Method for precise closing and positioning of container ship super-large sections

Abstract

The application discloses a container ship super-large section precise closing positioning method and belongs to the technical field of shipbuilding. The method comprises the following steps: taking the closing seam edge point cloud of a reference section and a to-be-closed section as input, and adopting nonlinear dimension reduction to extract the closing seam feature manifold; calculating the external normal invariant and the tangential invariant of the manifold at multiple discrete sampling points to form the Riemann metric tensor of the six-degree-of-freedom pose deviation of the section; mapping the tensor into the elastic elongation distribution coefficient of multiple mooring cables on the mooring system on both sides of the dock; controlling the mooring winch to apply forced displacement to the section so that the section tends to the reference section along the geodesic line path; and in the moving process, the pose is measured in real time by fusing laser targets and inertial sensors and repeated iteration is carried out until the Hausdorff distance of the closing seam edge point cloud is less than the allowable value. The application realizes high-precision autonomous docking control of the hull section.

Description

Technical Field

[0001] This invention relates to the field of shipbuilding technology, specifically a method for precise assembly and positioning of ultra-large sections of container ships. Background Technology

[0002] With the trend towards larger and more modular container ship construction, assembling ultra-large sections weighing hundreds of tons on water requires millimeter-level precision in spatial positioning and attitude adjustment. Existing positioning technologies for assembling ultra-large sections typically use total stations or laser trackers to measure the coordinates of discrete target points at both ends of the closure joint. Deviations are calculated through three-dimensional geometric fitting, and construction workers then repeatedly adjust mooring cables or hydraulic jacking devices based on the deviation to gradually push the section towards the reference section. This process relies on manual experience and multiple trial and error. Given the high flexibility of ultra-large sections and the fact that the closure joint edge is a three-dimensional free-form surface, discrete point measurements are insufficient to characterize the continuous geometric features of the closure joint edge. This makes it impossible to obtain detailed information such as local curvature changes and torsion, resulting in uneven joint gaps, local interference, and the potential introduction of residual stress within the structure during forced assembly.

[0003] Meanwhile, existing deviation compensation control is mostly based on six-degree-of-freedom independent adjustment in a fixed coordinate system, without considering the spatial motion constraint characteristics of the main segment under multi-point cable constraints. The mapping relationship between the distribution of cable release and take-off and the spatial pose deviation of the main segment depends on simplified geometric approximation, which causes the actual motion trajectory of the main segment to deviate from the optimal path, resulting in low closing efficiency and easy repeated collisions.

[0004] The problem requires reconstructing the complete closure seam surface manifold from the sparse point cloud at the seam edge, and extracting invariants describing the local deformation and deviation directions on this manifold to establish a metric tensor that accurately reflects the intrinsic geometric structure of the six-degree-of-freedom deviation. It also requires mapping this geometric metric tensor directly to the elastic elongation distribution of multiple mooring cables, enabling the entire segment to move along the optimal geodesic path determined by the closure seam manifold, achieving precise closure under compliant constraints. Summary of the Invention

[0005] A method for precise positioning and closure of ultra-large sections of container ships is provided. Starting from the point cloud at the edge of the closure joint, a characteristic manifold of the closure joint is constructed and geometric invariants are extracted to form a Riemannian metric tensor of six-degree-of-freedom pose deviation. The tensor is mapped to the elastic elongation distribution coefficient of the mooring cable, and the mooring winch is driven to achieve compliant and precise closure of the section along the geodesic path.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a method for precise assembly and positioning of ultra-large sections of container ships, comprising: Using the point clouds of the closure seam edges of the reference segment and the segment to be closed as input, a nonlinear dimensionality reduction method is used to extract the closure seam feature manifold. As a preferred embodiment of the invention, for each point in the closure seam edge point cloud, a K-dimensional tree is used to search for its neighboring points, and the Euclidean distance between neighboring points is calculated. A weighted adjacency graph is established using the Euclidean distance as the edge weight, and the geodesic distance between any two points is calculated using the Dijkstra algorithm to form a geodesic distance matrix. A multidimensional scaling transformation is applied to the geodesic distance matrix to obtain low-dimensional embedded coordinates, thereby preserving the geodesic distance relationship between point pairs in the low-dimensional space. The points corresponding to the edge contour of the segment in the low-dimensional embedded coordinates are sorted according to curvature continuity. The sorted point set is the closure seam feature manifold, which significantly reduces the data dimensionality while preserving the essential geometric structure of the closure seam.

[0007] The external normal and tangential invariants of the closure seam feature manifold at multiple discrete sampling points are calculated, and a Riemannian metric tensor for the total six-degree-of-freedom pose deviation is constructed accordingly. As a preferred approach, multiple discrete sampling points are selected on the closure seam feature manifold at fixed arc length intervals. For each discrete sampling point, a local quadratic surface is fitted using its neighborhood points to obtain the two principal curvature directions of the local quadratic surface. These two principal curvature directions are projected onto the tangential and normal planes of the closure seam, respectively, to obtain the external normal and tangential invariants of the discrete sampling point. The external normal and tangential invariants of all discrete sampling points are sequentially combined to form a high-dimensional eigenvector, and principal component analysis is performed on this high-dimensional eigenvector to reduce its dimensionality, resulting in a compact eigenvector. The compact feature vector is dynamically time-warped and matched with the corresponding compact feature vector of the reference segment to obtain the corresponding point of each discrete sampling point on the reference segment. Based on the positional difference between the corresponding point and the discrete sampling point, the spatial offset of each discrete sampling point in the overall segment coordinate system is calculated and transformed to the centroid of the overall segment to form a six-dimensional deviation vector. The covariance matrix of the six-dimensional deviation vector is calculated, and the inverse of the covariance matrix is ​​taken as the Riemann metric tensor. This Riemann metric tensor can accurately reflect the coupling relationship and adjustment stiffness between the degrees of freedom of pose deviation.

