A power tower phase line welding trajectory real-time compensation method and system

Abstract

The present application relates to the technical field of trajectory compensation, and more particularly, to a power tower intersecting line welding trajectory real-time compensation method and system, comprising: acquiring actual point cloud of an intersecting line region containing actual topography, and three-dimensional target point cloud of an intersecting line theoretical model; based on any data point in the actual point cloud, acquiring a local point array thereof, and based on the local point array, obtaining a normal vector and three characteristic values arranged from small to large. The present application accurately quantifies and strips the feature difference caused by abnormal topography by constructing local normal disorder degree and topological pseudo-state mutation rate, and then embeds the anti-noise confidence penalty weight into the registration algorithm, effectively avoiding the false traction of noise data on spatial registration, reducing the phenomena of miswelding and missed welding in the actual welding process, and greatly improving the trajectory compensation accuracy and connection quality of robot automatic welding.

Description

Technical Field

[0001] This invention relates to the field of trajectory compensation technology. More specifically, this invention relates to a method and system for real-time compensation of welding trajectories along intersecting lines of power transmission towers. Background Technology

[0002] With the continuous development of industrial automation, robotic automatic welding technology is frequently applied in the manufacturing of various large steel structures. To improve the automation level and precision of welding, a line laser vision sensor is usually installed at the end of the welding torch. This sensor scans the area to be welded to obtain three-dimensional point cloud data, and combines this data with computer algorithms to perform spatial registration between the theoretical model and the actual workpiece, thereby achieving automatic guidance and error compensation of the welding trajectory.

[0003] In the manufacturing of power transmission towers, there is a significant demand for welding at the intersections of numerous pipe components (such as the junctions of circular pipes). Because power transmission tower components are typically large and heavy, misalignment and macroscopic gaps inevitably occur during actual assembly. To achieve precise robotic welding, the system needs to acquire the point cloud of the intersection surface of the assembled pipe components and match it with the theoretical smoothing model from computer-aided design (CAD). The calculated spatial error is then sent to the robot in real time to drive the welding torch for dynamic trajectory compensation.

[0004] However, in actual power tower pipe assembly sites, workers often perform temporary spot welding around the bevel to achieve initial positioning of the pipes, and cutting burrs are easily left on the edges of the pipes during material cutting. These spot welds and burrs cause the surface of the actual welding area to exhibit extremely irregular three-dimensional lumps. If existing point cloud matching algorithms are used directly for trajectory calculation, the system will treat these randomly distributed burrs and spot welds as normal pipe wall data and forcibly compare them. This approach causes the algorithm to be erroneously guided by abnormal protrusions, resulting in a severely distorted spatial transformation matrix. This not only fails to accurately quantify the actual assembly misalignment gap, but also causes the welding trajectory ultimately compensated by the robot to deviate significantly from the actual intersection line joint, leading to severe off-center welding and missed welding during actual welding, thereby endangering the overall structural safety and connection strength of the power tower. Summary of the Invention

[0005] This invention provides a real-time compensation method and system for welding trajectories at intersections of power transmission towers. It aims to solve the problem in related technologies where existing point cloud matching algorithms are directly used for trajectory calculation. In such cases, the system treats randomly distributed contaminant data such as burrs and spot welds as normal pipe walls and performs forced comparisons, resulting in severe distortion of the calculated spatial transformation matrix. Consequently, the welding trajectory ultimately compensated by the robot deviates significantly from the actual intersection joint.

