchronization processing in real time, and outputting to the subsequent processing module.

[0056] In one possible implementation, when obtaining position feedback signals from each axis sensor, the original data can be collected by a combination of a three-axis acceleration sensor and a gyroscope.

[0057] For example, sensors are installed on a mobile robot to collect acceleration and angular velocity data along X, Y, and Z axes, respectively, with a sampling frequency of 100 Hz and a data range of ±2g. A low-pass filter is used to preprocess the signals to effectively remove high-frequency noise.

[0058] For example, a Butterworth low-pass filter with a cutoff frequency of 10 Hz is set to filter out high-frequency components caused by welding robot vibration or environmental interference, retaining smooth position-related signals. This improves data stability and ensures subsequent processing accuracy.

[0059] For example, when integrating standardized sensor data using the weighted average method, weights can be assigned to each axis data, such as 0.4 for the X axis, 0.4 for the Y axis, and 0.2 for the Z axis, based on sensor accuracy or axis importance. Assuming the X axis data is 1.2 m / s², the Y axis data is 1.0 m / s², and the Z axis data is 0.5 m / s², the weighted average of the unified acceleration value is 1.1 m / s². This method balances the differences between multi-axis data and improves the consistency of the data set.

[0060] In one embodiment, Kalman filtering is used to generate a set of real-time position information. The state transition matrix G represents the linear relationship between position and velocity, the control matrix B represents the influence of external input, which is set to 0 (no external control input), and the observation matrix H maps sensor data to the state vector. The process noise covariance Q is set to 0.01, reflecting environmental random disturbances; the measurement noise covariance R represents sensor errors, set to 0.1 (based on sensor accuracy). After predicting the state at each time step, the state is updated with new measurements.

[0061] For example, the initial position of the welding robot is (0, 0, 0), the predicted position at the next time step is (0.1, 0.2, 0), and the measured value is (0.12, 0.21, 0.01). After Kalman filtering, the corrected position is output as (0.11, 0.205, 0.005). This method significantly reduces position estimation error and improves real-time performance.

[0062] For example, for a set of real-time position information, if the time series continuity meets the first continuity threshold (e.g., no data missing for more than 0.02 seconds), moving average filtering can be used to smooth the data. The window size is set to 5 sampling points, and the smoothed position data is more continuous, reducing sudden changes.

[0063] For example, the original position sequence is [0.11, 0.13, 0.10, 0.12, 0.11], and after smoothing, it becomes [0.114, 0.112, 0.112], effectively reducing noise interference and improving trajectory smoothness.

[0064] In one possible implementation, the Pandas library is used for timestamp alignment to determine synchronized data streams.

[0065] For example, the sensor data timestamps are [2025-08-27 04:53:00.000, 04:53:00.010, 04:53:00.021], which are aligned to a uniform time interval of 0.01 seconds using Pandas, and missing data is filled. This ensures data stream synchronization and provides consistent input for subsequent modules.

[0066] For example, smooth position data is transmitted to subsequent modules in real time using the UDP protocol. The UDP server address is set to 192.168.1.100, port 5000, and a data packet is sent every 0.01 seconds. The efficiency of UDP ensures low-latency transmission, making it suitable for scenarios with high real-time requirements, such as robot navigation modules. This solution improves the accuracy, smoothness, and real-time performance of position data, supporting efficient subsequent processing.

[0067] S2, a timestamp marking method is used to apply a uniform clock reference to each signal in the real-time position information set, obtaining a time-synchronized feedback data set to eliminate asynchronization bias.

[0068] Optionally, this step also includes:

[0069] Step S21, obtain each signal from the real-time position information set, use the Network Time Protocol to mark the timestamp, apply a uniform clock reference, and generate a time-synchronized feedback data set.

[0070] Step S22, use a verification tool to detect the timestamp continuity of the time-synchronized feedback data set. If the continuity meets the pre-set second continuity threshold, it is determined to be a valid synchronization data set.

[0071] Step S23, using the valid synchronization data set, use the Pandas library of Python to extract time series segments, and obtain a local position feature set.

[0072] Step S24, according to the local position feature set, use the uniform_filter1d function of the SciPy library to smooth the time series data, and generate a smooth position feature set.

[0073] Step S25, obtain the data stream from the smooth position feature set, and split it according to the timestamp to obtain a sliced synchronization data set.

[0074] Step S26, for the sliced synchronization data set, use the real-time transmission protocol to output to the subsequent processing module, and generate a real-time synchronization data stream.

[0075] For example, when acquiring each signal from the real-time position information set, the position data of the X, Y, and Z axes can be collected by a multi-axis sensor respectively, and each axis signal is sampled at a frequency of 10 Hz to ensure high-frequency real-time performance. After signal acquisition, a network time protocol timestamp is required. The network time protocol can be synchronized with a time server to ensure that the timestamp accuracy of each sensor data reaches the millisecond level.

[0076] For example, assuming that a certain sensor node is synchronized with a time server and the timestamp is 2025-08-27 04:53:12.345, and another node is 04:53:12.347, the time difference between the two nodes is calibrated to within 0.002 seconds by a unified clock reference to generate a time synchronization feedback data set.

[0077] Specifically, for the time synchronization feedback data set, when using a verification tool to detect the continuity of the timestamp, a timestamp difference analysis can be used. Assuming that the interval of the last 10 timestamps is within 0.1 seconds ± 0.01 seconds, which meets the preset second continuity threshold, it is confirmed as a valid synchronization data set.

