Parallel directional control method and system for multiple welding devices

Abstract

The invention provides a parallel orientation control method and system for multiple welding devices, and the method comprises the steps: S1, collecting real-time three-dimensional space coordinate data and device attitude angle information from multiple operation devices through a multi-source sensor, and obtaining an initial positioning information set; s2, performing multi-source data fusion processing on the initial positioning information set by adopting a state estimation filtering algorithm to obtain a fused operation equipment position and attitude estimation result; s5, acquiring the corrected accurate equipment positioning information and the three-dimensional geometric model data of the target workpiece, and generating a task allocation scheme of multi-equipment collaborative operation; and S7, simulating a potential motion interference scene among multiple devices according to the preliminary movement path sequence and the device attitude adjustment specification, and judging whether a path conflict condition exists or not. According to the method, the positioning precision and the operation efficiency of multi-device collaborative operation are remarkably improved, path conflicts are effectively avoided, and the stability and the safety of coordinated control in a complex scene are ensured.

Description

Technical Field

[0001] This invention relates to the field of welding technology, and in particular to a method and system for parallel directional control of multiple welding devices. Background Technology

[0002] Welding technology, as one of the core processes in modern manufacturing, is widely used in aerospace, automobile manufacturing, shipbuilding, and other fields. Its precision and efficiency directly determine product quality and production efficiency. With the increasing demand for complex and large workpieces in industrial production, the collaborative operation of multiple welding machines has become crucial for improving efficiency. However, ensuring high-precision collaborative control of multiple machines in complex operating environments not only affects production efficiency but also has a profound impact on product quality and safety. Technological advancements in this field are of great significance for promoting intelligent manufacturing and industrial automation.

[0003] Current multi-welding equipment collaborative operation solutions mostly rely on single positioning technologies or simple communication protocols. While this approach is adequate for small, regular workpieces, it often reveals significant shortcomings when dealing with large workpieces with complex geometries. Especially when the workpiece surface has irregular curves or dynamic changes, existing methods struggle to adapt to the real-time coordination needs of multiple devices. Accumulated positioning errors and communication delays between devices can lead to welding path deviations or angular inaccuracies, thus affecting weld quality. For example, in the welding of large ships, if multiple devices cannot be precisely synchronized, it may result in uneven welds or even cracks.

[0004] In the parallel orientation control of multiple welding devices, the core technical challenge lies primarily in achieving high-precision positioning. Each device needs to determine its own position and attitude angle in three-dimensional space in real time, which requires the integration of multiple positioning technologies, such as laser ranging, visual recognition, and inertial navigation. However, the integration of these technologies is not a simple superposition; it requires addressing the issue of data consistency between different technologies in dynamic environments. For example, laser ranging may become inaccurate under smoke or light interference, while visual recognition is not adaptable enough to complex curved surfaces, making it difficult to guarantee the positioning accuracy of the equipment under complex working conditions.

[0005] Insufficient positioning accuracy directly leads to another key technical challenge: collaborative path planning among multiple devices. Even if each device can obtain its own position, if the group control center cannot calculate the movement path and orientation angle of each device in real time according to the overall task requirements, conflicts will occur when multiple devices are operating on the same workpiece. For example, when welding large aerospace components, if one device deviates from its designated position due to improper path planning, it may interfere with other devices, resulting in incorrect welding angles or equipment collisions, thereby affecting the stability and safety of the entire production process.

[0006] Therefore, how to achieve high-precision positioning of equipment in dynamic and complex environments by integrating multiple positioning technologies, and based on this, how to plan conflict-free movement paths and orientation angles for multiple devices in real time, has become a key issue in the parallel orientation control of multiple welding devices. Solving this problem requires not only technological breakthroughs but also ensuring the efficiency and reliability of multiple devices in collaborative operations to meet the demands of modern manufacturing for high-quality and high-efficiency production. Summary of the Invention

[0007] To address the technical problems mentioned in the background art, a first aspect of the present invention provides a parallel directional control method for multiple welding devices, the method comprising:

[0008] S1. Real-time three-dimensional spatial coordinate data and equipment attitude angle information are collected from multiple operating devices using multi-source sensors to obtain an initial positioning information set; S2. The initial positioning information set is subjected to multi-source data fusion processing to obtain fused operating device position and attitude estimation results; S3. Based on the fused operating device position and attitude estimation results, a relative spatial position relationship model among multiple devices is constructed to determine the coordination control parameters required for group control; S4. If the deviation between the fused operating device position and attitude estimation results and a preset position threshold exceeds the allowable range, the estimation results are corrected in real time to obtain the corrected accurate equipment position and attitude estimation results. S5. Obtain the corrected precise equipment positioning information and the three-dimensional geometric model data of the target workpiece, and generate a task allocation scheme for multi-equipment collaborative operation; S6. For the task allocation scheme, calculate the conflict-free movement trajectory of each working equipment from the current corrected position to the target working point, and obtain a preliminary movement path sequence; S7. Based on the preliminary movement path sequence and equipment attitude adjustment specifications, simulate potential motion interference scenarios between multiple devices, and determine whether there is a path conflict; S8. If there is a path conflict, iteratively optimize the movement path sequence based on the simulation interference results to obtain the final conflict-free equipment movement path and attitude adjustment plan.

[0009] Optionally, in step S1, real-time three-dimensional spatial coordinate data and equipment attitude angle information are collected from multiple operating devices using multi-source sensors to obtain an initial positioning information set, including:

[0010] Step S11: Collect the three-dimensional coordinates and attitude angles of multiple devices through multi-source sensors to generate an initial positioning dataset;

[0011] Step S12: Smooth the three-dimensional coordinates and attitude angles in the initial positioning dataset to obtain a smoothed positioning dataset;

[0012] Step S13: Based on the smooth positioning dataset, calculate the relative position of each device in three-dimensional space and determine the spatial relationship between the devices;

[0013] Step S14: Optimize and fuse the three-dimensional coordinates based on the spatial relationship between devices to generate an optimized positioning dataset;

[0014] Step S15: Fit the time series according to the optimized positioning dataset, calculate the residuals, and if the residuals exceed the preset residual threshold, they are detected as outliers and removed to obtain the outlier removal dataset.

[0015] Step S16: Based on the anomaly removal dataset, the particle filter algorithm is used to predict the future position and attitude of the device, where the particle represents the position and attitude state. The particle weights are updated by resampling to generate the predicted positioning dataset.

[0016] Step S17: By predicting the positioning dataset, connecting the location points, constructing the device's motion trajectory, and determining the dynamic change trend of the device.

[0017] Optionally, step S13, calculating the relative position of each device in three-dimensional space based on the smoothed positioning dataset and determining the spatial relationship between devices, includes:

[0018] The relative position is calculated using the following Euclidean distance formula:

[0019]

[0020] in, It refers to the relative position of each device in three-dimensional space. These are the coordinates of the first device. These are the coordinates of the second device.

[0021] Optionally, step S2 involves performing multi-source data fusion processing on the initial positioning information set to obtain the fused position and attitude estimation results of the operating equipment, including:

[0022] Step S21: Data is acquired from multiple sensors of the working equipment to form an initial information set containing position and attitude;

[0023] Step S22: Clean the initial information set to obtain a standardized data set;

[0024] Step S23: If the integrity of the standardized dataset meets the preset integrity threshold, then perform preliminary state estimation to obtain preliminary position and attitude estimation results.

[0025] Step S24: Based on the preliminary position and attitude estimation results, perform multi-source data fusion to obtain the optimized fusion result;

[0026] Step S25: If the error of the fusion result is lower than the preset error threshold, then determine the final position and pose estimation result.

[0027] Step S26: Update the status information of the operating equipment based on the final position and attitude estimation results;

[0028] Step S27: Generate dynamic position and attitude data of the working equipment based on the status information.