[0008] The Riemann metric tensor is mapped to elastic elongation distribution coefficients for multiple mooring cables arranged on both sides of the dock. For each mooring cable, its mooring point coordinates on the main section, its fixed point coordinates on the dock mooring system, axial stiffness, and original length are obtained. The direction vector of the cable in the current attitude is calculated. A projection matrix is ​​constructed based on the direction vector. The Riemann metric tensor G is multiplied by the projection matrix to obtain the stiffness contribution matrix of the cable to the center of gravity of the main section. The stiffness contribution matrices of all mooring cables are summed to obtain the overall cable system stiffness matrix. The overall cable system stiffness matrix is ​​multiplied by the six-dimensional deviation vector to obtain the cable force vector used to eliminate the six-dimensional deviation vector. Each component of the cable force vector is divided by the axial stiffness of the corresponding mooring cable to obtain the elastic elongation distribution coefficient. This mapping transforms the abstract pose deviation into directly executable cable extension and retraction control quantities, providing an explicit solution basis for achieving multi-cable coordinated positioning.

[0009] Based on the elastic elongation distribution coefficient, the corresponding mooring winch is controlled to apply forced displacement to the overall section, causing the overall section to move towards the reference section along the geodesic path determined by the tangential invariant. The elastic elongation distribution coefficient of each mooring cable is converted into the drum angle increment of the corresponding mooring winch; a speed command is sent to the drive motor of the mooring winch according to the drum angle increment, and the actual angle is fed back by the encoder installed on the drum shaft to form a position closed loop; the speed command is corrected according to the real-time tension value fed back by the tension sensor installed on each mooring cable, so that the ratio between the tension of each mooring cable and the elastic elongation distribution coefficient remains constant. Under the combined action of the position closed loop and tension correction, the mooring winch synchronously winds up and unwinds the cables, driving the overall section to move smoothly along the shortest geodesic path, avoiding suboptimal swaying and additional torque.

[0010] During the movement, the overall segment's pose is measured in real time using a fusion of laser targets positioned on both sides of the closure joint and inertial sensors. At least three laser targets are deployed at the edges of the closure joint on both sides of the segment to be joined, each equipped with multiple optical prisms. An inertial sensor containing a three-axis gyroscope and a three-axis accelerometer is installed at the segment's center of gravity. The three-dimensional coordinates of each optical prism are measured using a total station as position observations. The attitude angles output by the inertial sensors are used as attitude observations. The position and attitude observations are input into a Kalman filter to obtain an estimated overall segment pose, which includes six degrees of freedom: sway, pitch, heave, roll, pitch, and bow. This fusion measurement effectively suppresses noise and drift from individual sensors, providing high-precision real-time pose feedback during dynamic adjustments.

[0011] Repeat the above steps until the Hausdorff distance of the closure seam edge point cloud is less than the allowable value. In each repetition, based on the current real-time measured total segment pose estimate, calculate the coordinate transformation matrix of the segment to be closed relative to the reference segment, and update the closure seam edge point cloud of the segment to be closed using this matrix. Calculate the Hausdorff distance between the updated closure seam edge point cloud and the closure seam edge point cloud of the reference segment, and determine if the Hausdorff distance is less than the allowable value. If not, regenerate the Riemann metric tensor based on the location of the maximum deviation point corresponding to the Hausdorff distance and proceed to the next iteration. If yes, lock all mooring cables between the segment to be closed and the reference segment, and insert temporary positioning pins along the circumference of the closure seam. Using the Hausdorff distance as a closure criterion allows for a comprehensive evaluation of the overall fit of the closure seam, ensuring final positioning accuracy.

[0012] The process of locking all mooring cables between the section to be joined and the reference section is as follows: apply a mechanical brake to the mooring winch of each mooring cable and put the drive motor of the mooring winch in a zero-torque state; insert a pre-positioning pin at preset angles along the circumference of the joining seam and measure the joining seam gap between adjacent pre-positioning pins; according to the distribution of the joining seam gap, add auxiliary positioning pins at positions where the gap is greater than the set value; tighten all temporary positioning pins in the order of first the middle and then the two ends along the length of the joining seam. When tightening each temporary positioning pin, immediately monitor the position of the whole section using an inertial sensor and a laser target; if the deviation of the currently monitored position of the whole section relative to the reference section exceeds the allowable range, stop tightening the current temporary positioning pin and control the mooring winch corresponding to the area where the current temporary positioning pin is located to make a reverse adjustment so that the deviation returns to the allowable range; after the deviation returns to the allowable range, continue tightening the current temporary positioning pin, and then perform the same operation on the next temporary positioning pin until all temporary positioning pins are tightened. This symmetrical progressive locking strategy effectively prevents local deformation and stress concentration caused by single-point locking, ensuring uniform fit and firm connection of the closure seam.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: The closure seam feature manifold is extracted using nonlinear dimensionality reduction, and the outward normal and tangential invariants at discrete sampling points are calculated to construct the Riemannian metric tensor. A weighted adjacency graph is constructed from the point cloud at the closure seam edge, and low-dimensional embeddings are obtained using geodesic distance and multidimensional scaling. The closure seam feature manifold is then sorted according to curvature continuity, preserving the intrinsic geometric structure of the edge surface. Local quadric surfaces are sampled at arc-length intervals on this manifold, and the principal curvature directions are projected onto the tangential and normal planes to obtain invariants. Principal component analysis is then used to obtain compact eigenvectors. Dynamic time warping is performed on the compact eigenvectors of the reference segment and the total segment to be closed to obtain the spatial offsets of each discrete sampling point, which are then transformed to the centroid of the total segment. The inverse of the covariance matrix is ​​constructed using these offsets as the Riemannian metric tensor. This tensor characterizes the coupling relationship between the local deformation direction and torsion degree of the closure seam and the spatial pose deviation. The direction of motion of the whole segment determined by this tensor naturally follows the shortest geodesic from the current state to the target state of the closure seam. During the closure process of the whole segment, the edge of the seam gradually comes into contact, avoiding edge collisions and repeated adjustments caused by unreasonable selection of motion direction and step size, thus improving the closure stability.