[0078] For example, the timestamp sequence of a certain data set is 04:53:12.345, 04:53:12.445, and 04:53:12.545, and the interval is stable, indicating good data continuity. If a timestamp jumps to 04:53:12.645, the abnormal point needs to be removed to ensure data reliability.

[0079] In one possible implementation, when using the Pandas library to extract time series segments from the valid synchronization data set, the data can be divided into 1-second windows to obtain a local position feature set.

[0080] For example, the X-axis position data in a 1-second window is [10.2, 10.3, 10.4] meters, and the Y-axis is [5.1, 5.2, 5.3] meters. After extraction, a feature set containing position change trends is formed. This helps to analyze the local motion trajectory, such as the device moving at a constant speed of 0.2 meters along the X-axis in a short time.

[0081] For example, when using the uniform_filter1d function of the SciPy library to smooth time series data, the window size can be set to 5. For X-axis position data [10.2, 10.3, 10.5, 10.4, 10.3] meters, after smoothing, [10.3, 10.32, 10.34, 10.36, 10.34] meters are obtained, and a smoothed position feature set is generated. This can effectively reduce sensor noise interference and improve data stability.

[0082] Specifically, when acquiring data streams from the smoothed position feature set and dividing them by timestamp, a 0.5-second unit is used to generate a sharded synchronization data set.

[0083] For example, a certain data stream contains X-axis position [10.3, 10.32] meters and Y-axis [5.2, 5.25] meters from 04:53:12.000 to 04:53:12.500, forming a piece of data set. This piece method facilitates subsequent modules to process data in chronological order.

[0084] In a possible implementation, when using real-time transport protocol output for piece synchronization data set, RTSP protocol can be used to transmit data stream through local area network at a bandwidth of 100 Mbps.

[0085] For example, a certain piece of data set is transmitted at 500 bytes per packet, 20 packets per second, ensuring that the subsequent processing module receives data in real time. This improves data transmission efficiency and ensures that the subsequent module can respond in a timely manner.

[0086] For example, the overall process and Kalman filtering in historical dialogue to generate position information set complement each other. The time synchronization feedback data set can be used as the input of Kalman filtering to further optimize the position accuracy. The real-time transmission of the piece synchronization data set also complements the historical UDP protocol transmission to ensure seamless connection of data streams at different processing stages. This design improves the real-time performance and data consistency of the system.

[0087] S3, if the feedback data set after time synchronization has noise interference, the refined position data after noise removal is obtained through Kalman filtering algorithm to support accurate motion calculation. In a possible implementation, it is assumed that real-time position data is collected from multiple sensors, such as latitude and longitude coordinates of a robot, and the time stamp accuracy needs to reach milliseconds. The time stamp of sensor A recording position (100, 200) is 10:00:00.123, and the time stamp of sensor B recording (101, 201) is 10:00:00.125, which is adjusted to 10:00:00.120 through a unified clock reference. This high-precision synchronization lays a foundation for subsequent processing and ensures that data is aligned in the time dimension.

[0088] Specifically, when Kalman filtering algorithm is used to smooth the feedback data set after time synchronization, noise interference can be effectively reduced. Kalman filtering generates smoother position data through prediction and update steps, combining sensor measurements and motion models.

[0089] For example, the position data of a certain robot within 1 second contains noise, causing the trajectory to appear jittery. Kalman filtering can predict the next position based on historical position and velocity, and correct it by combining actual measurement values to generate a smoother coordinate sequence, such as smoothing (100, 200, 10:00:00) and (102, 202, 10:00:01) into a more continuous trajectory point.

[0090] In a possible implementation, for screening of abnormal values in the refined position data, the position change threshold can be set as 5 meters. If the position of a point suddenly changes to (150, 250) and the distance from the previous point is much greater than the position change threshold, it is determined to be abnormal and is removed.

[0091] For example, in normal operation of the robot, the position data usually changes smoothly between adjacent points. If a point deviates due to sensor failure, it can be removed through the position change threshold screening to ensure the stability of subsequent analysis.

[0092] For example, when filling in the missing points using the linear interpolation function of SciPy, the missing data in the time series can be supplemented. Assuming that a segment of data is missing at 10:00:02, the coordinates of 10:00:02 are calculated as (102, 202) through the previous and next points (100, 200, 10:00:01) and (104, 204, 10:00:03). This method ensures the continuity of the trajectory and facilitates subsequent motion parameter calculation.

[0093] Specifically, when calculating the velocity and acceleration based on the trajectory points, the difference function of NumPy can be used for efficient processing.

[0094] For example, the trajectory point sequence is (100, 200, 10:00:00) and (102, 202, 10:00:01), and the velocity is calculated as 2 meters per second through the difference. If the average distance of adjacent points is less than the preset distance average threshold of 0.5 meters, it is confirmed that the data is stable and suitable for further analysis. This method can accurately capture the motion characteristics.

[0095] In a possible implementation, when fitting the motion curve using the least square method, a quadratic curve can be fitted according to the position points to describe the motion trend of the robot.

[0096] For example, based on 5 trajectory points, the parabolic trajectory trend is determined to facilitate prediction of future positions. The final motion description data can be output as a time series, including position, velocity, acceleration and trend information, to support real-time monitoring.

[0097] For example, the generation of motion description data can be used for robot navigation optimization. The complete trajectory and motion parameters provide a basis for path planning, and the fitted motion curve can be used to predict potential collision risks. This multi-step processing ensures the integrity and reliability of data from collection to analysis, and provides support for high-precision positioning applications.