[0029] Optionally, step S3, based on the fused position and attitude estimation results of the operating equipment, constructs a relative spatial position relationship model among multiple devices and determines the coordination control parameters required for group control, including:

[0030] Step S31: Obtain position and attitude estimation data from multiple operating devices, and fuse them using a Kalman filter to obtain the fusion result;

[0031] Step S32: Calculate the relative spatial relationships between the multiple devices based on the fusion results;

[0032] Step S33: Extract relative distance and angle information from the relative spatial positional relationship to generate a positional relationship matrix;

[0033] Step S34: If the distance between devices in the position relationship matrix is ​​less than a preset distance threshold, then calculate the attitude offset based on the distance and adjust the device attitude to obtain the optimized spatial distribution.

[0034] Step S35: Based on the optimized spatial distribution, calculate the coordination parameters required for group control and generate a set of control parameters;

[0035] Step S36: By controlling the parameter set, motion commands are allocated to each device to determine the coordinated motion trajectory of the group;

[0036] Step S37: Update the position and attitude data of each device according to the coordinated motion trajectory of the group, and obtain a new fusion result through a Kalman filter.

[0037] Optionally, step S5, acquiring the corrected precise equipment positioning information and the three-dimensional geometric model data of the target workpiece, and generating a task allocation scheme for multi-equipment collaborative operation, includes:

[0038] Step S51: Obtain the corrected equipment positioning data and the three-dimensional geometric model of the target workpiece from the sensor, parse the data, extract the equipment coordinate information and workpiece geometric features, and obtain the spatial position of the equipment and the surface description of the workpiece.

[0039] Step S52: Based on the spatial location of the equipment and the surface description of the workpiece, group the equipment and determine the equipment cooperation group;

[0040] Step S53: Calculate task priority based on workpiece processing requirements and equipment operating status to obtain task allocation results;

[0041] Step S54: Based on the task allocation results and the task execution sequence extracted from the task allocation results, generate a scheduling sequence for multi-device collaborative operation and determine the order of device operation;

[0042] Step S55: Based on the equipment operation sequence, process the real-time data stream to monitor the equipment operation status. After determining that the operation status meets the workpiece processing requirements, optimize the task execution sequence to obtain the final collaborative operation plan.

[0043] Optionally, step S7, based on the preliminary movement path sequence and device attitude adjustment specifications, simulates potential motion interference scenarios between multiple devices to determine whether path conflicts exist, including:

[0044] Step S71: By parsing the movement path sequence, extract the spatial location data of each device in the time series to obtain the device movement trajectory;

[0045] Step S72: Obtain the spatial coordinates of each device at each time point from the device motion trajectory;

[0046] Step S73: Group the device motion trajectories, extract position feature vectors from the trajectory data, initialize the number of clusters to the number of devices using the K-Means class of the sklearn library, fit the data to obtain cluster labels, calculate the intersection probability based on the trajectory overlap within the cluster, obtain the spatial distribution characteristics of the trajectory, and determine the probability of trajectory intersection.

[0047] Step S74: If the probability of trajectory intersection is greater than a preset probability threshold, then the nearest distance between trajectories is calculated through spatial location analysis to determine potential conflict areas.

[0048] Step S75: For potential conflict areas, analyze the time overlap of the equipment in the conflict area by combining the motion time series, and determine whether there is a path conflict.

[0049] Step S76: Replan the device trajectories with path conflicts, search for alternative paths in the conflict area, and obtain the adjusted movement path sequence.

[0050] Step S77: Based on the adjusted movement path sequence, run time step simulation to verify whether the path conflict has been eliminated and obtain the final conflict judgment result.

[0051] Optionally, in step S74, if the probability of trajectory intersection is greater than a preset probability threshold, then through spatial location analysis, the nearest distance between trajectories is calculated to determine potential conflict areas, including:

[0052] Calculate the shortest distance between trajectories using the following formula:

[0053]

[0054] in, It is the shortest distance between two trajectories. These are the coordinates of the first trajectory point. These are the coordinates of the second trajectory point.

[0055] Optionally, in step S8, if path conflicts exist, the movement path sequence is iteratively optimized based on the simulated interference results to obtain a final conflict-free device movement path and attitude adjustment plan, including:

[0056] Step S81: If a path conflict is detected, the interference results are simulated and analyzed to determine the spatial location and timestamp of the conflict point, and the distribution data of the conflict area is obtained.

[0057] Step S82: Adjust the movement path sequence based on the distribution data of the conflict area to generate a candidate path set;

[0058] Step S83: Obtain at least one path that meets the conflict-free condition from the candidate path set, compare the path length and simulate the execution time, and determine the preferred path;

[0059] Step S84: For the preferred path, obtain the current attitude control parameters of the device, determine whether they meet the stability requirements of path execution, and obtain the attitude adjustment requirements;

[0060] Step S85: Generate attitude control parameters based on attitude adjustment requirements and determine the final attitude planning of the device.

[0061] Step S86: Obtain execution instructions from the preferred path and final attitude planning; determine and correct deviations in instruction execution by real-time monitoring of equipment status; and obtain the final equipment movement path and attitude adjustment plan.

[0062] A second aspect of the present invention provides a parallel orientation control system for multiple welding equipment, employing the method described above to perform parallel orientation control on multiple welding equipment. The system includes: a multi-source sensor data acquisition module, used to acquire real-time three-dimensional spatial coordinate data and equipment attitude angle information from multiple welding equipment using multi-source sensors to obtain an initial positioning information set; a multi-source data fusion processing module, used to perform multi-source data fusion processing on the initial positioning information set to obtain fused positioning and attitude estimation results for the welding equipment; a relative spatial position relationship modeling module, used to construct a relative spatial position relationship model among the multiple equipment based on the fused positioning and attitude estimation results, and determine the coordination control parameters required for group control; and a dynamic filtering correction module, used to correct errors if the deviation between the fused positioning and attitude estimation results and a preset position threshold exceeds a certain threshold. If the estimated result exceeds the allowable range, it is corrected in real time to obtain the corrected precise equipment positioning information. A collaborative task allocation generation module is used to acquire the corrected precise equipment positioning information and the three-dimensional geometric model data of the target workpiece, generating a task allocation scheme for multi-equipment collaborative operations. A conflict-free path planning module is used to calculate the conflict-free movement trajectory of each operating device from its current corrected position to the target operating point based on the task allocation scheme, obtaining a preliminary movement path sequence. A motion interference simulation module is used to simulate potential motion interference scenarios between multiple devices based on the preliminary movement path sequence and equipment attitude adjustment specifications, determining whether path conflicts exist. A path iteration optimization module is used to iteratively optimize the movement path sequence based on the simulated interference results if path conflicts exist, obtaining the final conflict-free equipment movement path and attitude adjustment plan. The technical solution provided by this invention has the following beneficial effects:

[0063] This invention discloses a parallel orientation control method and system for multiple welding devices, addressing the challenges of insufficient positioning accuracy, path conflicts, and difficult coordinated control faced by multiple devices operating collaboratively in three-dimensional space in complex industrial scenarios. The invention acquires real-time three-dimensional coordinates and attitude information of the devices, employs a state estimation filtering algorithm for multi-source data fusion to obtain high-precision positioning and attitude estimation results, and constructs a model of the relative spatial positions between devices to generate coordinated control parameters. If the positioning deviation exceeds a threshold, a dynamic filtering algorithm is used for real-time correction, and a task allocation scheme is generated based on the geometric model of the target workpiece. The invention calculates conflict-free movement trajectories using a path planning algorithm and iteratively optimizes the path through simulated interference scenarios, ultimately achieving conflict-free movement path and attitude adjustment planning. This invention significantly improves the positioning accuracy and operational efficiency of multi-device collaborative operations, effectively avoids path conflicts, and ensures the stability and safety of coordinated control in complex scenarios. Attached Figure Description

[0064] Figure 1This is a flowchart of a parallel directional control method for multiple welding devices according to the present invention.