[0014] The Riemann metric tensor is mapped to the elastic elongation distribution coefficients of multiple mooring cables. The direction vector is obtained by acquiring the mooring point coordinates and fixed point coordinates of each mooring cable. A projection matrix is ​​constructed and multiplied by the Riemann metric tensor to obtain the stiffness contribution matrix of each cable to the center of gravity of the overall section. The summation of all stiffness contribution matrices is multiplied by the six-dimensional deviation vector to obtain the cable force vector, which is then divided by the axial stiffness of each cable to obtain the elastic elongation distribution coefficient. When controlling the mooring winch, the drum angle increment is converted according to the elastic elongation distribution coefficient, and the cables are wound up and down under the combined action of position closed-loop control and tension correction. This distribution method directly correlates the elongation of each cable with the local geometric deviation of the closure joint. The coordinated action of multiple cables automatically maintains a constant ratio between the cable tension and the deviation vector. The cable force system experienced by the overall section during the closure movement can smoothly adapt to the changes in the closure joint contour, without generating excessive local traction. After closure, the gap at the joint is uniform, and the additional stress on the structure is suppressed. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a flowchart of the method for precise assembly and positioning of ultra-large sections of container ships; Figure 2 This is a flowchart of point cloud feature extraction and dimensionality reduction processing at the edge of the closure seam; Figure 3This is a flowchart of the process for tightening the temporary positioning pins and adjusting the position of the main section during assembly; Figure 4 These are the curves of the external normal and tangential invariants of the discrete sampling points of the characteristic manifold of the closure seam; Figure 5 This is a principal component analysis distribution diagram of the compact eigenvectors of the baseline segment and the total segment to be merged; Figure 6 It is the Hausdorff distance iteration convergence curve of the point cloud at the edge of the closure seam. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0018] See Figure 1 This invention provides a method for precise positioning and closure of ultra-large sections of a container ship. Using the point clouds of the closure seam edges of a reference section and the section to be closed as input, a nonlinear dimensionality reduction method is employed to extract the characteristic manifold of the closure seam. The outward normal and tangential invariants of the closure seam characteristic manifold at multiple discrete sampling points are calculated, and a Riemann metric tensor for the six-degree-of-freedom pose deviation of the section is constructed accordingly. This Riemann metric tensor is mapped to the elastic elongation distribution coefficients of multiple mooring cables arranged on both sides of the dock. Based on the elastic elongation distribution coefficients, the corresponding mooring winches are controlled to apply forced displacement to the section, causing it to move towards the reference section along a geodesic path determined by the tangential invariants. During the movement, the pose of the section is measured in real time using laser targets and inertial sensors arranged on both sides of the closure seam, and this process is repeated until the Hausdorff distance of the closure seam edge point cloud is less than the allowable value.

[0019] In specific implementation, please refer to Figure 2 Each point in the point cloud at the edge of the closure seam is defined as a sample point. For each sample point, a K-dimensional tree is used to search for its neighboring points, and the Euclidean distance between the neighboring points is calculated. The number of neighboring points K in the K-dimensional tree search is set according to the point cloud density, so that the neighborhood range covers the local geometric features at the sample point; K is set to a value of 15.

[0020] A weighted adjacency graph is constructed using Euclidean distance as the edge weights. In this graph, if two sample points are not neighbors, the edge weight is set to infinity. Dijkstra's algorithm is used to calculate the shortest path length between any two sample points in the weighted adjacency graph; this shortest path length is the geodesic distance between the two sample points. The geodesic distances between all pairs of sample points are then combined to form a geodesic distance matrix.

[0021] By applying a multidimensional scaling transformation to the geodesic distance matrix, we can find the low-dimensional embedding coordinates that minimize the difference between the Euclidean and geodesic distances of points in the low-dimensional space. The multidimensional scaling transformation is achieved by minimizing the following stress function:

[0022] in, Represents sample points With sample points The geodesic distance between them comes from the first element in the geodesic distance matrix. Line 1 Column elements; and Representing sample points respectively and sample points Embedded coordinate vectors in low-dimensional space; Represents the Euclidean norm; stress function To measure the degree to which low-dimensional embedded coordinates preserve geodesic distance; This represents the difference between the geodesic distance and the Euclidean distance in the low-dimensional space. The dimension of the low-dimensional embedding space is set to 1 to extract the one-dimensional manifold structure at the edge of the closure joint. The stress function is solved iteratively using the gradient descent method. Minimized The set yields the one-dimensional low-dimensional embedding coordinates of all sample points.

[0023] After obtaining the low-dimensional embedding coordinates, the sample points corresponding to the edge contour of the total segment in the low-dimensional embedding coordinates are sorted according to the curvature continuity. Since the low-dimensional embedding coordinates are one-dimensional, the sample points are directly arranged in ascending order of coordinate values, and the positional continuity of adjacent sample points in the original point cloud is checked to remove jump outliers, forming a closure seam feature manifold. The closure seam feature manifold is represented by an ordered set of points.

[0024] Discrete sampling points are selected at fixed arc length intervals on the closure seam feature manifold. The fixed arc length interval is determined based on the total arc length of the closure seam profile and the required pose calculation accuracy, and is set to 200 mm. On the closure seam feature manifold, starting from the starting point, a point is selected every 200 mm along the arc length as a discrete sampling point.

[0025] For each discrete sampling point, search for the nearest neighboring points in the point cloud of the closure seam edge, and fit a local quadratic surface using these neighboring points. The equation of the local quadratic surface is as follows: ,in , , The coordinates are in a local coordinate system with the discrete sampling points as the origin. , , , , , The parameters to be fitted are denoted as . After solving for the parameters using the least squares method, the first principal curvature direction and the second principal curvature direction of the local quadratic surface at the discrete sampling points are calculated.

[0026] The first and second principal curvature directions are projected onto the tangent and normal planes of the closure joint, respectively. The tangent directions of the characteristic manifold of the closure joint at discrete sampling points are used as the normals of the tangent plane to construct the tangent and normal planes. The first principal curvature direction is projected onto the normal plane to obtain the external normal invariant; the second principal curvature direction is projected onto the tangent plane to obtain the tangential invariant. The external normal invariant reflects the bending characteristics of the closure joint along the normal direction at discrete sampling points, while the tangential invariant reflects the torsional characteristics along the tangential direction.

[0027] The outward normal and tangential invariants of all discrete sampling points are concatenated sequentially to form a high-dimensional eigenvector. Principal component analysis (PCA) is then performed on the high-dimensional eigenvector to reduce its dimensionality. During PCA, the covariance matrix of the high-dimensional eigenvector is calculated, and the eigenvalues ​​and eigenvectors are solved. Principal components are selected according to the eigenvalues ​​from largest to smallest, ensuring that the cumulative variance contribution rate of the first few principal components exceeds 95%. The high-dimensional eigenvector is then projected onto the selected principal component directions to obtain a compact eigenvector.

[0028] See Figure 4 The figure uses the arc length of the closure seam (in meters) as the horizontal axis to show the curve trends of the external normal invariant and tangential invariant at discrete sampling points on the characteristic manifold of the closure seam as a function of the arc length. The blue solid line to the left of the vertical axis represents the external normal invariant, and the orange dashed line to the right of the vertical axis represents the tangential invariant, both in units of 1 / meter.