[0098] S4, according to the refined position data after noise removal, a high-speed data bus transmission channel is obtained, whether the transmission load exceeds the preset transmission load threshold is judged, if it exceeds, the data packet size is dynamically adjusted, and an optimized transmission stream is obtained to ensure the bandwidth utilization.

[0099] Optionally, this step also includes:

[0100] Step S41, obtain the high-speed data bus transmission channel from the refined position data, use the Wireshark tool to analyze the data stream to extract channel allocation information, and obtain the identification and state of the first transmission channel.

[0101] Step S42, according to the identification and state of the first transmission channel, obtain real-time transmission load data, use the Prometheus tool to calculate the current load value L, where L is equal to the total number of bytes of the current transmission load divided by the current transmission time, and determine the load level.

[0102] Step S43, if the load level exceeds the preset transmission load threshold, use the tcpdump tool to split the data packet to adjust the transmission load data size, and obtain the adjusted data packet sequence.

[0103] Step S44, for the adjusted data packet sequence, use the ApacheKafka tool to reallocate channel resources, and obtain the optimized transmission stream configuration.

[0104] Step S45, extract bandwidth utilization rate data from the optimized transmission stream configuration, and determine whether the bandwidth utilization rate reaches the preset target value, and obtain the bandwidth utilization rate evaluation result.

[0105] Step S46, if the bandwidth utilization rate evaluation result does not reach the preset target value, obtain the real-time state of the second transmission channel, if the bandwidth utilization rate evaluation result of the second transmission channel also does not reach the preset target value, use the PID controller to adjust the channel allocation, and obtain the updated channel allocation scheme.

[0106] Step S47, extract transmission performance data from the updated channel allocation scheme, use the iPerf tool to calculate the transmission efficiency E, where E is equal to the number of bytes of the optimized transmission stream divided by the transmission time, and obtain the final transmission stream optimization result.

[0107] In one possible implementation, obtaining the high-speed data bus transmission channel from the refined position data needs to ensure the stability and efficiency of the data stream. The refined position data is usually obtained from sensors or positioning devices and contains timestamp and coordinate information.

[0108] For example, the GPS module of a robot generates refined position data, collects 100 position points per second, and needs to be transmitted to the central processor through a high-speed bus. When using the Wireshark tool to analyze the data stream, the data packets on the bus can be captured to extract channel allocation information.

[0109] For example, Wireshark shows that the first transmission channel is identified as "CH01", the status is "active", and the data packets are transmitted at a rate of 5000 bytes per second, indicating that the channel is running normally. This intuitive and efficient analysis can quickly locate channel problems.

[0110] Specifically, based on the identification and status of the first transmission channel, real-time transmission load data can be obtained by monitoring the channel performance using the Prometheus tool. Prometheus records the current load value L through a time series database.

[0111] For example, if a channel currently transmits 50000 bytes in 10 seconds, the current load value L is 5000 bytes / second. If the preset transmission load threshold is 6000 bytes / second, the current load is not over-standard, indicating that the channel load level is normal. The visualization interface of Prometheus can clearly show the load trend, facilitating real-time monitoring.

[0112] In one embodiment, if the load level exceeds the preset transmission load threshold, for example, L reaches 7000 bytes / second, the transmission load data needs to be segmented using the tcpdump tool to optimize transmission. Tcpdump can capture and analyze data packets, and split large packets into smaller units.

[0113] For example, a 2000-byte data packet is split into two 1000-byte packets, reducing transmission delay. The adjusted data packet sequence can more evenly allocate network resources, improving transmission stability.

[0114] For example, through the adjusted data packet sequence, the channel resources need to be redistributed using Apache Kafka. Kafka, as a distributed stream processing platform, can dynamically adjust channel priorities. Assuming that the system has three channels, Kafka assigns high priority to "CH01" based on load data. The optimized transmission flow configuration can increase bandwidth utilization from 70% to 85%. This configuration optimization can significantly improve data flow smoothness.

[0115] Specifically, bandwidth utilization data is extracted from the optimized transmission flow configuration to determine whether the preset target value, for example, 90%, is reached. If the bandwidth utilization is 85%, it is not up to standard and needs to be further optimized. At this time, the real-time status of the second transmission channel is obtained, for example, "CH02" is displayed as "low load". The channel allocation is adjusted by the PID controller to dynamically increase the data sharing ratio of "CH02".

[0116] For example, originally "CH01" bears 80% of the traffic and "CH02" bears 20%, after adjustment, both of them bear 50%, so as to balance the load.

[0117] In a possible implementation, the transmission performance data is extracted from the updated channel allocation scheme, and the transmission efficiency E is calculated using the iPerf tool.

[0118] For example, the iPerf test shows that the adjusted channel transmits 60,000 bytes in 10 seconds, with an efficiency E of 6,000 bytes / s, which is 20% higher than before optimization. This efficient transmission stream optimization result ensures the real-time and reliability of data, providing a solid foundation for subsequent motion calculation.

[0119] S5, extract the axis coordination instruction parameters from the optimized transmission stream and distribute them to the controller to determine the unified motion trajectory planning to match the complex workpiece geometry. In a possible implementation, when parsing the axis instructions from the transmission stream data, the instruction stream in the high-speed bus can be analyzed by a data packet decoder.