[0065] Figure 2 This is a schematic diagram of a parallel directional control method for multiple welding devices according to the present invention.

[0066] Figure 3 This is another schematic diagram of a parallel directional control method for multiple welding devices according to the present invention.

[0067] Figure 4 This is a schematic diagram of the structure of a parallel directional control system for multiple welding equipment according to the present invention. Detailed Implementation

[0068] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] like Figures 1-3 As shown, in a first aspect, the present invention provides a method for parallel directional control of multiple welding devices, the method specifically including:

[0070] S1 collects real-time three-dimensional spatial coordinate data and equipment attitude angle information from multiple operating devices through multi-source sensors to obtain an initial positioning information set.

[0071] Optionally, this step also includes:

[0072] Step S11: Collect the three-dimensional coordinates and attitude angles of multiple devices through multi-source sensors to generate an initial positioning dataset.

[0073] Step S12: The Kalman filter algorithm is used to smooth the three-dimensional coordinates and attitude angles in the initial positioning dataset to obtain a smoothed positioning dataset.

[0074] Step S13: Calculate the relative position of each device in three-dimensional space based on the smooth positioning dataset, and determine the spatial relationship between the devices.

[0075] Preferably, the relative position is calculated using the following Euclidean distance formula:

[0076]

[0077] in, It refers to the relative position of each device in three-dimensional space. These are the coordinates of the first device. These are the coordinates of the second device.

[0078] Step S14: Based on the spatial relationship between devices, use the least_squares function of SciPy to optimize and fuse the three-dimensional coordinates, where the sum of squared residuals is minimized to generate an optimized positioning dataset.

[0079] Step S15: Based on the optimized localization dataset, use the ARIMA model of statsmodels to fit the time series, calculate the residuals, and if the residuals exceed the preset residual threshold, they are detected as outliers and removed to obtain the outlier removal dataset.

[0080] Step S16: Based on the anomaly removal dataset, the particle filter algorithm is used to predict the future position and attitude of the device, where the particles represent the position and attitude states. The particle weights are updated by resampling to generate a predicted positioning dataset.

[0081] Step S17: By predicting the location dataset, the location points are connected using the plot function of matplotlib to construct the device's motion trajectory and determine the dynamic change trend of the device.

[0082] For example, in a smart factory, multiple welding robots collect 3D coordinates and attitude angles using multi-source sensors to form an initial positioning dataset. The sensors include LiDAR, inertial measurement units, and vision cameras. Robot 1's coordinates are (2.5, 3.0, 1.2) and attitude angles are (0.1, 0.2, 0.3 radians), while Robot 2's coordinates are (4.0, 5.0, 1.5) and attitude angles are (0.2, 0.1, 0.4 radians). The initial data may fluctuate due to noise, affecting positioning accuracy.

[0083] In one possible implementation, Kalman filtering is used for smoothing. Kalman filtering reduces the impact of noise by combining sensor data with a motion model through prediction and update steps.

[0084] For example, robot 1's coordinates are smoothed to (2.48, 3.02, 1.18), and its attitude angles are smoothed to (0.09, 0.19, 0.31 radians), generating a smoothed localization dataset. Smoothing improves data stability and contributes to the accuracy of subsequent calculations.

[0085] Specifically, the relative positions between devices are calculated based on a smoothed dataset. Using the Euclidean distance formula, the distance between Robot 1 and Robot 2 is... Meters. This spatial relationship clarifies the geometric distribution between devices, providing a foundation for optimized integration.

[0086] For example, SciPy's `least_squares` function is used to optimize fusion. It minimizes the sum of squared residuals and adjusts the coordinates to minimize the overall error.

[0087] For example, after optimization, robot one's coordinates are adjusted to (2.50, 3.01, 1.20), and robot two's coordinates are adjusted to (3.99, 4.98, 1.49). Optimizing the positioning dataset improves positioning accuracy and enhances the reliability of equipment collaboration.

[0088] In one embodiment, the ARIMA model fits and optimizes the time series of the dataset to predict the trend of coordinate changes.

[0089] For example, if the x-coordinate sequence of Robot 1 is [2.50, 2.51, 2.49...], and the residual after fitting exceeds the residual threshold of 0.1, it is marked as an anomaly and removed. The anomaly removal dataset reduces interference from erroneous data and improves prediction reliability.

[0090] Specifically, the particle filter algorithm predicts future position and orientation. Particles represent possible states; for example, the particle set of Robot 1 contains 1000 position and orientation combinations. By resampling and updating the weights, the predicted coordinates after 10 seconds are (2.60, 3.10, 1.25). This predicted localization dataset provides the basis for dynamic programming.

[0091] For example, matplotlib's `plot` function connects predicted location points to plot the trajectory of a welding robot. Robot One's trajectory is displayed as a smooth curve, reflecting its slow movement along the x-axis. Trajectory visualization helps analyze equipment dynamics, optimize path planning, and improve factory automation efficiency.

[0092] S2, the initial positioning information set is processed by multi-source data fusion using a state estimation filtering algorithm to obtain the fused position and attitude estimation results of the working equipment.

[0093] Optionally, this step also includes:

[0094] Step S21: Data is acquired from multiple sensors of the working equipment to form an initial information set containing position and attitude.

[0095] Step S22: The initial information set is cleaned using Python's Pandas library to obtain a standardized data set.

[0096] Step S23: If the integrity of the standardized dataset meets the preset integrity threshold, the Kalman filter algorithm is used to perform preliminary state estimation to obtain preliminary position and attitude estimation results.

[0097] Step S24: Based on the preliminary position and attitude estimation results, the particle filter algorithm is used to fuse multi-source data to obtain the optimized fusion result.

[0098] Step S25: If the error of the fusion result is lower than the preset error threshold, then determine the final position and attitude estimation result.

[0099] Step S26: Based on the final position and attitude estimation results, update the status information of the working equipment using NumPy array operations.

[0100] Step S27: Generate dynamic position and attitude data of the working equipment based on the status information.

[0101] For example, when acquiring data from operational equipment using multi-source sensors, information can be collected from sensors such as GPS, IMU, and LiDAR installed on the equipment. GPS provides the equipment's three-dimensional coordinates, the IMU provides attitude angles, and the LiDAR provides environmental distance information. This data forms the initial information set, containing raw data on position and attitude.

[0102] For example, the initial data of a welding robot may include coordinates (100, 200, 50) and attitude angles (30°, 45°, 0°).

[0103] It should be noted that the initial data may contain outliers due to sensor noise or environmental interference, such as coordinate shifts caused by GPS signal drift.

[0104] In one possible implementation, when cleaning data using the Pandas library, missing and outlier values ​​can be detected and handled.

[0105] For example, check if there are any empty coordinate values ​​in the dataset, or if the attitude angle is outside a reasonable range, such as exceeding 360°.

[0106] Specifically, missing values ​​can be filled with the average of the preceding and following data; outliers can be removed by setting an outlier threshold, such as considering coordinate changes exceeding 10 meters per second as outliers. After cleaning, a standardized dataset is obtained, ensuring data consistency and integrity.

[0107] For example, the cleaned dataset may contain 1,000 records, each including three-dimensional coordinates and attitude angles, with an integrity of over 95%.

[0108] For example, when using Kalman filtering for preliminary state estimation, its prediction-update mechanism can be used to smooth the data. Kalman filtering estimates the true state of the device by fusing sensor data and motion models.