[0006] In a first aspect, the present invention provides a real-time compensation method for welding trajectories of intersecting lines of power transmission towers, comprising: acquiring an actual point cloud containing the actual topography of the intersecting line region, and a three-dimensional target point cloud of the theoretical model of the intersecting line; acquiring a local point matrix based on any data point in the actual point cloud, and obtaining a normal vector and three eigenvalues ​​arranged in ascending order based on the local point matrix; calculating the local normal disorder of the data point to characterize the consistency of the local normal direction of the data point; using the product of the local normal disorder of the data point and the ratio of the minimum value of its eigenvalues ​​to the sum of all eigenvalues ​​as the topological pseudomorphic mutation rate of the data point; calculating the noise resistance confidence penalty weight of the data point, wherein the noise resistance confidence penalty weight is negatively correlated with the local normal disorder; registering the actual point cloud and the target point cloud based on an iterative nearest-point algorithm, wherein the registration process includes minimizing a weighted error function containing the penalty weight to obtain a pose transformation between the source point cloud and the target point cloud; and applying the pose transformation to the coordinates of the target point cloud to obtain the compensated welding trajectory. This method constructs local normal disorder and topological pseudo-state mutation rate to accurately distinguish between the real pipe wall and contamination interference points from both micro-geometry and macro-structure dimensions. Based on this, a noise-resistant confidence penalty weight is designed and deeply embedded into the weighted error function of the iterative nearest-point registration algorithm. This mechanism enables the system to automatically shield against noise interference when solving spatial pose transformations, accurately calculate actual assembly misalignments, and generate high-precision dynamic welding compensation trajectories in real time. This fundamentally eliminates the risks of off-center welding and incomplete welding by the robot, greatly improving the quality and structural safety of automated welding.

[0007] Further, calculating the local normal disorder of the data point includes: performing a dot product operation on the normal vector of the data point and the normal vectors of its neighboring points within its local lattice to obtain the dot product value; and constructing the local normal disorder of the data point based on the sum of the squares of the complements of the dot product values. By measuring the consistency of the local normal direction through the dot product operation to construct the local normal disorder, the microscopic irregularities on the surface of pipe fittings caused by temporary spot welds and cutting burrs in actual assembly sites can be accurately and quantitatively characterized. In actual industrial scanning scenarios with significant noise, this provides a robust and reliable mathematical evaluation standard for effectively identifying and removing abnormally protruding interference areas.

[0008] Furthermore, the calculation of the topological pseudomorphic mutation rate of data points needs to satisfy the following relationship: In the formula, Representing data points The topological pseudo-state mutation rate; For data points The local normal disorder; These represent the three eigenvalues ​​of the local lattice covariance matrix, from smallest to largest. By combining the eigenvalues ​​representing the microscopic normal divergence with those representing the macroscopic deviation, a product operation is performed. This results in a geometrically amplified topological pseudostate mutation rate when dealing with three-dimensional protruding areas such as weld spots on-site. This mathematically achieves a significant step-like distinction between on-site contamination data and actual pipe wall geometry data, improving the system's anti-interference identification capability when dealing with surface defects in complex components.

[0009] Furthermore, obtaining the normal vector based on the local point matrix includes: calculating the covariance matrix of the local point matrix; performing singular value decomposition on the covariance matrix to obtain multiple eigenvalues ​​and corresponding eigenvectors, wherein the eigenvector associated with the smallest eigenvalue is the normal vector of that data point. Singular value decomposition of the covariance matrix of the local point matrix based on principal component analysis can quickly and accurately extract the principal direction of the three-dimensional distribution and its dispersion in the local region. In the scanning and calculation of complex curved surface structures of pipe fittings, it efficiently captures stable normal vectors and thickness features reflecting the actual assembly state of the component, laying a solid foundation for subsequent accurate calculation of morphological abrupt change rates.

[0010] Furthermore, the weighted error function The following relationship must be satisfied: In the formula, This represents the total number of point pairs participating in the matching; This represents the data points in the actual point cloud dataset; Data points representing the target point cloud in the theoretical intersection point set; For data points The noise resistance confidence penalty weight; Let be the three-dimensional rotation matrix to be solved; Let be the three-dimensional spatial translation vector to be solved. By dynamically introducing noise-resistant confidence penalty weights into the error function, the algorithm can make differentiated adjustments for the reliability of each matching point pair when evaluating the spatial pose transformation between the theoretical model and the actual acquired data. In the automated welding torch positioning and matching process, the error influence caused by abnormal topography points is automatically shielded, ensuring the accuracy and robustness of the final solved rigid body transformation matrix.