[0120] For example, assuming that the transmission stream is 1,000 data packets per second, containing motion instructions for X, Y, and Z axes, a dedicated decoding software can be used to extract the instruction sequence for each axis, obtaining allocation information such as 200 instructions per second for the X axis and 300 instructions per second for the Y axis. This method ensures efficient and accurate instruction parsing, providing a reliable data foundation for subsequent processing.

[0121] For example, the X-axis instruction stream is decomposed into low-frequency components with a frequency below 50 Hz, representing smooth motion trends, and high-frequency components reflecting instantaneous jitter. Through this decomposition, the coordination parameters of each axis can be clearly extracted, such as the synchronization delay time between the X-axis and the Y-axis being 0.02 seconds. This method helps to separate noise and core motion characteristics.

[0122] For example, the X-axis instruction stream is decomposed into low-frequency components with a frequency below 50 Hz, representing smooth motion trends, and high-frequency components reflecting instantaneous jitter. Through this decomposition, the coordination parameters of each axis can be clearly extracted, such as the synchronization delay time between the X-axis and the Y-axis being 0.02 seconds. This method helps to separate noise and core motion characteristics.

[0123] In a possible implementation, if the coordination parameters do not match the preset coordination parameter threshold, such as a synchronization delay exceeding the delay threshold of 0.01 seconds, linear interpolation can be used for adjustment.

[0124] For example, if the difference between the X-axis and Y-axis instruction timestamps is detected to be 0.03 seconds, exceeding the delay threshold of 0.01 seconds, the timestamp sequence is interpolated to generate a new timestamp sequence, reducing the difference to 0.008 seconds and generating a smooth instruction set. This adjustment ensures smooth instruction sequences and reduces motion jitter.

[0125] For example, for generating a unified motion trajectory using A-path planning,

[0126] Based on the smooth instruction set, the motion trajectory of each axis can be calculated using the A algorithm.

[0127] For example, set the target as three-axis cooperation to complete straight line motion, the algorithm generates a continuous trajectory point set according to the speed of 0.5 m / s for the X-axis and 0.3 m / s for the Y-axis. This way ensures smooth trajectory and consistent coordination of each axis.

[0128] In a possible implementation, the real-time control system can be used by distributing the trajectory data to each axis execution unit through the controller.

[0129] For example, the trajectory point set is distributed to the servo motors of the X, Y, and Z axes at a frequency of 100 times per second, and the feedback data such as the actual displacement deviation is 0.01 meters. This feedback data provides the basis for subsequent optimization. For adjusting the instruction distribution according to the execution feedback data using the PID controller.

[0130] For example, if the X-axis feedback deviation is 0.01 meters, the instruction output ratio can be adjusted by the PID controller.

[0131] For example, increase the ratio parameter to 1.2, reduce the deviation to 0.005 meters, and finally generate a coordinated motion scheme. This way ensures accurate and stable motion through dynamic adjustment.

[0132] It can be understood that the implementation methods of the above-mentioned subjects are closely combined with the business scenarios of high-speed data bus transmission, and focus on optimizing transmission and control around refined position data. The analysis instruction, wavelet transform, parameter adjustment, path planning, data distribution, and PID control form a complete logical chain, which jointly supports an efficient and coordinated motion control scheme.

[0133] S6, according to the proportional integral derivative control algorithm, real-time adjustment of each axis speed and position is performed according to the unified motion trajectory planning, and a coordinated motion output is obtained to maintain the stability of the welding process. For example, when obtaining the deviation data of the speed and position of each axis from the sensor, the running state of the multi-axis motion system can be collected in real time through a high-precision encoder. The encoder collects 1000 data per second to generate the actual speed and position values of each axis, and the deviation data is obtained by comparing with the expected values. This way can ensure the real-time and accuracy of the deviation data, and provides a reliable basis for subsequent calculation.

[0134] In a possible implementation, the sampling frequency of the sensor can be adjusted according to the complexity of the workpiece. For example, for a complex curved surface workpiece, the sampling frequency can be increased to 2000 times per second to capture subtle deviations.

[0135] Specifically, the preliminary speed correction value P, the position adjustment increment I, and the trajectory smoothing correction value D are calculated according to the deviation data. When calculating the preliminary speed correction value P, the proportional gain Kp can be preset to 5 according to the system characteristics. Assuming that the deviation e(t) of a certain axis is 0.2 mm, the preliminary speed correction value P = 5 × 0.2 = 1 mm / s. This method directly affects the speed adjustment by amplifying the deviation, and quickly responds to system changes.

[0136] It should be noted that the selection of the proportional gain Kp needs to be combined with the workpiece material and the movement speed, for example, an aluminum alloy workpiece may need a smaller Kp to avoid over-regulation.

[0137] In a possible implementation, the calculation of the position adjustment increment I can eliminate steady-state errors by accumulating the deviation.

[0138] For example, the deviation e(t) is accumulated to 1 mm in 10 consecutive samples, and the integral gain Ki is set to 0.1, then the position adjustment increment I = 0.1 x 1 = 0.1 mm / s. The integral term can record historical deviations, which is suitable for processing continuous deviations caused by thermal deformation during welding.

[0139] Preferably, the integral time window can be set to 0.5 seconds to prevent excessive accumulation from causing system oscillation.

[0140] For example, the calculation of the trajectory smoothing correction value D captures the dynamic trend through the deviation change rate. Assuming the deviation change rate de(t) / dt is 0.1 mm / s, and the differential gain Kd is set to 2, then the trajectory smoothing correction value D = 2 x 0.1 = 0.2 mm / s. This method can predict the change in deviation and smooth the trajectory adjustment, which is particularly suitable for high-speed welding scenarios.