[0029] The curve starts at an arc length of 0 meters and ends at approximately 100 meters. Both the external normal invariant and the tangential invariant exhibit obvious periodic fluctuations. The value of the external normal invariant fluctuates approximately between -0.03 and 0.03, while the value of the tangential invariant fluctuates approximately between -0.025 and 0.025. The two invariant curves show a phase difference, with the peaks of the external normal invariant corresponding to the troughs of the tangential invariant, and they alternate in local regions, reflecting the alternating distribution of bending and torsion characteristics of the closure seam edge in the normal and tangential directions.

[0030] The curve's trend indicates that the first and second principal curvature directions of the closure seam feature manifold, calculated through local quadratic surface fitting, effectively characterize the geometric shape changes at different locations along the closure seam edge using the outward normal and tangential invariants obtained through projection. This provides fundamental data support for the subsequent construction of high-dimensional and compact feature vectors. Furthermore, the figure demonstrates the specific effect of nonlinear dimensionality reduction in the above embodiment for extracting the closure seam feature manifold and sampling and calculating invariants. It shows the spatial curvature properties of the closure seam feature manifold along the arc length of the closure seam, satisfying the geometric feature description required for accurate closure and positioning of the closure seam.

[0031] In practice, dynamic time warping matching is performed on the compact feature vectors of the segment to be merged and the corresponding compact feature vectors of the reference segment. The dynamic time warping matching takes the sequences of compact feature vectors of the segment to be merged and the corresponding compact feature vector sequences of the reference segment as inputs. Euclidean distance is used as the local distance metric between the elements of the two sequences. A cumulative distance matrix is ​​constructed through dynamic programming, and the warping path that minimizes the cumulative distance is searched within the cumulative distance matrix. The warping path provides a one-to-one correspondence between each discrete sampling point in the compact feature vector of the segment to be merged and its corresponding point in the compact feature vector of the reference segment, thus obtaining the position of each discrete sampling point on the reference segment.

[0032] Based on the positional difference between each corresponding point and its corresponding discrete sampling point, the spatial offset of each discrete sampling point in the overall segment coordinate system is calculated. The spatial offset is a three-dimensional vector, containing offsets along the X, Y, and Z axes of the overall segment coordinate system. Based on the coordinates of each discrete sampling point in the overall segment coordinate system, the spatial offset at the discrete sampling point is equivalently transformed to the centroid of the overall segment using the force screw equivalence principle, resulting in the linear and angular displacements at the centroid of the overall segment. The three components of the linear displacement and the three components of the angular displacement are arranged sequentially to form a six-dimensional deviation vector.

[0033] The covariance matrix of the total six-degree-of-freedom pose deviations is calculated using a six-dimensional deviation vector. Multiplying the six-dimensional deviation vector by its transpose yields a 6x6 matrix, which is used as an estimate of the covariance matrix. The inverse of the covariance matrix is ​​then used as the Riemann metric tensor. The Riemann metric tensor is a positive definite symmetric matrix.

[0034] Obtain the coordinates of the mooring point on the main section, the coordinates of the anchor points on the mooring systems on both sides of the dock, the axial stiffness of the mooring line, and the original length of the mooring line. The axial stiffness of the mooring line is obtained by multiplying the elastic modulus of the mooring line material by the cross-sectional area of ​​the mooring line; the axial stiffness of the mooring line is the factory-calibrated value. The coordinates of the mooring point and the anchor points are obtained through three-dimensional measurements in the dock coordinate system.

[0035] According to the The mooring point coordinate vector of the root mooring cable and the first Calculate the coordinate vector of the anchor point of the root mooring cable. The direction vector of the root mooring cable in the current attitude. The direction vector points from the fixed point to the mooring point, and its calculation formula is:

[0036] in, This is a numbered index for the mooring cables. Take a positive integer, ranging from 1 to the total number of mooring cables; Indicates the first The direction vector of the root mooring cable; Indicates the first The coordinate vector of the mooring point on the main section of the root mooring cable; Indicates the first The coordinate vectors of the fixed points of the root mooring cables on the mooring systems on both sides of the dock; The Euclidean norm of a vector; This represents a vector pointing from a fixed point to a mooring point; This indicates the distance between the mooring point and the fixed point.

[0037] Construct the projection matrix for each mooring cable based on the direction vector. For the ... For the root mooring cable, the projection matrix is ​​a 1x6 row vector, with the first three elements of the projection matrix being the direction vectors. The three components, the last three elements of the projection matrix are taken as vectors. The three components, of which This represents the position vector of the mooring point relative to the center of gravity of the entire segment. This represents the vector cross product operation. Multiplying the Riemannian metric tensor by the projection matrix yields the stiffness contribution matrix of each mooring cable to the center of gravity of the entire segment. The multiplication method is to multiply the transpose of the projection matrix by the Riemannian metric tensor, and then multiply by the projection matrix again, i.e., the first... The stiffness contribution matrix of the root mooring cable is , Indicates the first The projection matrix of the root mooring cable. This represents the Riemannian metric tensor.

[0038] The overall stiffness contribution matrix is ​​obtained by summing the stiffness contribution matrices of all mooring lines element-wise. This matrix is ​​then multiplied by the six-dimensional deviation vector to obtain the line force vector, which is used to eliminate the deviation vector. Each component of the line force vector corresponds to the axial force required to be applied to one mooring line. Dividing each component of the line force vector by the axial stiffness of the corresponding mooring line yields the elastic elongation distribution coefficient for each mooring line. This coefficient represents the change in line length required for each mooring winch to operate.

[0039] See Figure 5 In the figure, the coordinate axes represent the first and second principal components obtained after dimensionality reduction by principal component analysis of the compact feature vectors. The horizontal axis represents the first principal component, and the vertical axis represents the second principal component. In the legend, the blue dots represent sample points of the compact feature vector of the reference segment closure joint, and the red triangles represent sample points of the compact feature vector of the closure joint of the total segment to be closed. The blue and red dashed ellipses in the figure represent the 95% confidence ellipses of the compact feature vectors of the reference segment and the total segment to be closed, respectively.