[0011] Further, the noise resistance confidence penalty weight of the data point is calculated, including: calculating the hyperbolic sine function value of the topological pseudomorphic mutation rate, and using the reciprocal of the hyperbolic sine function value as the noise resistance confidence penalty weight of the data point. By utilizing the characteristic of the hyperbolic sine function to exhibit exponential growth when the independent variable is large, the reciprocal of the weight is calculated, allowing the noise weight falling within abnormal patches or burr regions to decay to zero at an extremely fast rate, while remaining stable in the actual pipe wall region. This achieves rapid and nonlinear suppression of the influence of noise points in actual complex component point cloud registration scenarios, greatly enhancing the noise resistance performance of the compensation algorithm.

[0012] Furthermore, obtaining the compensated welding trajectory includes performing pose matrix multiplication on the welding trajectory point set corresponding to the intersection line theoretical model to obtain the compensated welding trajectory. By directly using a transformation matrix containing accurate assembly misalignment information to perform multiplication on the welding trajectory corresponding to the theoretical model, a corrected compensation path can be quickly generated. When this path is sent to the robot motion controller to drive the welding torch, real-time dynamic compensation for spatial assembly errors is achieved, meeting the timeliness and precise control requirements for efficient online welding of large structural components.

[0013] Furthermore, obtaining the local point matrix of the data point includes: performing a neighborhood search on the data point using a three-dimensional spatial kd-tree data structure to obtain the surrounding area of ​​the data point. Each neighboring point constitutes a local point matrix. A three-dimensional kd-tree data structure is used to search for neighboring points in the collected actual point cloud data. This enables high-speed local geometric point matrix partitioning in large-scale, high-density scan datasets, significantly reducing the computational overhead of subsequent feature extraction processes. This meets the stringent timeliness requirements for real-time trajectory calculation in the automated on-site processing of large equipment such as power transmission towers.

[0014] Furthermore, acquiring the actual point cloud of the intersection line region containing the actual shape includes: scanning the intersection line region of the actually assembled pipe fitting using a line laser vision sensor to obtain the actual point cloud containing the true surface shape. By using a line laser vision sensor installed at the end effector of the robot to perform online scanning of the assembled intersection line region, initial 3D data containing the true surface shape and assembly misalignment can be captured non-contactly and with high fidelity in complex assembly environments where heavy workpieces such as large pipe fittings exist. This provides high-quality, comprehensive, and realistic environmental input for the entire error correction process.

[0015] In a second aspect, a real-time compensation system for the welding trajectory of intersecting lines of power transmission towers is also provided, comprising a processor and a memory, characterized in that the memory stores a computer program, and the processor executes the computer program to implement the real-time compensation method for the welding trajectory of intersecting lines of power transmission towers as described in any of the above claims.

[0016] Beneficial effects: By constructing local normal disorder and topological pseudo-state mutation rate, feature differences caused by abnormal morphology can be accurately quantified and removed. Then, the noise resistance confidence penalty weight is embedded in the registration algorithm, which effectively avoids the erroneous guidance of noise data on spatial registration. In complex application scenarios such as on-site assembly of power tower pipe fittings, it can accurately reflect the real assembly misalignment gap, reduce the phenomenon of off-center welding and missing welding in the actual welding process, and significantly improve the trajectory compensation accuracy and connection quality of robotic automated welding. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart illustrating a real-time welding trajectory compensation method according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the initial state comparison before compensation registration according to an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the precise alignment after weighted compensation registration according to an embodiment of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0019] like Figure 1 As shown, S101: Data acquisition and neighborhood feature extraction.

[0020] In this embodiment, multi-source point cloud data is first acquired and preprocessed. Specifically, the system imports a theoretical 3D model of the intersection line of power transmission towers from computer-aided design software. By discretizing the model surface, the error-free theoretical intersection line point set is extracted and denoted as the target point cloud. Simultaneously, a line laser vision sensor mounted on the end effector of the welding robot scans the intersection area of ​​the actually assembled pipe fittings, acquiring a real point cloud dataset containing the true surface morphology (including spot welds, burrs, and assembly misalignments), denoted as the source point cloud. .

[0021] To analyze the local geometric features of each data point in the source point cloud, the system uses a three-dimensional spatial method for any given data point. Tree( This data structure enables efficient neighborhood search, where the search radius can be set to... To obtain the area around that point Each neighboring point forms a local point matrix. In this embodiment, the neighboring points constitute a local point matrix. The range of values ​​for can be: .