[0141] It should be noted that a too high differential gain Kd value may cause noise amplification, so the sensor data can be preprocessed by a low-pass filter.

[0142] Specifically, when the Arduino PID library integrates P, I, and D terms, the trajectory smoothing correction value can be limited by a preset correction value threshold of 0.5 mm / s.

[0143] For example, the total control output u(t) = 1.3 mm / s of P = 1, I = 0.1, and D = 0.2 exceeds the correction value threshold, so it is automatically limited to 0.5 mm / s by the library function. This way ensures that the trajectory adjustment is not excessive and maintains system stability.

[0144] Preferably, the Arduino can communicate with the controller through a serial port, and the control output is updated in real time every 10 milliseconds.

[0145] In a possible implementation, real-time updating of the speed and position of each axis can be performed by a servo motor.

[0146] For example, the optimized trajectory data indicates that the speed of a certain axis is adjusted to 10 mm / s and the position is corrected by 0.3 mm, and the servo motor can accurately respond. When continuously monitoring the output deviation of the welding process, the laser range finder can detect the weld position every second, and the deviation exceeding 0.1 mm triggers a loop adjustment. This closed-loop control can correct the deviation in real time and ensure the consistency of the weld.

[0147] For example, the cycle of inputting the deviation data to the calculation of the preliminary speed correction value P, the position adjustment increment I and the trajectory smoothing correction value D can be realized by the embedded controller. Assuming that the deviation of 0.15 mm is detected during the welding process, the control output is updated after recalculating the P, I and D terms. This continuous cycle mechanism can dynamically adapt to external factors such as workpiece deformation or equipment wear, maintaining the stability of the welding process.

[0148] It should be noted that the cycle frequency can be set to 100 Hz to ensure the balance between real-time performance and computational efficiency.

[0149] S7, monitor the weld path deviation by coordinating the motion output, and if the deviation exceeds the preset path deviation threshold, iteratively update the instruction parameters until the deviation meets the preset path deviation threshold, and generate the final welding control sequence.

[0150] Optionally, this step further comprises:

[0151] Step S71, real-time acquisition of motion output data by sensors.

[0152] Step S72, calculating the Euclidean distance between the actual coordinates and the theoretical coordinates to obtain the path deviation value.

[0153] Preferably, the path deviation value is calculated using the following formula:

[0154]

[0155] wherein, is the path deviation value, (X is the actual coordinate, (X is the theoretical coordinate.

[0156] Step S73, if the path deviation value exceeds the preset path deviation threshold, define the loss function as the sum of the squares of the deviation values, and use the minimize function in the SciPy library of Python to realize the gradient descent algorithm to optimize the control parameters, and obtain the updated control parameters.

[0157] Step S74, driving the welding equipment by the updated control parameters, reacquiring the motion output data, and generating new weld path data.

[0158] Step S75, calculating the new deviation value by comparing the new weld path data with the theoretical path, and judging whether the accuracy requirement is met.

[0159] Step S76, if the new deviation value still exceeds the preset path deviation threshold, extracting the features based on the new weld path data, training the support vector regression model to fit the relationship between the path features and the parameters using the SVR function of the sklearn library, and obtaining the fine-tuned control parameters.

[0160] Step S77, generate the final welding control sequence through the fine-tuned control parameters, and drive the welding equipment to perform precise path motion.

[0161] Step S78, according to the final welding control sequence, real-time monitor the weld path deviation, and judge whether the accuracy requirement is continuously met.

[0162] For example, in the path control of welding equipment, real-time acquisition of motion output data through sensors is the core to ensure weld accuracy. Sensors usually include laser range finders and inertial measurement units to capture the three-dimensional position of the welding head. Assuming that a welding equipment is running, the sensor detects the actual coordinates as (100.0, 50.0, 10.0), while the theoretical coordinates are (100.5, 50.2, 10.1). Calculate the path deviation by the Euclidean distance formula, and get the deviation value d = 0.3 mm. If the preset path deviation threshold is 0.2 mm, the deviation is out of tolerance and needs to be optimized.

[0163] Specifically, after defining the loss function as the sum of the squares of the deviation values, the minimize function in the SciPy library of Python can be used for gradient descent optimization. The initial control parameters such as the speed of the servo motor are set to 500 rpm, and after optimization, they are adjusted to 510 rpm to reduce the deviation. This method gradually approaches the theoretical path by iteratively adjusting the parameters, ensuring the consistency of the weld.

[0164] In one embodiment, if the optimized deviation is still out of the path deviation threshold, the features of the weld path such as curvature and speed variation can be extracted, and the SVR model of sklearn can be used for fitting. Assuming that the extracted features include the curvature radius of the weld path as 20 mm and the speed variation rate as 0.5 mm / s, the SVR model can predict more accurate control parameters, such as adjusting the motor speed to 512 rpm. This method uses machine learning to capture the relationship between complex path features and parameters, further improving accuracy.

[0165] For example, after the final welding control sequence is generated, the equipment performs path motion according to the sequence. Real-time monitoring shows that the new deviation is reduced to 0.15 mm, meeting the accuracy requirement. This closed-loop control ensures that the weld path always adheres to the theoretical trajectory through continuous feedback and adjustment.

[0166] Preferably, the monitoring system can combine visual sensors to detect the appearance quality of the weld, further verifying the path accuracy.