[0109] For example, the coordinates of a welding robot may be smoothed from (100, 200, 50) to (99.8, 200.2, 50.1), and the attitude angle may be adjusted from (30°, 45°, 0°) to (29.8°, 45.2°, 0.1°).

[0110] It should be noted that this method can effectively reduce the impact of noise and improve the accuracy of state estimation.

[0111] In one possible implementation, when particle filtering is used for multi-source data fusion, the possible states of the device can be represented by a large number of particles.

[0112] For example, 1000 particles are generated, each containing a set of position and attitude assumptions. The particle weights are updated using sensor data, and the optimal state is finally selected. If the error of the fusion result is lower than a preset error threshold, such as a position error of less than 0.5 meters and an attitude error of less than 1°, the result is considered reliable.

[0113] For example, the position of the welding robot after fusion may be optimized to (99.9, 200.1, 50.0).

[0114] For example, the final position and orientation results can be used to update device status information through NumPy array operations.

[0115] Specifically, the optimized coordinates and orientation can be stored as NumPy arrays, and the device status table can be updated periodically.

[0116] For example, the status table of a welding robot may record its latest position as (99.9, 200.1, 50.0) and its posture as (29.9°, 45.1°, 0.0°).

[0117] In one possible implementation, when generating dynamic position and attitude data based on state information, the data can be serialized into a time series to record the movement trajectory of the equipment in the smart factory.

[0118] For example, the trajectory data of a welding robot over one hour can reflect its movement path from point A to point B, making it easier to monitor the equipment's operating status.

[0119] It should be noted that the above method forms a complete technical chain through multi-level processing, from data cleaning to state estimation and then to dynamic data generation.

[0120] For example, data cleaning ensures data quality, Kalman filtering and particle filtering improve estimation accuracy, and NumPy operations and dynamic data generation support real-time monitoring. These steps support each other, ensuring accurate and reliable position and orientation information for welding equipment, facilitating plant management optimization.

[0121] S3. Based on the fused position and attitude estimation results of the operating equipment, construct a relative spatial position relationship model among multiple devices and determine the coordination control parameters required for group control.

[0122] Optionally, this step also includes:

[0123] Step S31: Obtain position and attitude estimation data from multiple operating devices, and fuse them using a Kalman filter to obtain the fusion result.

[0124] Step S32: Based on the fusion results, calculate the relative spatial position relationship between the multiple devices using the Euclidean distance formula.

[0125] Step S33: Extract relative distance and angle information from the relative spatial position relationship, and generate a position relationship matrix using a NumPy array.

[0126] Step S34: If the distance between devices in the position relationship matrix is ​​less than a preset distance threshold, then calculate the attitude offset based on the distance and adjust the device attitude to obtain the optimized spatial distribution.

[0127] Step S35: Based on the optimized spatial distribution, use a particle filter to calculate the coordination parameters required for swarm control and generate a set of control parameters.

[0128] Step S36: By controlling the parameter set, motion commands are allocated to each device, and the A* algorithm is used to determine the coordinated motion trajectory of the group.

[0129] Step S37: Update the position and attitude data of each device according to the coordinated motion trajectory of the group, and obtain a new fusion result through a Kalman filter.

[0130] For example, in a smart factory, multiple robots need to work collaboratively to complete welding tasks, and acquiring the position and attitude data of each device is crucial. Position data can be obtained through a GPS module, while attitude data relies on an inertial measurement unit (IMU), which includes angular velocity and acceleration information. When the Kalman filter fuses this data, it comprehensively considers the GPS latitude and longitude data and the attitude angles of the IMU to generate a more accurate estimate of the device's state.

[0131] For example, a robot's GPS position is 120.5 degrees east longitude and 30.2 degrees north latitude, with a yaw angle of 30 degrees. The Kalman filter smooths sensor noise through prediction and update steps, and outputs the fused position and attitude.

[0132] Specifically, when calculating the relative spatial relationships between multiple devices based on the fusion results, the Euclidean distance formula can be used to determine the straight-line distance between devices.

[0133] For example, if the merged positions of two robots are (100, 200, 5) and (110, 210, 6), the distance between the two devices is calculated to be approximately 14.14 meters using the Euclidean distance formula. The relative angle can be calculated using the vector angle, yielding an azimuth difference of approximately 45 degrees. This information is stored in a position relationship matrix generated by a NumPy array, where the matrix elements represent the distance and angle between the devices.

[0134] For example, the matrix might show that the distance between Robot 3 and Robot 4 is 14.14 meters and the angle is 45 degrees, while the distance between Robot 4 and Robot 5 is 3.2 meters.

[0135] In one embodiment, if the position relationship matrix shows that the distance between two robots is less than a preset distance threshold of 1 meter, the robot's posture needs to be adjusted to avoid a collision. The posture deviation calculated based on distance can be analyzed by angle difference to adjust the robot's posture angle.

[0136] For example, robot three adjusts its attitude angle by 5 degrees to maintain a safe distance. The optimized spatial distribution reflects a reasonable layout between devices, and the particle filter further calculates the coordination parameters for swarm control.

[0137] For example, the motion trend of each device is evaluated by a particle filter to generate a set of control parameters, including instructions such as a speed of 0.5 m / s and a steering angle of 10 degrees.

[0138] For example, when A* algorithm is used to plan the coordinated movement trajectory of a group, it calculates the optimal path based on the factory terrain and equipment location.

[0139] For example, if a robot needs to navigate around obstacles to reach a target point (120, 220, 5), the A* algorithm generates a trajectory containing multiple path points, ensuring efficiency while avoiding obstacles. Based on this trajectory, the position and attitude data of each device are updated, and then fused again using a Kalman filter to obtain a new state estimate.

[0140] For example, robot three's new position might be updated to (115, 215, 5), with its attitude angle adjusted to 35 degrees. This method ensures the accuracy and coordination of multi-device collaborative operations, effectively improving the operational efficiency and safety of the smart factory.

[0141] S4. If the deviation between the fused position and attitude estimation results of the working equipment and the preset position threshold exceeds the allowable range, a dynamic filtering algorithm is used to perform real-time correction processing on the estimation results to obtain the corrected accurate equipment positioning information.

[0142] Optionally, this step also includes:

[0143] Step S41: If the deviation between the fused device position and attitude estimation results and the preset position threshold exceeds the allowable range, then sensor input data is acquired, and the estimation results are corrected in real time using the Kalman filter algorithm to obtain the corrected device positioning information.

[0144] Step S42: Based on the corrected device positioning information, use the NumPy library to calculate the weighted average and fuse multi-source sensor data, where the weights are based on the inverse variance of each sensor data to determine the error distribution of the fusion result.

[0145] Step S43: If the error distribution of the fusion result exceeds the preset accuracy threshold, the Kalman filter parameters are updated using the gradient descent optimizer. The process involves setting the learning rate to 0.01, calculating the gradient of the parameters with respect to the error, iteratively updating the parameter values, and obtaining the optimized filter model.

[0146] Step S44: Based on the optimized filtering model, reprocess the sensor input data to obtain updated device position and attitude estimation results.

[0147] Step S45: If the updated estimation result still deviates from the preset position threshold within the allowable range, then the historical positioning data is fitted by the least squares method, where the fitting equation is L=ax+b, x is the time series, L is the historical position value, a is the slope, and b is the intercept, and the correction compensation coefficients a and b are determined.

[0148] Step S46: Adjust the current equipment positioning information according to the correction compensation coefficients a and b to obtain the final accurate equipment positioning information.

[0149] Step S47: Update the historical records of device position and attitude by storing the final accurate device positioning information to the database.