[0022] Subsequently, principal component analysis was performed on the local lattice. This analysis was based on the principle that calculating the covariance matrix of the local lattice reveals its principal distribution direction in three-dimensional space. Singular value decomposition of the covariance matrix yielded three eigenvalues ​​arranged in ascending order. And the corresponding eigenvectors. Among them, the smallest eigenvalue... The associated eigenvector is the normal vector of the data point in that local region. These three eigenvalues ​​quantitatively describe the dispersion or variance of the local lattice along the three principal axes, providing basic data for the subsequent construction of the topological pseudomorphic mutation rate.

[0023] S102: Construct the local normal disorder of each data point.

[0024] In actual power tower assembly sites, the surfaces of areas such as spot welds and cutting burrs exhibit extreme microscopic irregularities, directly leading to severe divergence of their surface normal vectors in space. Conversely, even with macroscopic curvature, the local normal vectors of real pipe walls or bevel surfaces should maintain a high degree of consistency. To quantitatively describe this difference, this embodiment constructs a local normal disorder.

[0025] This index is constructed based on the principle of measuring the consistency of local normal directions by calculating the cosine of the angle between the normal vector of the center point and the normal vectors of all points in its neighborhood, i.e., the dot product. The calculation formula is as follows: In the formula, Representing data points The local normal disorder; This represents the total number of neighboring points in the local lattice. Indicates the current central data point The normal vector; Represents the first point within the local lattice The normal vector of each neighboring point; It is an exponential function of the natural constant.

[0026] As can be seen from the above formula, when data points When it is on a smooth, real pipe wall or bevel surface, its normal vector direction is highly consistent with that of its neighboring points, i.e. The dot product value approaches At this point, the summation term The value is extremely small, resulting in local normal disorder. The value is smaller. Conversely, when the data point... When located in abnormal areas such as spot weld spots or burrs, the neighborhood normal vector diverges in all directions in three-dimensional space, resulting in a dot product value. This significantly reduces or even causes negative values. This leads to a sharp increase in the summation term, and the exponential function's power becomes a large negative number, ultimately resulting in local normal disorder. As the value increases, the indicator thus precisely isolates the feature differences caused by abnormal morphology at the micro-geometric level.

[0027] S103: Construct the topological pseudo-mutation rate for each data point.

[0028] Understandably, relying solely on local normal disorder is insufficient to completely distinguish noise from real structures, especially when dealing with large gaps or macroscopic steps caused by assembly misalignment. At real pipe intersections, although the bevel point cloud has a certain curvature, its local three-dimensional point cloud still macroscopically conforms to a two-dimensional manifold surface. Weld spots, burrs, or misaligned edges, however, exhibit a three-dimensional cluster distribution that bulges outward or concave inward. To integrate macroscopic and microscopic features, this embodiment constructs a topological pseudomorphic abrupt change rate.

[0029] The basis for constructing this indicator is to characterize the divergence of microscopic normals. Multiplying this by the surface change rate, which characterizes the deviation of the macroscopic three-dimensional structure, amplifies and enhances the abnormal region. The surface change rate can be quantified using the eigenvalues ​​obtained in step S101. The calculation formula is constructed as follows: In the formula, Representing data points The topological pseudo-state mutation rate; For data points The local normal disorder; These represent the three eigenvalues ​​of the local lattice covariance matrix from smallest to largest.

[0030] In the above formula, the sum of eigenvalues This represents the overall scattering variance of the point set in space, while the minimum eigenvalue... This characterizes the degree of deviation of the point set along the surface normal, i.e., the thickness. For an ideal two-dimensional manifold surface, such as a normal pipe wall, its thickness along the normal is minimal, i.e. Approaching At this time, regardless of For any value, it makes Extremely low. However, when three-dimensional protrusions such as weld spots or burrs are present, the point cloud experiences thickness accumulation in the normal direction. Significantly increased. At this point, if the microscopic normals of this region are also in a divergent state, that is... If the independent variables are large, multiplying two positively increasing independent variables will lead to a larger dependent variable. It is magnified geometrically. This design creates a significant, step-like distinction between contamination data and actual pipe wall data on a mathematical level.