[0167] In one embodiment, the performance of the welding equipment on different workpiece materials such as steel and aluminum can be adapted by adjusting the parameters of the SVR model.

[0168] For example, when welding steel, the model can prioritize the impact of thermal deformation and optimize the speed parameters; when welding aluminum, more attention is paid to path smoothness. This flexibility allows the control sequence to adapt to various working conditions, improving the versatility of the equipment.

[0169] For example, when monitoring the deviation of the weld path in real time, multiple threshold levels can be set, such as 0.1 mm for optimal and 0.2 mm for qualified, and the optimization frequency is dynamically adjusted. Assuming that the monitoring deviation is 0.18 mm, the system can only fine-tune the parameters instead of retraining the model, saving computing resources. This hierarchical strategy balances between high precision and efficiency.

[0170] In one embodiment, the frequency of sensor data collection can be dynamically adjusted according to the welding speed. For example, when welding at high speed, the collection frequency is increased to 100 Hz to ensure that subtle deviations are captured; when welding at low speed, the frequency is reduced to 50 Hz to reduce energy consumption. This adaptive collection method optimizes data processing efficiency and ensures real-time performance.

[0171] S8, extract efficiency index data from the final welding control sequence and compare it with historical records to determine system optimization adjustment parameters to improve overall production performance.

[0172] Optionally, this step further includes:

[0173] Step S81, obtain efficiency index data from the welding control sequence, load sequence data using the Pandas library and extract welding time and yield as efficiency indexes to obtain a first index data set.

[0174] Step S82, normalize the first index data set using the MinMaxScaler class to obtain a normalized index set.

[0175] Step S83, if the cosine similarity between the normalized index set and the pre-established historical data set is greater than the preset similarity threshold 0.9, query the corresponding historical record from the database to obtain a historical performance benchmark.

[0176] Step S84, according to the difference between the historical performance benchmark and the normalized index set as the deviation, use the LinearRegression class of the Scikit-learn library to fit the time trend of the deviation sequence to determine the performance deviation value.

[0177] Step S85, simulate adjustment of the welding control sequence through the performance deviation value, and use the gradient descent algorithm to iteratively update the sequence parameters to obtain an adjustment parameter set.

[0178] Step S86, modify the voltage and current values of the welding control sequence through the adjustment parameter set to obtain optimized sequence data.

[0179] Step S87, calculate the average welding efficiency of the optimized sequence data, compare with the original sequence data, if the improvement rate exceeds 5%, judge that the production performance improvement is effective.

[0180] For example, in the efficiency index analysis of the welding control sequence, welding time and yield are used as key indicators, which reflect the core of production efficiency. Pandas library can be used to load sequence data, assuming that a data table containing timestamp, welding time and hourly yield is read from a CSV file.

[0181] For example, a record shows that a welding process takes 120 seconds and produces 50 pieces, which can be extracted as the first indicator data set. This data set reflects the actual running efficiency of the welding equipment and lays the foundation for subsequent analysis.

[0182] In one possible implementation, when normalizing the first indicator data set, MinMaxScaler scales the data to the range of 0 to 1, which helps to unify indicators of different dimensions.

[0183] For example, welding time ranges from 100 seconds to 200 seconds, and yield ranges from 40 pieces to 60 pieces. After normalization, the time of a certain record may become 0.5 and the yield may become 0.6. This standardization facilitates subsequent comparison and calculation of similarity, avoiding the interference of dimensional differences on the analysis results.

[0184] Specifically, when calculating the cosine similarity between the normalized indicator set and the historical data set, the efficiency indicators of the past week can be extracted from the database as the historical data set. If the similarity calculated at a certain time is 0.92, which exceeds the similarity threshold of 0.9, the corresponding historical record is queried to obtain the performance benchmark at that time, such as average welding time of 110 seconds and yield of 55 pieces. This benchmark provides a reference for subsequent deviation analysis.

[0185] For example, based on the difference between the historical performance benchmark and the normalized indicator set, the performance deviation value can be calculated. Assuming that the normalized time is 0.5 and the historical benchmark is 0.45, the deviation is 0.05. When fitting the time trend of the deviation sequence using LinearRegression, the deviation data of the past 10 days can be analyzed to observe whether the deviation increases or decreases over time.

[0186] For example, if the fitting result shows that the deviation has a linear growth trend, it indicates that the welding efficiency gradually deviates from the expectation, and the control sequence needs to be adjusted.

[0187] In one possible implementation, the gradient descent algorithm is used to iteratively update the parameters of the welding control sequence.

[0188] For example, for voltage and current, the initial voltage is 220V and the current is 100A. Through multiple iterations, the voltage is adjusted to 218V and the current is adjusted to 102A, forming an adjusted parameter set. This adjustment is based on the deviation trend and optimizes the running parameters of the sequence.

[0189] Specifically, the optimized sequence data is re-run on the welding equipment with the adjusted voltage and current values.

[0190] For example, the optimized welding time is shortened to 115 seconds and the output is increased to 58 pieces. When calculating the average welding efficiency, assuming the original efficiency is 0.416 pieces per second and the optimized efficiency is 0.504 pieces per second, the improvement rate = (0.504-0.416) / 0.416 ≈ 21%, which is much higher than the efficiency index threshold of 5%, indicating that the performance improvement is effective. This improvement is reflected in production, which can significantly improve equipment utilization and output stability. For example, when monitoring the optimized sequence data in real time, the welding time and output can be continuously collected, and the deviation from the theoretical efficiency can be calculated.