[0040] As can be seen from the data distribution and ellipse shape, the compact feature vector of the baseline segment is mainly distributed in the range of approximately -3 to 2 for the first principal component and between approximately -2 and 2 for the second principal component, showing a clear concentration trend. The compact feature vector of the segment to be merged is generally shifted towards the positive direction of the first principal component compared to the baseline segment, with a distribution range of approximately -1 to 4 for the first principal component and between approximately -4 and 2 for the second principal component. There is some overlap, but the central position is significantly shifted.

[0041] The feature vector distribution reflects the geometrical differences in the closure seam feature manifold extracted by principal component analysis in the above embodiment. The compact features of the segment to be closed and the reference segment exhibit certain pose and shape differences. The 95% confidence ellipses of the two sets of data show significant discriminative power, providing an effective feature differentiation basis for dynamic time-warped matching.

[0042] In practice, the elastic elongation distribution coefficient of each mooring cable is converted into the corresponding mooring winch drum rotation angle increment. The conversion relationship between the mooring winch drum rotation angle increment and the elastic elongation distribution coefficient is determined by the radius of the mooring winch drum. The radius of the mooring winch drum is taken as the distance from the center line of the wire rope on the drum to the drum's rotation axis. The radius of the mooring winch drum is obtained from the mooring winch design drawings and is preset as a fixed parameter in the control system. The elastic elongation distribution coefficient is divided by the mooring winch drum radius to obtain the required radian value of the mooring winch drum rotation. This radian value is then converted into an angle to obtain the mooring winch drum rotation angle increment.

[0043] Based on the increment of the mooring winch drum's rotation angle, a speed command is sent to the drive motor of the mooring winch. The speed command for the drive motor is generated using proportional control. The proportional control coefficient is set based on the ratio of the drive motor's rated speed to the maximum allowable increment of the mooring winch drum's rotation angle, ensuring that within a single control cycle, the drive motor speed corresponding to the increment of the mooring winch drum's rotation angle does not exceed 80% of the drive motor's rated speed.

[0044] An encoder mounted on the mooring winch drum shaft provides real-time feedback on the actual rotation angle of the mooring winch drum. The incremental rotation angle is used as a setpoint and compared with the actual rotation angle fed back by the encoder to obtain the rotation angle deviation. This deviation is processed by the position loop controller to generate a corrected speed value. This corrected speed value is then superimposed on the speed command to form a position closed loop. The parameters of the position closed loop controller are determined by the inertia and damping characteristics of the mooring winch drum drive system and are tuned through step response tests during the commissioning phase.

[0045] Tension sensors installed on each mooring line collect the tension value of the mooring line in real time, obtaining the real-time tension value. The control system calculates the proportional distribution of the real-time tension values ​​of all mooring lines and compares the proportional distribution of the real-time tension values ​​with the ratio of the elastic elongation distribution coefficient. When the ratio of the real-time tension value of a certain mooring line to the real-time tension values ​​of other mooring lines deviates from the ratio of the corresponding mooring line in the elastic elongation distribution coefficient by more than a preset threshold, the speed command corresponding to that mooring line is corrected. The correction method is to superimpose an adjustment amount proportional to the proportional deviation on the speed command, so that the ratio between the tension of each mooring line and the elastic elongation distribution coefficient remains constant. The above adjustment amount is processed by low-pass filtering to filter out the interference of tension measurement noise on the speed command.

[0046] Under the combined effect of position closed-loop control and tension correction, the mooring winches synchronously raise and lower the mooring cable, driving the main section to move along the geodesic path. Synchronization refers to the coordinated movement of each mooring winch within the same control cycle, based on its respective elastic elongation distribution coefficient.

[0047] At least three laser targets are deployed at both edges of the closure joint of the section to be closed. The laser targets are evenly distributed along the circumference of the closure joint, and the arc length interval between adjacent laser targets is not less than one-quarter and not more than one-third of the circumference of the closure joint. Multiple optical prisms are installed on each laser target, and the optical prisms on the same laser target are spatially non-collinearly distributed. The number of optical prisms is not less than four, so that the total station can capture at least three of the optical prisms when observing the laser target from different angles.

[0048] An inertial sensor is installed at the center of gravity of the main segment. The center of gravity of the main segment is calculated based on the three-dimensional model of the main segment, and the coordinates of the inertial sensor mounting base are pre-calibrated on the main segment structure. The inertial sensor includes a three-axis gyroscope and a three-axis accelerometer. The three-axis gyroscope is used to measure the angular velocity about the three axes of the main segment's coordinate system, and the three-axis accelerometer is used to measure the acceleration along the three axes of the main segment's coordinate system.

[0049] The three-dimensional coordinates of each optical prism were measured from different stations using multiple total stations. A unified measurement coordinate system was established using resection of the total stations, and this system was aligned with the dock coordinate system. The three-dimensional coordinates of each optical prism were used as position observations, which were then incorporated into the observation update process of the Kalman filter.

[0050] The attitude angles output by the inertial sensor are used as attitude observations. The three-axis gyroscope in the inertial sensor outputs angular velocity signals, which are used by the attitude calculation algorithm to calculate the roll, pitch, and yaw angles of the entire segment in real time. The three-axis accelerometer provides a gravity direction reference to correct the accumulated error of the attitude calculation when there is no large translational acceleration in the entire segment. The attitude observations include three components: roll, pitch, and yaw.

[0051] The position and attitude observations are input into a Kalman filter. The Kalman filter employs an error-state Kalman filter framework, with the state vector representing the six degrees of freedom (DOF) of the overall segment's pose error, including sway, pitch, heave, roll, pitch, and yaw. The state equations are established based on a rigid body kinematics model, using the angular velocity and acceleration output from the inertial sensor as control inputs, and introducing zero-mean Gaussian white noise as process noise. The observation equations correlate the three-dimensional coordinates of the optical prism obtained from the total station with the position error in the state vector, and directly correlate the attitude observations with the attitude error in the state vector. At each sampling time, the Kalman filter performs a prediction step and an update step. The prediction step recursively derives the state vector using inertial sensor data, while the update step corrects the state vector using the position and attitude observations to obtain an estimate of the overall segment's pose, which includes six DDF components: sway, pitch, heave, roll, pitch, and yaw.