[0031] S104: Construct noise-resistant confidence penalty weights for each data point.

[0032] After obtaining the topological pseudo-state mutation rate that can significantly distinguish noise, it needs to be transformed into a rate that can be directly applied to the iterative nearest point (…). The weight parameters of the algorithm. These weights should have the following characteristics: for actual pipe wall points, the weights should be stable at a constant value. For noisy points, the weights should decay rapidly and non-linearly to [the desired level]. Therefore, this embodiment constructs a noise-resistant confidence penalty weight.

[0033] This indicator is constructed based on the property that the hyperbolic sine function exhibits exponential growth when the independent variable is large. Placing it in the denominator position enables rapid convergence of the weights. Its calculation formula is as follows: In the formula, Representing data points The noise resistance confidence penalty weight; For data points The topological pseudo-state mutation rate; This represents the hyperbolic sine function.

[0034] As can be seen from this formula, when the scanning point is the actual pipe wall region, The value is extremely small, and the value of the hyperbolic sine term is also close to... This makes the penalty weight Stable and maintained at Nearby. Once the point cloud falls into an abnormal patch or spike area, An increase in will lead to an increase in the denominator The term expands rapidly, which in turn increases the penalty weight. At an extremely high speed Convergence. For example, when from Increase to hour, From the agreement plummeted to about This effectively suppresses the influence of noise points.

[0035] S105: Improved iterative closest point algorithm and trajectory compensation implementation.

[0036] Traditional The algorithm solves for the optimal rigid body transformation matrix by minimizing the sum of squared Euclidean distances between all corresponding point pairs. In this embodiment, the noise-resistant confidence penalty weights generated in step S104 are embedded into... In the core objective function of the algorithm, a weighted iterative objective function has been reconstructed.

[0037] The weighted objective function is constructed by assigning a confidence weight to the error term of each point pair, enabling the algorithm to automatically ignore or severely downplay the errors caused by noise points during iterative optimization. Its weighted error function... The calculation formula is as follows: In the formula, This represents the total number of point pairs participating in the matching; This represents the data points in the actual point cloud dataset; Represents the target data points in the theoretical intersection point set; For data points The noise resistance confidence penalty weight; Let be the three-dimensional rotation matrix to be solved; Let be the three-dimensional spatial translation vector to be solved.

[0038] In the trajectory compensation implementation phase, the system iteratively solves the aforementioned weighted objective function until its value converges to a minimum. In each iteration, for points such as burrs... Approaching The data points, and the resulting matching error Will be multiplied by a near The weights, thus in the overall error The calculations are effectively masked. This allows the optimization process to automatically eliminate the erroneous influence of burrs and weld spots on spatial registration, ultimately resulting in a more accurate three-dimensional spatial rotation matrix. Translation vector in three-dimensional space It can accurately reflect the actual misalignment of pipe fittings during assembly.

[0039] Finally, based on the solved 3D rotation matrix and 3D translation vector, the transformation matrix is ​​obtained. This transformation matrix is ​​then used to perform pose matrix multiplication on the coordinates of the theoretical intersection point set to generate a compensated real-time welding path. This path is sent to the robot motion controller in real time, driving the welding torch to dynamically compensate for spatial errors caused by assembly misalignment, thereby achieving high-precision automated welding.

[0040] like Figure 2 As shown in the figure, two sets of point clouds appear simultaneously. It can be clearly observed that the point array of the actual surface data collected by the sensor on site has obvious rotation and positional offset relative to the point array of the theoretical intersection path generated by the computer. This represents the assembly error generated during the actual assembly of the power tower pipe fittings.

[0041] like Figure 3 As shown, although the actual point cloud after trajectory compensation still retains those protruding burrs, the core curve representing the main body of the pipe wall has achieved a high degree of overlap with the theoretical intersection line point cloud.

[0042] It has been demonstrated that the weighted calculation process effectively shields the errors caused by burrs. Even with irregular protrusions, the algorithm still accurately extracts the true assembly deviation values.