[0191] For example, if the theoretical efficiency is 0.5 pieces per second and the actual efficiency is 0.49 pieces per second, the deviation is within an acceptable range, indicating that the optimized sequence can run stably. This continuous monitoring ensures that the accuracy and efficiency of the welding process remain at a high level for a long time, providing reliable protection for production.

[0192] For example, if the theoretical efficiency is 0.5 pieces per second and the actual efficiency is 0.49 pieces per second, the deviation is within an acceptable range, indicating that the optimized sequence can run stably. This continuous monitoring ensures that the accuracy and efficiency of the welding process remain at a high level for a long time, providing reliable protection for production. Figure 2As shown, the second aspect of the present application provides a precise control system for multi-axis linkage of welding equipment, which precisely controls the multi-axis linkage of welding equipment by using the method described above. The system mainly includes: a signal acquisition and integration module A1, which is used to obtain position feedback signals from each axis through sensors and integrate them into a unified data set to obtain a real-time position information set; a time synchronization processing module A2, which is used to apply a unified clock reference to each signal in the real-time position information set by using a timestamp marking method to obtain a time-synchronized feedback data set; a noise filtering processing module A3, which is used to smooth the noise-removed refined position data by using Kalman filtering algorithm if there is noise interference in the time-synchronized feedback data set; a transmission optimization control module A4, which is used to obtain a high-speed data bus transmission channel according to the noise-removed refined position data, judge whether the transmission load exceeds the preset transmission load threshold, and if it exceeds, dynamically adjust the data packet size to obtain an optimized transmission stream to ensure the bandwidth utilization; a trajectory planning generation module A5, which is used to extract the coordinated instruction parameters of each axis from the optimized transmission stream and distribute them to the controller to determine a unified motion trajectory planning; a motion coordination control module A6, which is used to adjust the speed and position of each axis in real time according to the unified motion trajectory planning by using a proportional-integral-derivative control algorithm to obtain a coordinated motion output; a deviation monitoring and iteration module A7, which is used to monitor the weld path deviation through the coordinated motion output, and if the deviation exceeds the preset path deviation threshold, iteratively update the instruction parameters until the deviation meets the preset path deviation threshold to generate a final welding control sequence; a performance optimization analysis module A8, which is used to extract efficiency index data from the final welding control sequence and compare it with historical records to determine system optimization adjustment parameters. Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A precise control method for multi-axis linkage of welding equipment, characterized in that, The method includes: S1. Acquire position feedback signals from each axis using sensors and integrate them into a unified dataset to obtain a real-time position information set. S2. Apply a unified clock reference to each signal in the real-time position information set to obtain a time-synchronized feedback dataset. S3. If noise interference exists in the time-synchronized feedback dataset, smoothing is performed to obtain refined position data after noise removal. S4. Obtain a high-speed data bus transmission channel based on the refined position data after noise removal, and determine if the transmission load exceeds a preset transmission load threshold. If it does, dynamically adjust the data packet size to obtain an optimized transmission flow. S5. Extract coordination command parameters for each axis from the optimized transmission flow and distribute them to the controller to determine a unified motion trajectory plan. S6. Adjust the speed and position of each axis in real time according to the unified motion trajectory plan to obtain coordinated motion output. S7. Monitor weld path deviation through the coordinated motion output. If the deviation exceeds a preset path deviation threshold, iteratively update the command parameters until the deviation meets the preset path deviation threshold, generating the final welding control sequence. S8. Extract efficiency index data from the final welding control sequence and compare it with historical records to determine system optimization adjustment parameters.

2. The precise control method for multi-axis linkage of welding equipment according to claim 1, characterized in that, Step S1 involves acquiring position feedback signals from each axis using sensors and integrating them into a unified dataset to obtain a real-time position information set, including: Step S11: Obtain position feedback signals from each axis sensor, perform signal preprocessing using a low-pass filter to obtain standardized sensor data, and integrate the standardized sensor data into a unified dataset. Step S12: Generate a set of real-time location information using a unified dataset; Step S13: For the real-time location information set, if the time series continuity meets the preset first continuity threshold, then perform data smoothing processing on the real-time location information set and determine the data stream required for synchronization processing. Step S14: Transmit the data stream required for synchronous processing in real time and output it to the subsequent processing module.

3. The precise control method for multi-axis linkage of welding equipment according to claim 2, characterized in that, Step S12, generating a real-time location information set using a unified dataset, includes: generating the real-time location information set using Kalman filtering; first, initializing the state transition matrix G, control matrix B, observation matrix H, process noise covariance Q, and measurement noise covariance R; then, predicting the state for each time step and updating it based on new measurements, outputting the corrected location set; wherein, the state transition matrix G represents the linear relationship between position and velocity, the control matrix B represents the influence of external input, the observation matrix H maps sensor data to a state vector, the process noise covariance Q represents random environmental disturbances, and the measurement noise covariance R represents sensor error.

4. The precise control method for multi-axis linkage of welding equipment according to claim 3, characterized in that, The control matrix B is set to 0, and the measurement noise covariance R is set to 0.

1.