[0052] In practice, at the start of each repetitive cycle, the estimated position of the overall segment at the current moment is obtained by fusing the inertial sensor and the laser target. The estimated position of the overall segment contains six degrees of freedom components: sway, pitch, heave, roll angle, pitch angle, and bow angle, forming a six-dimensional vector. Based on the three translational and three rotational components in the estimated position of the overall segment, the coordinate transformation matrix of the segment to be joined relative to the reference segment is calculated using the homogeneous transformation matrix construction method in rigid body kinematics. The coordinate transformation matrix is ​​a 4x4 matrix, containing a 3x3 rotation matrix and a 3x1 translation vector, and its form is uniquely determined by the estimated position of the overall segment.

[0053] Using the calculated coordinate transformation matrix, the coordinates of each point in the closure seam edge point cloud of the segment to be closed are transformed. The homogeneous coordinate vector of each point in the closure seam edge point cloud of the segment to be closed is multiplied by the coordinate transformation matrix to obtain the updated closure seam edge point cloud. The updated closure seam edge point cloud represents the closure seam contour position of the segment to be closed in the current estimated pose.

[0054] Calculate the Hausdorff distance between the updated closure seam edge point cloud and the closure seam edge point cloud of the baseline segment. The Hausdorff distance measures the maximum mismatch between two point sets, and its calculation formula is as follows:

[0055] in, This represents the Hausdorff distance between the updated point cloud of the closure joint edge of the main segment to be closed and the point cloud of the closure joint edge of the reference segment. This represents the updated set of point clouds at the edges of the closure seam of the main segment to be closed; This represents the set of point clouds at the edge of the closure joint of the reference segment; yes A three-dimensional coordinate point; yes A three-dimensional coordinate point; Point and points The Euclidean distance between them; Indicates from From any point in the middle to The maximum value of the nearest distance between sets, i.e., the forward Hausdorff distance; Indicates from From any point in the middle to The maximum value of the nearest distance between sets, i.e., the backward Hausdorff distance; the final Hausdorff distance. Take the maximum value between the forward Hausdorff distance and the backward Hausdorff distance.

[0056] The process checks if the Hausdorff distance is less than an allowable value, which is set according to the joining process requirements and is 3 mm. If the Hausdorff distance is greater than or equal to the allowable value, the iteration is considered non-converged. In this case, the Riemann metric tensor is regenerated based on the location of the maximum deviation point corresponding to the Hausdorff distance. The maximum deviation point is the 3D coordinate point that reaches the maximum forward or backward Hausdorff distance during the Hausdorff distance calculation. When the maximum deviation point belongs to... At that time, extract the local area point cloud corresponding to the maximum deviation point from the point cloud of the closure seam edge of the updated closure section; when the maximum deviation point belongs to If the maximum deviation point is found, the local region point cloud corresponding to the maximum deviation point is extracted from the point cloud of the closure seam edge of the reference segment, and the corresponding local region point cloud closest to the maximum deviation point is determined from the point cloud of the closure seam edge of the total segment to be closed. Using the extracted local region point cloud as new input, the steps of closure seam feature manifold extraction, discrete sampling, calculation of external normal and tangential invariants, and compact feature vector generation are repeated, and the six-dimensional deviation vector and the corresponding Riemann metric tensor are recalculated before entering the next iteration loop.

[0057] When the Hausdorff distance is less than the allowable value, it is determined that the closure joint between the section to be closed and the reference section has reached the target positioning accuracy. At this point, all mooring cables between the section to be closed and the reference section are locked. The locking method for all mooring cables is as follows: control all mooring winch drive motors to stop torque output, and simultaneously apply mechanical brakes to all mooring winches to prevent the winch drums from rotating, thus maintaining the length of the mooring cables. After all mooring cables are locked, temporary positioning pins are inserted circumferentially along the closure joint. The temporary positioning pins are inserted at fixed angular intervals along the circumferential direction of the closure joint, and the diameter of the temporary positioning pins is clearance-fitted with the diameter of the positioning pin holes on the closure joint. After insertion, the temporary positioning pins maintain the relative positional relationship between the section to be closed and the reference section.

[0058] See Figure 6 In the figure, the vertical axis represents the Hausdorff distance in millimeters, and the horizontal axis represents the number of iterations, ranging from 0 to 500. The green solid curve shows the trend of Hausdorff distance between the updated edge point cloud of the closure joint of the segment to be closed and the edge point cloud of the closure joint of the reference segment as the number of iterations increases; the red dashed line represents the allowable value of Hausdorff distance, which is fixed at 3 millimeters.

[0059] As shown in the figure, the Hausdorff distance was relatively large at the initial iteration, approximately 27 mm, indicating a significant positional deviation between the closure joint of the segment to be closed and the reference segment. With increasing iterations, the Hausdorff distance exhibited a monotonically decreasing trend, with a relatively rapid rate of decrease in the first 200 iterations. This demonstrates that repeatedly fusing inertial sensor and laser target measurement data and updating the coordinate transformation matrix effectively reduced the maximum point set distance between the two closure joints, thus improving the matching accuracy of the closure joint.

[0060] Around 300 iterations, the Hausdorff distance began to approach the allowable value of 3 mm and then stabilized. Finally, after 400 iterations, the Hausdorff distance remained at around 3 mm, indicating that the maximum mismatch between the point clouds at the edge of the closure seam had reached the preset target positioning accuracy, and the precise docking of the closure segment was completed.

[0061] In specific implementation, please refer to Figure 3 The application of mechanical brakes to the mooring winches for each mooring cable is achieved by controlling the built-in brake actuator of the mooring winch. This actuator employs a spring-force locking and hydraulic pressure release structure. Upon receiving a brake command, the hydraulic oil is depressurized, and the spring force pushes the brake shoes to press against the brake disc of the mooring winch drum, preventing the drum from rotating. Simultaneously with applying the mechanical brakes, the drive motor of the mooring winch is brought to a zero-torque state. This zero-torque state is achieved by setting the torque current setpoint of the drive motor to zero, while simultaneously keeping the position and speed loops of the drive motor out of regulation.

[0062] A pre-positioning pin is inserted at preset angles along the circumference of the closure joint. The preset angles are determined based on the circumferential length of the closure joint and the total number of temporary positioning pins. The circumferential length of the closure joint is obtained by measuring the arc length of the closure joint profile of the reference section. The total number of temporary positioning pins is calculated by dividing the circumferential length of the closure joint by 800 mm and rounding up. The preset angle is equal to 360 degrees divided by the total number of temporary positioning pins. The diameter of the pre-positioning pin is 0.5 mm smaller than the diameter of the positioning pin hole on the closure joint. When inserting the pre-positioning pin, a thrust is applied along the axis of the pin hole. The pre-positioning pin is not tightened after entering the pin hole; it is only used to initially constrain the relative displacement on both sides of the closure joint.