[0043] This invention also provides a real-time compensation system for the welding trajectory of intersecting lines of power transmission towers. The system includes a processor and a memory, the memory storing computer program instructions. When the processor executes the computer program instructions, it implements the real-time compensation method for the welding trajectory of intersecting lines of power transmission towers according to the first aspect of this invention.

[0044] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and therefore will not be described in detail here.

[0045] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this invention can be implemented using computer-readable / executable instructions stored or otherwise maintained on such a computer-readable medium.

Claims

1. A method for real-time compensation of welding trajectory along the intersection line of power transmission towers, characterized in that, include: Obtain the actual point cloud of the intersection line region containing the actual shape, and the three-dimensional target point cloud of the intersection line theoretical model; Based on any data point in the actual point cloud, its local point matrix is ​​obtained, and based on the local point matrix, the normal vector and three feature values ​​arranged from smallest to largest are obtained. Calculate the local normal disorder of the data point to characterize the consistency of the local normal direction of the data point; the product of the local normal disorder of the data point and the ratio of the minimum value of its eigenvalues ​​to the sum of all its eigenvalues ​​is used as the topological pseudomorphic mutation rate of the data point. The noise immunity confidence penalty weight of the data points is calculated, and the noise immunity confidence penalty weight is negatively correlated with the local normal disorder. The actual point cloud and the target point cloud are registered based on the iterative nearest point algorithm. The registration process includes minimizing the weighted error function containing the penalty weight to obtain the pose transformation between the source point cloud and the target point cloud. The pose transformation is applied to the coordinates of the target point cloud to obtain the compensated welding trajectory.

2. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, Calculating the local normal disorder of the data points includes: The normal vector of the data point is multiplied by the normal vectors of its neighboring points in its local lattice to obtain the dot product value. The local normal disorder of the data point is constructed based on the sum of the squares of the complements of the dot product value.

3. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, The calculation of the topological pseudomorphic mutation rate of data points needs to satisfy the following relationship: ； In the formula, Representing data points The topological pseudo-state mutation rate; For data points The local normal disorder; These represent the three eigenvalues ​​of the local lattice covariance matrix from smallest to largest.

4. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, The normal vector is obtained based on the local point matrix, including: Calculate the covariance matrix of the local point matrix; perform singular value decomposition on the covariance matrix to obtain multiple eigenvalues ​​and corresponding eigenvectors, wherein the eigenvector associated with the smallest eigenvalue is the normal vector of the data point.

5. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, The weighted error function The following relationship must be satisfied: ； In the formula, This represents the total number of point pairs participating in the matching; This represents the data points in the actual point cloud dataset; Data points representing the target point cloud in the theoretical intersection point set; For data points The noise resistance confidence penalty weight; Let be the three-dimensional rotation matrix to be solved; Let be the three-dimensional spatial translation vector to be solved.

6. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, Calculating the noise resilience confidence penalty weight for the data points includes: The hyperbolic sine function value of the topological pseudostate mutation rate is calculated, and the reciprocal of the hyperbolic sine function value is used as the noise resistance confidence penalty weight for that data point.

7. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, The compensated welding trajectory includes: The welding trajectory point set corresponding to the intersection line theoretical model is subjected to pose matrix multiplication to obtain the compensated welding trajectory.

8. The real-time compensation method for the welding trajectory of the intersection line of power transmission towers according to claim 1, characterized in that, Obtaining a local point matrix of data points includes: A three-dimensional kd-tree data structure is used to perform a neighborhood search on the data point to obtain the neighborhood of the data point. Each neighboring point forms a local lattice.

9. The real-time compensation method for welding trajectory of intersecting lines of power transmission towers according to claim 1, characterized in that, Obtain the actual point cloud of the intersection region containing the actual topography, including: A line laser vision sensor is used to scan the intersection area of ​​the assembled pipe fittings to obtain an actual point cloud containing the true surface morphology.

10. A real-time compensation system for welding trajectory of intersecting lines of power transmission towers, comprising a processor and a memory, characterized in that, The memory stores a computer program, and the processor executes the computer program to implement the real-time compensation method for the welding trajectory of the intersection line of the power tower as described in any one of claims 1-9.

Images