5. The precise control method for multi-axis linkage of welding equipment according to claim 1, characterized in that, Step S2 involves applying a unified clock reference to each signal in the real-time location information set to obtain a time-synchronized feedback dataset, including: Step S21: Obtain each signal from the real-time location information set, mark the timestamp using the network time protocol, apply a unified clock reference, and generate a time synchronization feedback dataset; Step S22: Detect the continuity of timestamps in the time synchronization feedback dataset. If the continuity meets the preset second continuity threshold, then determine the valid synchronization dataset. Step S23: By effectively synchronizing the dataset, extract time series segments to obtain a set of local location features; Step S24: Based on the local location feature set, smooth the time series data to generate a smoothed location feature set; Step S25: Obtain the data stream from the smoothed location feature set, split it by timestamp, and obtain the sharded synchronization dataset; Step S26: For the sharded synchronization dataset, output to the subsequent processing module to generate a real-time synchronization data stream.

6. The precise control method for multi-axis linkage of welding equipment according to claim 1, characterized in that, Step S4 involves obtaining a high-speed data bus transmission channel based on the refined position data after noise removal, determining whether the transmission load exceeds a preset transmission load threshold, and if so, dynamically adjusting the data packet size to obtain an optimized transmission flow, including: Step S41: Obtain the high-speed data bus transmission channel from the refined position data, parse the data stream to extract the channel allocation information, and obtain the identifier and status of the first transmission channel; Step S42: Based on the identifier and status of the first transmission channel, obtain real-time transmission load data, calculate the current load value, and determine the load level; Step S43: If the load level exceeds the preset transmission load threshold, adjust the transmission load data size to obtain the adjusted data packet sequence. Step S44: Reallocate channel resources to the adjusted data packet sequence to obtain the optimized transport stream configuration; Step S45: Extract bandwidth utilization data from the optimized transport flow configuration to obtain bandwidth utilization evaluation results; Step S46: If the bandwidth utilization evaluation result does not reach the preset target value, the real-time status of the second transmission channel is obtained. If the bandwidth utilization evaluation result of the second transmission channel also does not reach the preset target value, the PID controller is used to adjust the channel allocation to obtain an updated channel allocation scheme. Step S47: Extract transmission performance data from the updated channel allocation scheme and calculate transmission efficiency to obtain the final transmission flow optimization result.

7. The precise control method for multi-axis linkage of welding equipment according to claim 6, characterized in that, In step S42, real-time transmission load data is obtained based on the identifier and status of the first transmission channel, and the current load value is calculated to determine the load level. The current load value is equal to the total number of bytes of the current transmission load divided by the current transmission time.

8. The precise control method for multi-axis linkage of welding equipment according to claim 7, characterized in that, In step S47, transmission performance data is extracted from the updated channel allocation scheme and transmission efficiency is calculated to obtain the final transmission stream optimization result. The transmission efficiency is equal to the number of bytes in the optimized transmission stream divided by the transmission time.

9. The precise control method for multi-axis linkage of welding equipment according to claim 1, characterized in that, Step S7 involves monitoring weld path deviation through coordinated motion output. If the deviation exceeds a preset path deviation threshold, the command parameters are iteratively updated until the deviation meets the preset path deviation threshold, generating the final welding control sequence, including: Step S71: Collect motion output data in real time using sensors; Step S72: Calculate the Euclidean distance between the actual coordinates and the theoretical coordinates to obtain the path deviation value; Step S73: If the path deviation value exceeds the preset path deviation threshold, optimize the control parameters to obtain the updated control parameters. Step S74: Drive the welding equipment with the updated control parameters, re-acquire motion output data, and generate new weld path data; Step S75: By comparing the new weld path data with the theoretical path, calculate the new deviation value and determine whether the accuracy requirements are met. Step S76: If the new deviation value still exceeds the preset path deviation threshold, then extract features based on the new weld path data, fit the relationship between path features and parameters, and obtain the fine-tuned control parameters. Step S77: Using the finely tuned control parameters, generate the final welding control sequence to drive the welding equipment to perform precise path movements; Step S78: Based on the final welding control sequence, monitor the weld path deviation in real time to determine whether the accuracy requirements are continuously met.

10. A precision control system for multi-axis linkage of welding equipment, characterized in that, The system employs the method described in any one of claims 1-9 for precise control of multi-axis linkage in welding equipment. The system comprises: a signal acquisition and integration module (A1) for acquiring position feedback signals from each axis via sensors and integrating them into a unified dataset to obtain a real-time position information set; a time synchronization processing module (A2) for applying a unified clock reference to each signal in the real-time position information set to obtain a time-synchronized feedback dataset to eliminate asynchronous deviations; a noise filtering processing module (A3) for smoothing the time-synchronized feedback dataset if noise interference exists, to obtain refined position data after noise removal; and a transmission optimization control module (A4) for acquiring a high-speed data bus transmission channel based on the refined position data after noise removal and determining whether the transmission load exceeds a preset transmission load. The system employs several mechanisms: a load threshold, dynamically adjusting data packet size if exceeded to optimize transmission flow and ensure bandwidth utilization; a trajectory planning generation module (A5), extracting coordination command parameters for each axis from the optimized transmission flow and distributing them to the controller to determine a unified motion trajectory plan; a motion coordination control module (A6), adjusting the speed and position of each axis in real time according to the unified motion trajectory plan to obtain coordinated motion output; a deviation monitoring and iteration module (A7), monitoring weld path deviation through coordinated motion output, iterating and updating command parameters if the deviation exceeds a preset path deviation threshold until the deviation meets the preset path deviation threshold, generating the final welding control sequence; and a performance optimization analysis module (A8), extracting efficiency index data from the final welding control sequence and comparing it with historical records to determine system optimization adjustment parameters.

Images