[0150] For example, in multi-device collaborative operation scenarios, the estimated position and attitude of the devices may deviate beyond preset position thresholds due to sensor noise or environmental interference. For real-time correction using Kalman filtering, assume three operating devices, each equipped with GPS and IMU sensors. GPS provides position data, and the IMU provides attitude data. The initial estimated position deviation is 0.5 meters, and the attitude deviation is 3 degrees, exceeding the allowable ranges of 0.2 meters and 1 degree, respectively. Kalman filtering, through prediction and update steps, fuses the sensor input data to correct the position deviation to 0.15 meters and the attitude deviation to 0.8 degrees. The correction process utilizes a state transition matrix to predict device motion and combines sensor observations to update the estimate, reducing the impact of noise.

[0151] Specifically, based on the corrected positioning information, a weighted average is calculated using the NumPy library to fuse multi-source sensor data.

[0152] For example, the variances of the GPS data from the three devices are 0.1, 0.2, and 0.3, respectively, while the variances of the IMU data are 0.05, 0.08, and 0.1. The weights are calculated based on the reciprocal of the variances, with GPS weights at 1 / 0.1, 1 / 0.2, and 1 / 0.3, and IMU weights similarly. After fusion, the position error distribution is 0.12 meters, and the attitude error is 0.7 degrees, close to the accuracy requirements. This method improves data reliability and optimizes the fusion results through weighting.

[0153] In one embodiment, if the fusion result error still exceeds the accuracy threshold of 0.1 meters, the Kalman filter parameters can be adjusted using a gradient descent optimizer. A learning rate of 0.01 is set, and the parameters are iteratively updated based on the gradient of the error.

[0154] For example, excessively high initial covariance matrix parameters led to filtering instability. After 10 iterations, the parameters converged, and the error decreased to 0.09 meters. The optimized filtering model reprocessed the sensor data, reducing the position deviation to 0.1 meters and the attitude deviation to 0.5 degrees, meeting the requirements. This method dynamically adjusts parameters to improve filtering robustness.

[0155] For example, if the updated estimation results still do not meet the standards, historical positioning data can be fitted using the least squares method. Assuming the device's position data for the past 10 seconds is a time series, the fitting equation is L=ax+b, with a slope of a=0.02 and an intercept of b=1.5, yielding the correction compensation coefficients. After adjustment, the device's position deviation is reduced to 0.08 meters, and its attitude deviation to 0.4 degrees. This method utilizes historical data patterns to compensate for the deficiencies of real-time data.

[0156] Specifically, the final location information is stored in the database, and the device's historical records are updated.

[0157] For example, the database records the device coordinates and attitude every second, generating a time-series table for subsequent analysis. This storage method ensures data traceability and facilitates the optimization of group control strategies. Through the above methods, device positioning accuracy is significantly improved, ensuring the stability of multi-device collaborative operations.

[0158] S5, acquire the corrected precise equipment positioning information and the three-dimensional geometric model data of the target workpiece, and generate a task allocation scheme for multi-equipment collaborative operation.

[0159] Optionally, this step also includes:

[0160] Step S51: Obtain the corrected equipment positioning data and the three-dimensional geometric model of the target workpiece from the sensor. Use Python's Pandas library to parse the data, extract the equipment coordinate information and workpiece geometric features, and obtain the equipment spatial position and workpiece surface description.

[0161] Step S52: Based on the spatial location of the equipment and the surface description of the workpiece, the k-means clustering algorithm is used to group the equipment and determine the cooperative grouping of the equipment.

[0162] Step S53: Calculate task priorities based on workpiece processing requirements and equipment operating status using hierarchical clustering in the SciPy library to obtain task allocation results.

[0163] Step S54: Based on the task allocation results and the task execution sequence extracted from the task allocation results, generate a scheduling sequence for multi-device collaborative operation and determine the order of device operation.

[0164] Step S55: Based on the equipment operation sequence, use Apache Kafka to process real-time data streams and monitor the equipment's operating status. After determining that the operating status meets the workpiece processing requirements, use NumPy arrays to optimize the task execution timing to obtain the final collaborative operation plan. In one possible implementation, when acquiring calibrated equipment positioning data and the 3D geometric model of the target workpiece from sensors, multi-source sensors, such as LiDAR and inertial measurement units, can be used to collect the spatial coordinates of the equipment and the geometric features of the workpiece. When parsing this data using the Pandas library, the sensor data can be organized into a data frame, and the X, Y, and Z coordinates of the equipment, as well as the curvature and boundary point cloud data of the workpiece surface, can be extracted.

[0165] For example, the equipment coordinates might be (10.5, 20.3, 5.2), and the workpiece surface description might include key points with a radius of curvature of 0.8 meters. This method efficiently organizes multidimensional data, facilitating subsequent analysis.

[0166] Specifically, when using the k-means clustering algorithm to group devices, 10 devices can be divided into 3 collaborative groups based on their location coordinates and operating status.

[0167] For example, the equipment closest to the left side of the workpiece is grouped together, with its center point coordinates at (12.0, 18.0, 6.0). Grouping is based on the Euclidean distance between the equipment and their load capacity to ensure efficient collaboration. After grouping, each group of equipment can perform tasks more coordinatedly, reducing resource conflicts.

[0168] In one embodiment, when calculating task priorities using hierarchical clustering of the SciPy library, high-precision tasks can be prioritized based on the required precision of the workpiece processing and the real-time status of the equipment, such as prioritizing equipment with a processing speed of 2 meters per minute.

[0169] For example, if a workpiece surface requires a machining accuracy of 0.01 mm, then equipment with stable performance should be prioritized. This method ensures that task allocation meets machining requirements and improves job quality.

[0170] For example, when generating a scheduling sequence for multi-device collaborative operations, robot α can process the left side of the workpiece first, and robot β can process the right side subsequently, based on task priority and execution timing.

[0171] For example, robot α's operation time is 0 to 10 minutes, and robot β's is 12 to 20 minutes. This scheduling sequence optimizes cooperation between devices and avoids time overlap.

[0172] Specifically, when using Apache Kafka to process real-time data streams, you can monitor the operating status of devices, such as whether the device's vibration frequency is 50Hz and within the normal range. Kafka transmits status information in real time through data streams, ensuring timely feedback on abnormal states.

[0173] For example, if the temperature of the robot γ exceeds 80 degrees Celsius, the system will pause its task. This method ensures the stability of the processing.

[0174] In one embodiment, when using NumPy arrays to optimize task execution timing, the job time of each device can be stored as an array, and the task interval can be adjusted to reduce waiting time.

[0175] For example, the work interval between robots α and β can be optimized from 5 minutes to 3 minutes. This optimization can improve overall work efficiency and shorten the workpiece processing cycle.

[0176] Understandably, the above methods closely revolve around the business areas of equipment positioning and collaborative operations, echoing the positioning correction technology in historical information. The implementation examples at each stage, through data analysis, grouping, priority allocation, and scheduling optimization, form a logically rigorous solution that ensures efficient equipment collaboration while meeting workpiece processing requirements.

[0177] S6. For the task allocation scheme, a path planning algorithm is used to calculate the conflict-free movement trajectory of each working device from the current correction position to the target working point, and a preliminary movement path sequence is obtained.

[0178] Optionally, this step also includes:

[0179] Step S61: Obtain the calibration position coordinates of each working device and the coordinates of the target working point, and use the Global Positioning System to determine the position data.

[0180] Step S62: Based on the location data, the A-path planning algorithm is used to calculate the preliminary movement trajectory of each working device from the corrected position coordinates to the target working point, and obtain the preliminary path sequence.

[0181] Step S63: For the preliminary path sequence, use a geometric intersection calculation tool to detect whether there are intersections or overlaps between trajectories. If so, record the coordinates of the conflict points.

[0182] Step S64: Based on the coordinates of the conflict points, the Dijkstra algorithm is used to adjust the initial path sequence and generate a conflict-free trajectory.