[0063] The procedure for measuring the gap between adjacent pre-positioning pins is as follows: Use a feeler gauge to select several measurement points in the area between two adjacent pre-positioning pins along the length of the gap. After inserting the feeler gauge into the gap, read the gap value and record the gap value of each measurement point. Take the average value of the gap values ​​of all measurement points as the gap between adjacent pre-positioning pins.

[0064] Based on the distribution of the gap in the closure joint, auxiliary positioning pins are added at locations where the gap exceeds a set value. The set value is 5 mm, determined by the allowable gap unevenness in the closure process being one-third of the closure joint plate thickness tolerance. The closure joint plate thickness tolerance is 1.5 mm, and one-third is 0.5 mm. The set value is based on this, taking the maximum allowable gap of 5 mm for the closure joint. The auxiliary positioning pins are added as follows: at locations where the measured gap exceeds 5 mm, select the nearest existing positioning pin hole and insert the auxiliary positioning pin. The diameter of the auxiliary positioning pin is the same as that of the pre-existing positioning pin. If there is no pre-existing positioning pin hole at this location, the auxiliary positioning pin is replaced with a temporary clamp plate. The temporary clamp plate compresses the gap using clips and wedge-shaped shims welded to both sides of the closure joint.

[0065] Tighten all temporary locating pins in the order of first the middle and then the two ends along the length of the closure joint. The list of the order for tightening the temporary locating pins is arranged symmetrically from the geometric center of the closure joint to both ends. First, tighten the temporary locating pins located in the geometric center area, then tighten the temporary locating pins on both sides adjacent to the geometric center area, and so on, expanding outwards pair by pair until the outermost temporary locating pin.

[0066] During the tightening of each temporary positioning pin, the overall segment's pose is immediately monitored using inertial sensors and laser targets. The inertial sensors, including a three-axis gyroscope and a three-axis accelerometer, are installed at the segment's center of gravity and output real-time estimates of the overall segment's pose, including roll, pitch, yaw, sway, and heave. Laser targets are positioned at the edges of the closure joint, each equipped with multiple optical prisms. A total station measures the three-dimensional coordinates of these prisms in real-time, providing positional observations. The inertial sensor data and total station measurement data are input into a Kalman filter, which outputs an estimate of the overall segment's pose, comprising six degrees of freedom: yaw, pitch, heave, roll, yaw, and yaw.

[0067] The deviation of the currently monitored overall segment pose from the reference segment is compared with the allowable range. The allowable range is set to no more than 2 mm for each of the three degrees of freedom (sway, pitch, and heave) and no more than 0.1 degrees for each of the three degrees of freedom (roll, pitch, and yaw). If the deviation of the currently monitored overall segment pose from the reference segment exceeds the allowable range, tightening of the current temporary locating pin is stopped. Stopping the tightening of the current temporary locating pin is achieved by sending a stop signal to the tightening tool, which is a hydraulic torque wrench. The wrench stops rotating after the hydraulic torque wrench's oil supply circuit is closed.

[0068] The control module reverses the adjustment of the mooring winch corresponding to the area where the current temporary positioning pin is located. The area where the current temporary positioning pin is located is defined as a section extending one-sixth of the closure joint length to both sides of the current temporary positioning pin, along the closure joint length direction. This section is structurally covered by the nearest set of mooring cables. The mooring winch control module obtains the geometric correspondence between the mooring cables and the closure joint area from the overall section structure diagram, determining the mooring winch that needs to be activated for the area where the current temporary positioning pin is located. The direction of the reverse adjustment is the opposite direction of the overall section pose deviation vector. The magnitude of the reverse adjustment is determined based on the amplitude of the six-degree-of-freedom pose deviation of the overall section. A gain coefficient is used to convert the pose deviation amplitude into the mooring winch drum angle adjustment amount. The gain coefficient is obtained by inverting the pre-calibrated relationship matrix between the unit extension / retraction of the mooring cable and the overall section pose change. After the mooring winch performs the reverse adjustment, the deviation returns to the allowable range.

[0069] Once the deviation returns to the allowable range, continue tightening the current temporary locating pin. Stop tightening when the tightening torque reaches 50% of the yield strength of the temporary locating pin material. Then perform the same operation on the next temporary locating pin, processing each temporary locating pin in sequence from the middle to both ends, until all temporary locating pins are tightened.

[0070] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for precise assembly and positioning of ultra-large sections of container ships, characterized in that, include: Using the point clouds of the closure seam edges of the reference segment and the total segment to be closure as input, a nonlinear dimensionality reduction method is used to extract the closure seam feature manifold. Calculate the outward normal invariant and tangential invariant of the closure seam feature manifold at multiple discrete sampling points, and construct the Riemann metric tensor of the total segment six-degree-of-freedom pose deviation accordingly; The Riemannian metric tensor is mapped to the elastic elongation distribution coefficient of multiple mooring cables arranged on the mooring system on both sides of the dock. Based on the elastic elongation distribution coefficient, the corresponding mooring winch is controlled to apply forced displacement to the main section, so that the main section tends towards the reference section along the geodesic path determined by the tangential invariant; During the movement, the total segment pose is measured in real time by fusing laser targets arranged on both sides of the closure seam with inertial sensors, and the steps are repeated until the Hausdorff distance of the point cloud at the edge of the closure seam is less than the allowable value.

2. The method for precise assembly and positioning of ultra-large container ship sections according to claim 1, characterized in that, The extraction of the closure seam feature manifold using a nonlinear dimensionality reduction method includes: For each point in the point cloud at the edge of the closure seam, a K-dimensional tree is used to search for its neighboring points, and the Euclidean distance between the neighboring points is calculated. A weighted adjacency graph is constructed using the Euclidean distance as the edge weight, and then the geodesic distance between any two points is calculated using the Dijkstra algorithm to form a geodesic distance matrix. Applying a multidimensional scaling transformation to the geodesic distance matrix yields low-dimensional embedded coordinates; The points corresponding to the edge contour of the total segment in the low-dimensional embedded coordinates are sorted according to the curvature continuity, and the sorted point set is the closure seam feature manifold.