[0183] Step S65: Extract the movement paths of each working device from the conflict-free trajectory to determine the final path sequence. Step S66: Generate movement instructions for each working device based on the final path sequence and output the instruction sequence.

[0184] Step S67: According to the instruction sequence, the real-time position of each working device is obtained using a Kalman filter. It is determined whether the device deviates from the final path sequence. If it does, the path is recalculated and adjusted using the A-path planning algorithm based on the real-time position and the final path sequence.

[0185] For example, in smart manufacturing scenarios, obtaining the calibrated position coordinates of each working device and the coordinates of the target work point is fundamental to achieving collaborative operation of multiple devices. The Global Positioning System (GPS) can obtain the precise coordinates of devices and work points through high-precision differential positioning technology. Suppose a factory has three mobile robots located at coordinates (10,20,0), (15,25,0), and (5,30,0) respectively, with the target work point at (50,50,0). These coordinates are corrected in real time by a GPS module, ensuring the error is within centimeters. This positioning data provides reliable input for subsequent path planning.

[0186] Specifically, when calculating the initial movement trajectory based on the A* path planning algorithm, the A* algorithm combined with a rasterized representation of the factory map can be used. The A* algorithm generates the shortest path from the starting coordinates to the target point by comprehensively considering the distance from the current position of the equipment to the target point and the obstacle situation through heuristic search.

[0187] For example, robot 1's initial path might be a straight line from (10,20,0) to (50,50,0), but it needs to bypass obstacles and generate a sequence of path points. This method ensures the efficiency of the path.

[0188] In one embodiment, when detecting trajectories intersecting or overlapping, a geometric intersection calculation tool can determine whether the paths intersect through vector analysis. For example, if the paths of robot 1 and robot 2 intersect at (30, 35, 0), the tool will record the coordinates of this conflict point. This detection provides crucial information for subsequent path optimization, avoiding the risk of collisions between devices.

[0189] For example, when adjusting paths using Dijkstra's algorithm, the path can be replanned based on conflict points.

[0190] For example, robot 1 bypasses (30,35,0) and selects a suboptimal path point (32,37,0) to generate a conflict-free trajectory. This adjustment ensures the safety and executability of the path sequence. After the final path sequence is extracted, each device obtains an independent movement path, such as robot 1's path points being (10,20,0), (20,30,0), (32,37,0), and (50,50,0).

[0191] Specifically, when generating movement instructions, the path sequence is converted into specific action instructions for the device.

[0192] For example, Robot 1's instructions include "move forward 10 meters, turn northeast, move 15 meters." These instructions are issued through the control system to ensure that the equipment executes them in sequence. The clarity of the instruction sequence improves the reliability of task execution.

[0193] In one embodiment, a Kalman filter is used for real-time position monitoring. Assuming robot 1 shifts position to (21,31,0) due to uneven ground during movement, the filter corrects the position deviation by fusing GPS data and inertial navigation data. This real-time monitoring ensures the device always stays close to the predetermined path.

[0194] For example, if a deviation is detected, the A* algorithm is recalculated and the path adjusted. Robot 1 generates a new path from (21,31,0) to (50,50,0) to avoid known obstacles. This dynamic adjustment improves the adaptability and stability of the operation.

[0195] Understandably, the above solution, through multi-level optimization—from positioning and path planning to real-time adjustment—ensures the efficiency and safety of multi-device collaborative operation. The implementation examples of each technical step support each other, jointly achieving seamless integration of equipment collaboration.

[0196] S7. Based on the preliminary movement path sequence and device attitude adjustment specifications, simulate potential motion interference scenarios between multiple devices and determine whether there are path conflicts.

[0197] Optionally, this step also includes:

[0198] Step S71: By parsing the movement path sequence, the spatial location data of each device in the time series is extracted to obtain the device movement trajectory.

[0199] Step S72: Obtain the spatial coordinates of each device at each time point from the device motion trajectory.

[0200] Step S73: The K-means clustering algorithm is used to group the device motion trajectories, the position feature vectors are extracted from the trajectory data, the number of clusters is initialized to the number of devices using the K-Means class of the sklearn library, the cluster labels are obtained by fitting the data, the intersection probability is calculated based on the trajectory overlap within the cluster, the spatial distribution characteristics of the trajectory are obtained, and the probability of trajectory intersection is determined.

[0201] Step S74: If the probability of trajectory intersection is greater than a preset probability threshold, then through spatial location analysis, the Euclidean distance formula is used to calculate the shortest distance between trajectories to determine potential conflict areas.

[0202] Preferably, the shortest distance between trajectories is calculated using the following formula:

[0203]

[0204] in, It is the shortest distance between two trajectories. These are the coordinates of the first trajectory point. These are the coordinates of the second trajectory point.

[0205] Step S75: For potential conflict areas, analyze the time overlap of the devices in the conflict areas by combining motion time series to determine whether there is a path conflict.

[0206] Step S76: Replan the device trajectories with path conflicts, use Algorithm A to search for alternative paths in the conflict area, and obtain the adjusted movement path sequence.

[0207] Step S77: Based on the adjusted movement path sequence, use MATLAB's Simulink tool to load the path sequence and run time-step simulation to verify whether the path conflict has been eliminated and obtain the final conflict judgment result.

[0208] For example, in a smart factory scenario, multiple welding robots need to move from different starting points to a designated welding position to complete the welding task. When analyzing the movement path sequence, the spatial coordinates of each robot in the time series can be recorded using timestamps to form a motion trajectory.

[0209] For example, robot A is located at (0,0,0) at t=0 seconds, moves to (1,2,0) at t=1 seconds, and so on, obtaining complete trajectory data. This data extraction is based on a high-precision positioning system, ensuring coordinate accuracy to the centimeter level, providing a reliable foundation for subsequent analysis.

[0210] In one possible implementation, the K-means clustering algorithm is used to group the trajectories. Assuming there are 5 robots in a smart factory, the initial number of clusters is 5, and clustering is performed based on the spatial feature vectors of the trajectories (such as the rate of change of coordinates and orientation angle).

[0211] For example, the trajectories of robot A and robot B are grouped into the same cluster because they are close to the same welding area. The high overlap of trajectories within the cluster indicates a potential risk of intersection. When calculating the probability of intersection, the spatial density of trajectory points within the cluster can be statistically analyzed to determine if it exceeds a preset probability threshold of 0.8. If it does, it indicates a high probability of trajectory intersection, requiring further analysis.

[0212] Specifically, for trajectories with a high probability of intersection, the Euclidean distance formula is used to calculate the shortest distance between trajectories.

[0213] For example, the trajectory point (2,3,0) of robot A is close to the trajectory point (2.1,3.2,0) of robot B, indicating a potential conflict area. This distance calculation is simple and efficient, and can quickly locate risk points. Combined with time series analysis, if both robots pass through this area simultaneously at t=2 seconds, then a time overlap is confirmed, and it is determined to be a path conflict.

[0214] For example, in conflict zones, the A* algorithm is used to replan the path. Suppose robot A's original path passes through channel one, which is prone to conflict with robot B. The A* algorithm can search for alternative paths, such as detouring to channel two, ensuring that the conflict zone is avoided. The adjusted path sequence must meet time constraints to avoid affecting overall welding efficiency.

[0215] One possible implementation uses MATLAB's Simulink tool for path verification. The adjusted path sequence is loaded, and a time step of 0.1 seconds is set to simulate the robot's movement.

[0216] For example, the lack of trajectory overlap between robots A and B in the simulation indicates that the conflict has been eliminated. This simulation verification is intuitive and efficient, enabling the early detection of potential problems and ensuring the reliability of path planning.

[0217] It should be noted that the above method focuses on optimizing equipment movement trajectories and is applicable to dynamic scheduling scenarios. Through clustering, distance calculation, and simulation verification, the solution forms a complete closed loop from core conflict detection to path adjustment, flexibly adapting to different work layouts and improving equipment coordination efficiency.