3. The method for precise assembly and positioning of ultra-large container ship sections according to claim 2, characterized in that, The calculation of the outward normal invariant and tangential invariant of the closure seam feature manifold at multiple discrete sampling points includes: Multiple discrete sampling points are selected at fixed arc length intervals on the characteristic manifold of the closure seam. For each discrete sampling point, a local quadratic surface is fitted using its neighborhood points to determine the two principal curvature directions of the local quadratic surface; By projecting the two principal curvature directions onto the tangent plane and normal plane of the closure seam respectively, the external normal invariant and tangential invariant of the discrete sampling point are obtained; The outward normal invariants and tangential invariants of all the discrete sampling points are sequentially combined to form a high-dimensional feature vector, and principal component analysis is performed on the high-dimensional feature vector to reduce its dimensionality, resulting in a compact feature vector.

4. The method for precise assembly and positioning of ultra-large container ship sections according to claim 3, characterized in that, The Riemannian metric tensor constituting the total six-degree-of-freedom pose deviation includes: The compact feature vector is dynamically time-warped and matched with the compact feature vector corresponding to the reference segment to obtain the corresponding point of each discrete sampling point on the reference segment. Based on the positional difference between the corresponding point and the discrete sampling point, calculate the spatial offset of each discrete sampling point in the overall coordinate system; The spatial offset is converted to the centroid of the whole segment to form a six-dimensional deviation vector; The covariance matrix of the total six-degree-of-freedom pose deviation is calculated using the six-dimensional deviation vector, and the inverse of the covariance matrix is ​​used as the Riemann metric tensor.

5. The method for precise assembly and positioning of ultra-large container ship sections according to claim 4, characterized in that, The mapping of the Riemann metric tensor to the elastic elongation distribution coefficient of multiple mooring cables arranged on both sides of the dock includes: Obtain the mooring point coordinates of each mooring cable on the main section, the fixed point coordinates on the mooring system on both sides of the dock, the axial stiffness, and the original length. Based on the mooring point coordinates, the fixed point coordinates, and the original length, calculate the direction vector of each mooring cable in the current attitude; Based on the direction vector, construct the projection matrix of each mooring cable, and multiply the Riemann metric tensor with the projection matrix to obtain the stiffness contribution matrix of each mooring cable to the center of gravity of the whole segment. The stiffness contribution matrices of each mooring cable are summed and multiplied with the six-dimensional deviation vector to obtain the cable force vector used to eliminate the six-dimensional deviation vector. The elastic elongation distribution coefficient is obtained by dividing each component of the cable force vector by the axial stiffness of the corresponding mooring cable.

6. The method for precise assembly and positioning of ultra-large container ship sections according to claim 5, characterized in that, The step of controlling the corresponding mooring winch to apply forced displacement to the main section based on the elastic elongation distribution coefficient includes: First, convert the elastic elongation distribution coefficient of each mooring cable into the corresponding mooring winch drum rotation increment; Based on the drum rotation angle increment, a speed command is sent to the drive motor of the mooring winch, and the actual rotation angle is fed back through the encoder installed on the drum shaft to form a position closed loop; The speed command is then corrected based on the real-time tension value fed back by the tension sensor installed on each mooring cable, so that the ratio between the tension and the elastic elongation distribution coefficient of each mooring cable remains constant. Under the combined effect of the position closed loop and tension correction, the mooring winch synchronously winds up and unwinds the cable, driving the main section to move along the geodesic path.

7. The method for precise assembly and positioning of ultra-large container ship sections according to claim 6, characterized in that, The method of using laser targets arranged on both sides of the closure seam and inertial sensors to measure the overall segment pose in real time includes: At least three laser targets are arranged at the edges of the closure seam on both sides of the section to be closed, and multiple optical prisms are installed on each laser target; The inertial sensor, which includes a three-axis gyroscope and a three-axis accelerometer, is installed at the center of gravity of the main segment. The three-dimensional coordinates of each optical prism are measured using a total station, and the three-dimensional coordinates are used as position observation values, while the attitude angles output by the inertial sensor are used as attitude observation values. The position observations and attitude observations are input into a Kalman filter to obtain an estimated value of the total segment pose, which includes six degrees of freedom components: sway, pitch, heave, roll, pitch and yaw.

8. The method for precise assembly and positioning of ultra-large container ship sections according to claim 7, characterized in that, The repeated steps until the Hausdorff distance of the point cloud at the closure seam edge is less than the allowable value include: Each time it is repeated, the coordinate transformation matrix of the segment to be joined relative to the reference segment is calculated based on the estimated value of the overall segment pose obtained by the current real-time measurement, and the closing seam edge point cloud of the segment to be joined is updated with the coordinate transformation matrix. Calculate the Hausdorff distance between the updated point cloud of the closure seam edge and the point cloud of the closure seam edge of the reference segment; Determine whether the Hausdorff distance is less than the allowable value. If not, regenerate the Riemann metric tensor based on the location of the maximum deviation point corresponding to the Hausdorff distance and proceed to the next iteration. If so, lock all mooring cables between the section to be joined and the reference section, and insert temporary positioning pins along the circumference of the joining seam.

9. The method for precise assembly and positioning of ultra-large container ship sections according to claim 8, characterized in that, The step of locking all mooring cables between the section to be joined and the reference section, and inserting temporary positioning pins circumferentially along the joining seam, includes: Apply a mechanical brake to the mooring winch of each of the mooring cables and bring the drive motor of the mooring winch to a zero-torque state; Insert a pre-positioning pin at preset angles along the circumference of the closure joint, and measure the closure joint gap between each adjacent pre-positioning pin; Based on the distribution of the gap in the closure joint, auxiliary positioning pins are added at positions where the gap is greater than the set value, and all temporary positioning pins are tightened in the order of first the middle and then the two ends along the length of the closure joint.

10. The method for precise assembly and positioning of ultra-large container ship sections according to claim 9, characterized in that, The process of tightening all temporary positioning pins in the order of first the middle and then the two ends along the length of the closure seam is as follows: As each of the temporary positioning pins is tightened, the overall segment pose is immediately monitored using the inertial sensor and the laser target; If the deviation of the current total segment pose relative to the reference segment exceeds the allowable range, then stop tightening the current temporary positioning pin and control the mooring winch corresponding to the area where the current temporary positioning pin is located to make a reverse adjustment so that the deviation returns to the allowable range. Once the deviation returns to the allowable range, continue tightening the current temporary locating pin, and then perform the same operation on the next temporary locating pin until all temporary locating pins are tightened.

Images