[0218] S8. If path conflicts exist, the movement path sequence is iteratively optimized based on the simulated interference results to obtain the final conflict-free device movement path and attitude adjustment plan.

[0219] Optionally, this step also includes:

[0220] Step S81: If a path conflict is detected, the interference results are analyzed through Monte Carlo simulation to determine the spatial location and timestamp of the conflict point, and the distribution data of the conflict area is obtained.

[0221] Step S82: Based on the distribution data of the conflict area, adjust the movement path sequence using the A* algorithm to generate a candidate path set.

[0222] Step S83: Obtain at least one path that meets the conflict-free condition from the candidate path set, and determine the preferred path by calculating the Euclidean distance, comparing the path length and simulating the execution time.

[0223] Step S84: For the preferred path, obtain the current attitude control parameters of the device, determine whether they meet the stability requirements of path execution, and obtain the attitude adjustment requirements.

[0224] Step S85: Based on the attitude adjustment requirements, use a proportional-integral-derivative controller to generate attitude control parameters and determine the final attitude plan of the device.

[0225] Step S86: Obtain execution instructions from the preferred path and final attitude planning; determine and correct deviations in instruction execution by real-time monitoring of equipment status; and obtain the final equipment movement path and attitude adjustment plan.

[0226] For example, in multi-device cooperative movement scenarios, after detecting path conflicts, the interference results can be analyzed using Monte Carlo simulation. Monte Carlo simulation generates a large number of possible device motion trajectories through random sampling, and combines them with time series to determine the location and time of the conflict point.

[0227] For example, assuming two devices may collide at spatial coordinates (10,20,5) and (12,21,6), after 1000 simulations, it was found that 80% of the trajectories overlapped around time t=5 seconds, yielding the probability distribution of the collision area. This helps in accurately locating high-risk areas.

[0228] Specifically, based on conflict area distribution data, the A* algorithm can be used to adjust routes. The A* algorithm generates a set of candidate routes through heuristic search, combining path length and conflict avoidance priority.

[0229] For example, in a 10m × 10m two-dimensional plane, the algorithm plans a path for device A from (0,0) to (10,10), avoiding the trajectory (5,5) of device B. Candidate paths may include path 1, which moves along the boundary, and path 2, which bypasses the obstacle. The shorter path is selected by calculating the path lengths to be 14m and 16m respectively.

[0230] In one embodiment, the preferred path needs to be verified for stability in conjunction with the device's attitude control parameters. Assume device A needs to maintain a horizontal attitude, with a current pitch angle of 5 degrees, but the preferred path requires a pitch angle within 3 degrees. By analyzing the path curvature and velocity, it is determined whether the stability requirements are met. If not, the attitude needs to be adjusted.

[0231] For example, a proportional-integral-derivative (PID) controller can be used to generate attitude control parameters. Based on the target pitch angle of 3 degrees and the current deviation, the controller dynamically adjusts the motor output, gradually correcting the attitude to the target value.

[0232] For example, device A completes the pitch angle adjustment within 2 seconds and maintains stable movement.

[0233] Specifically, after the execution command is generated, the device status is monitored in real time to correct any deviations.

[0234] For example, when device A moves along the preferred path, the sensor detects a lateral deviation of 0.5 meters and adjusts its direction through feedback control to ensure the accuracy of path execution. This method can effectively reduce the risk of collisions.

[0235] In one embodiment, the final path and attitude planning need to be comprehensively verified.

[0236] For example, in the simulation, the adjusted paths of devices A and B showed no overlap, and their attitude parameters met stability requirements. Real-time monitoring data indicated that the deviation was controlled within 0.1 meters, ensuring the reliability of the coordinated movement of multiple devices.

[0237] like Figure 4 As shown, in a second aspect, the present invention provides a parallel orientation control system for multiple welding equipment, which uses the method described above to perform parallel orientation control on multiple welding equipment. The system mainly includes: a multi-source sensor data acquisition module, used to acquire real-time three-dimensional spatial coordinate data and equipment attitude angle information from multiple working equipment through multi-source sensors to obtain an initial positioning information set; and a multi-source data fusion processing module, used to perform multi-source data fusion processing on the initial positioning information set using a state estimation filtering algorithm to obtain the fused position and attitude estimation results of the working equipment. The relative spatial position relationship modeling module is used to construct a relative spatial position relationship model among multiple devices based on the fused position and attitude estimation results of the operating equipment, and to determine the coordination control parameters required for group control. The dynamic filtering and correction module is used to perform real-time correction processing on the estimation results by using a dynamic filtering algorithm if the deviation between the fused position and attitude estimation results of the working equipment and the preset position threshold exceeds the allowable range, so as to obtain the corrected accurate equipment positioning information; the collaborative task allocation generation module is used to acquire the corrected accurate equipment positioning information and the three-dimensional geometric model data of the target workpiece, and generate a task allocation scheme for multi-equipment collaborative operation; the conflict-free path planning module is used to calculate the conflict-free movement trajectory of each working equipment from the current corrected position to the target working point using a path planning algorithm for the task allocation scheme, so as to obtain a preliminary movement path sequence; The motion interference simulation module is used to simulate potential motion interference scenarios between multiple devices based on the preliminary movement path sequence and device attitude adjustment specifications, and to determine whether there are path conflicts. The path iteration optimization module is used to iteratively optimize the movement path sequence based on the simulation interference results if path conflicts exist, so as to obtain the final conflict-free device movement path and attitude adjustment plan.

[0238] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for parallel orientation control of a plurality of welding apparatuses, characterized by, The method comprises: S1, collecting real-time three-dimensional spatial coordinate data and device attitude angle information from multiple working devices through multi-source sensors to obtain an initial positioning information set; S2, performing multi-source data fusion processing on the initial positioning information set to obtain a fused working device position and attitude estimation result; S3, constructing a relative spatial position relationship model between multiple devices according to the fused working device position and attitude estimation result, and determining a coordinated control parameter required for group control; S4, if a deviation of the fused working device position and attitude estimation result from a preset position threshold exceeds an allowable range, performing real-time correction processing on the estimation result to obtain corrected accurate device positioning information; S5, obtaining the corrected accurate device positioning information and three-dimensional geometric model data of a target workpiece, and generating a task allocation scheme for multi-device collaborative work; S6, calculating a conflict-free movement trajectory of each working device from a current corrected position to a target work point for the task allocation scheme to obtain a preliminary movement path sequence; S7, simulating a potential motion interference scenario between multiple devices according to the preliminary movement path sequence and a device attitude adjustment specification, and judging whether there is a path conflict situation; S8, if there is a path conflict situation, iteratively optimizing the movement path sequence according to the simulated interference result to obtain a final conflict-free device movement path and attitude adjustment plan.

2. The method of claim 1, wherein the plurality of welding devices are controlled in parallel in a direction of the welding device with the highest temperature. The step S1 comprises: Step S11, collecting three-dimensional coordinates and attitude angles of multiple devices through multi-source sensors to generate an initial positioning data set; Step S12, performing smoothing processing on the three-dimensional coordinates and attitude angles in the initial positioning data set to obtain a smoothed positioning data set; Step S13, calculating a relative position of each device in a three-dimensional space according to the smoothed positioning data set, and determining a spatial relationship between devices; Step S14, performing optimization fusion on the three-dimensional coordinates through the spatial relationship between devices to generate an optimized positioning data set; Step S15, fitting a time sequence according to the optimized positioning data set, calculating a residual error, detecting an abnormal point and removing it if the residual error exceeds a preset residual error threshold, and obtaining an abnormal removal data set; Step S16, predicting a future position and attitude of a device according to the abnormal removal data set by using a particle filtering algorithm, wherein a particle represents a position and attitude state, and a particle weight is updated through resampling to generate a predicted positioning data set; Step S17, connecting position points through the predicted positioning data set to construct a device motion trajectory and determine a device dynamic change trend.

3. The method of claim 2, wherein the plurality of welding devices are controlled in parallel in a direction of the welding device with the highest priority. The step S13 comprises: The relative position is calculated through the following Euclidean distance formula: wherein, is the relative position of each device in three-dimensional space, is the coordinate of the first device, is the coordinate of the second device.

4. The method of claim 1, wherein, The step S2 comprises: Step S21, obtaining data from multiple sensors of the working device to form an initial information set containing position and attitude; Step S22, cleaning the initial information set to obtain a standardized data set; Step S23, if the integrity of the standardized data set meets a preset integrity threshold, performing preliminary state estimation to obtain preliminary position and attitude estimation results; Step S24, performing multi-source data fusion according to the preliminary position and attitude estimation results to obtain an optimized fusion result; Step S25, if the error of the fusion result is lower than a preset error threshold, determining the final position and attitude estimation results; Step S26, updating the state information of the working device according to the final position and attitude estimation results; Step S27, generating dynamic position and attitude data of the working device according to the state information.

5. The method of claim 1, wherein, The step S3, according to the fused working device position and attitude estimation results, a relative spatial position relationship model between multiple devices is constructed, and coordination control parameters required for group control are determined, including: Step S31, obtaining position estimation and attitude estimation data of multiple working devices, and fusing to obtain a fusion result through a Kalman filter; Step S32, calculating the relative spatial position relationship between multiple devices according to the fusion result; Step S33, extracting relative distance and angle information from the relative spatial position relationship to generate a position relationship matrix; Step S34, if the distance between devices in the position relationship matrix is less than a preset distance threshold, calculating the attitude offset based on the distance and adjusting the device attitude to obtain an optimized spatial distribution; Step S35, calculating the coordination parameters required for group control according to the optimized spatial distribution to generate a control parameter set; Step S36, distributing the motion instructions of each device through the control parameter set to determine the group coordinated motion trajectory; Step S37, updating the position and attitude data of each device according to the group coordinated motion trajectory to obtain a new fusion result through a Kalman filter.

6. The method of claim 1, wherein, The step S5, obtaining the corrected accurate device positioning information and the three-dimensional geometric model data of the target workpiece to generate a task allocation scheme for multi-device collaborative work, including: Step S51, obtaining corrected device positioning data and three-dimensional geometric model of the target workpiece from the sensor, analyzing the data, extracting device coordinate information and workpiece geometric features, and obtaining device spatial position and workpiece surface description; Step S52, grouping the devices according to the device spatial position and workpiece surface description to determine the device collaboration group; Step S53, calculating the task priority according to the workpiece processing requirements and the device running state to obtain a task allocation result; Step S54, generating a scheduling sequence for multi-device collaborative work according to the task allocation result and the task execution time sequence extracted from the task allocation result to determine the device operation sequence; Step S55, according to the device operation sequence, processing real-time data stream to monitor the device running state, and after determining that the running state meets the workpiece processing requirements, optimizing the task execution time sequence to obtain a final collaborative work scheme.

7. The method of claim 1, wherein the plurality of welding devices are controlled in parallel in a direction of the welding device with the welding device being controlled in parallel in a direction of the welding device. The step S7 simulates a potential motion interference scenario among the devices according to the preliminary movement path sequence and the device posture adjustment specification, and judges whether there is a path conflict, including: Step S71, by analyzing the movement path sequence, extracting the spatial position data of each device in the time sequence, obtaining the device motion trajectory; Step S72, obtaining the spatial coordinates of each device at each time point from the device motion trajectory; Step S73, grouping the device motion trajectory, extracting the position feature vector from the trajectory data, initializing the cluster number using the K-Means class of the sklearn library, fitting the data to obtain the cluster label, calculating the cross possibility according to the trajectory overlap degree in the cluster, obtaining the spatial distribution feature of the trajectory, and judging the possibility of trajectory intersection; Step S74, if the possibility of trajectory intersection is greater than the preset possibility threshold, calculating the nearest distance between the trajectories through spatial position analysis to determine the potential conflict area; Step S75, for the potential conflict area, combining the motion time sequence, analyzing the time overlap of the devices in the conflict area, and judging whether there is a path conflict; Step S76, re-planning the device trajectory with path conflict, searching for an alternative path in the conflict area, and obtaining an adjusted movement path sequence; Step S77, according to the adjusted movement path sequence, running time step simulation to verify whether the path conflict is eliminated, and obtaining the final conflict judgment result.

8. The method of claim 7, wherein the plurality of welding apparatuses are controlled in parallel in a direction of the welding apparatuses. The step S74, if the possibility of trajectory intersection is greater than the preset possibility threshold, calculating the nearest distance between the trajectories through spatial position analysis to determine the potential conflict area, including: The nearest distance between the trajectories is calculated using the following formula: wherein, is the closest distance between the two trajectories, is the coordinate of the first trajectory point, is the coordinate of the second trajectory point.

9. The method of claim 1, wherein, The step S8, if there is a path conflict, iteratively optimizing the movement path sequence according to the simulation interference result to obtain the final device movement path and posture adjustment planning, including: Step S81, if a path conflict is detected, simulating and analyzing the interference result to determine the spatial position and timestamp of the conflict point, and obtaining the distribution data of the conflict area; Step S82, adjusting the movement path sequence according to the distribution data of the conflict area to generate a candidate path set; Step S83, obtaining at least one path that meets the conflict-free condition from the candidate path set, comparing the path length and the simulation evaluation execution time to determine the preferred path; Step S84, for the preferred path, obtaining the current posture control parameter of the device, judging whether it meets the stability requirement of path execution, and obtaining the posture adjustment requirement; Step S85, generating the posture control parameter according to the posture adjustment requirement to determine the final posture planning of the device; Step S86, obtaining the execution instruction from the preferred path and the final posture planning, judging the deviation of the instruction execution through real-time monitoring of the device state and correcting it to obtain the final device movement path and posture adjustment planning.

10. A multi-welding apparatus parallel orientation control system characterized by, The method according to any one of claims 1-9 is used for parallel directional control of multiple welding equipment, and the system comprises: a multi-source sensor data acquisition module for acquiring real-time three-dimensional spatial coordinate data and equipment attitude angle information from multiple working equipment through multi-source sensors to obtain an initial positioning information set; a multi-source data fusion processing module for performing multi-source data fusion processing on the initial positioning information set to obtain fused working equipment position and attitude estimation results; a relative spatial position relationship modeling module for constructing a relative spatial position relationship model among multiple equipment according to the fused working equipment position and attitude estimation results, and determining coordination control parameters required for group control; a dynamic filtering correction module for performing real-time correction processing on the estimation results if a deviation of the fused working equipment position and attitude estimation results from a preset position threshold exceeds an allowable range, to obtain corrected accurate equipment positioning information; a collaborative task allocation generation module for obtaining the corrected accurate equipment positioning information and three-dimensional geometric model data of a target workpiece, and generating a task allocation scheme for multiple equipment collaborative work; a collision-free path planning module for calculating collision-free moving trajectories of each working equipment from a current corrected position to a target working point for the task allocation scheme, to obtain a preliminary moving path sequence; a motion interference simulation module for simulating potential motion interference scenarios among multiple equipment according to the preliminary moving path sequence and equipment attitude adjustment specifications, and judging whether there is a path conflict situation; and a path iterative optimization module for iteratively optimizing the moving path sequence according to the simulated interference results if there is a path conflict situation, to obtain a final collision-free equipment moving path and attitude adjustment plan.

Images