A high-precision measurement and automatic docking method adapted to ship block without allowance

Abstract

The present invention discloses a high-precision measurement and automatic docking method suitable for ship block without allowance, including: first, self-calibrating the measurement equipment, measuring the visible key points of the block docking surface and the coordinate positions of each trolley target ball; performing rough matching on the fixed block and the moving block to obtain the initial rotation matrix and the initial translation matrix of the key point set of the block docking surface, and obtaining the accurate rotation matrix and the accurate translation matrix; determining the alignment method of the moving block, determining the position of the trolley relative to the moving block according to the coordinate positions of the trolley target balls, and combining the accurate rotation matrix and the accurate translation matrix to obtain the motion planning trajectory of the trolley; compensating the motion trajectory of the trolley to obtain the compensated path docking trajectory; fusing the trolley position data measured by the two measurement devices to feedback and control the motion trajectory of the trolley. The method of the present invention realizes the automatic docking of ship blocks through self-calibration of the measurement equipment, data calculation, automatic attitude adjustment, compensation prediction and control feedback.

Description

Technical Field

[0001] The present invention relates to the technology of automatic butt - joint of ship blocks, and specifically to a high - precision measurement and automatic butt - joint method suitable for ship blocks. Background Art

[0002] In the ship manufacturing industry, the butt - joint of ship blocks is directly related to the overall manufacturing level of the ship. The quality of welding depends to a large extent on the butt - joint accuracy of ship blocks. There are many factors affecting ship blocks, and three of them have the greatest impact: measurement error, pose calculation error, and attitude adjustment error. Most patents reduce errors by proposing measurement and attitude adjustment methods. For example, Chinese Patent CN200810229112.9 discloses a method for butt - joint of two ship blocks during shipbuilding, which measures the coplanarity of the butt - joint seam through a laser theodolite. However, it does not consider the installation error of the equipment itself, does not take into account the occlusion of key points in the block during measurement, and does not consider the data conversion error, thus affecting the butt - joint accuracy of ship blocks. Summary of the Invention

[0003] Object of the Invention: The object of the present invention is to provide a high - precision measurement and automatic butt - joint method suitable for ship blocks, which is an automatic butt - joint method based on ship - block measurement, data calculation, automatic attitude adjustment, compensation prediction, and control feedback.

[0004] Technical Solution: A high - precision measurement and automatic butt - joint method suitable for ship blocks without allowance of the present invention includes the following steps:

[0005] S1. Fix the fixed block, place the moving block on an adjustable trolley, set common reference points on the foundations on both sides of the fixed block and the moving block, set corresponding key points on the butt - joint surfaces of the fixed block and the moving block, install measurement equipment. The measurement equipment includes a first laser tracker, a second laser tracker, and a laser rangefinder, and calibrate the measurement equipment through the common reference points to calculate the position of the measurement equipment in the global coordinate system. Measure the coordinates of each key point with the first laser tracker, and measure the position coordinates of the three - degree - of - freedom moving plane of the trolley with the second laser tracker; convert the coordinates of each key point and the position coordinates of the three - degree - of - freedom moving plane of the trolley into the global coordinate system;

[0006] S2. Coarsely match the fixed section and the moving section, and obtain the initial rotation matrix and the initial translation matrix of the key point sets of the docking surfaces of the fixed section and the moving section. Then, use the Iterative Closest Point (ICP) algorithm to further obtain the precise rotation matrix and the precise translation matrix. This includes: drawing circles with the key points of the moving section as the centers and a preset distance as the radius, calculating the weights of the key points of the moving section according to the number of pipelines in each circle; selecting three key points with the largest sum of weights and not collinear to form a plane, calculating the centroid of this plane, and according to the concept of random combination, selecting another three key points with the largest sum of weights and not collinear to form a plane, calculating the centroid of this plane, and so on to calculate the centroids corresponding to the key points of the moving section; calculating the centroids corresponding to the key points on the fixed section in the same way, selecting three pairs of corresponding and non-collinear centroids on the fixed section and the moving section, respectively establishing the coordinate system of the docking surface of the moving section and the coordinate system of the docking surface of the fixed section through the three-point alignment method, and resolving the two coordinate systems to obtain the initial rotation matrix and the initial translation matrix, thus realizing the coarse matching of the moving section and the fixed section; substituting the initial rotation matrix and the initial translation matrix into the ICP algorithm to solve for the precise rotation matrix and the precise translation matrix;

[0007] S3. Determine the docking method of the moving section. The docking method of the moving section includes two stages. The first stage: adjust the three degrees of freedom xyz displacement and the rotation direction around xz of the moving section; the second stage: adjust the displacement in the y direction of the moving section. Then, according to the position coordinates of the three-degree-of-freedom moving plane of the trolley in the global coordinate system obtained in step S1, determine the position of the trolley relative to the moving section, and combine the precise rotation matrix and the precise translation matrix obtained in step S2. Solve the distances that each degree of freedom of the trolley needs to move through inverse kinematics, and use the fifth-order polynomial fitting method to plan the path docking trajectory of the trolley to obtain the motion planning trajectory of the trolley;

[0008] S4. Use a BP neural network to predict the deformation error generated at the connection between the trolley and the moving section during the path docking adjustment. Train the BP neural network with historical data. The BP neural network includes an input layer, three hidden layers, and an output layer. Through training and prediction, obtain the error values of the deformation caused to the bottom structure of the moving section at different speeds, accelerations, and different weights under the same decision, so as to compensate the distances that each degree of freedom of the trolley obtained in step S3 during the path docking adjustment of the trolley, and recalculate the motion planning trajectory of the trolley in step S3 to obtain the compensated path docking trajectory;

[0009] S5. When adjusting the control trolley, the second laser tracker measures the coordinates of the trolley target balls on each trolley. Meanwhile, the laser rangefinder measures the moving distances of each trolley in three degrees of freedom directions. The data measured by the second laser tracker and the laser rangefinder are fused and processed to provide real-time feedback on the adjustment state of the trolley until the actual path docking trajectory of the trolley is consistent with the compensated path docking trajectory in step S4.

[0010] Further, in step S1, target balls are installed at each key point position and on the three-degree-of-freedom moving plane of the trolley. The three-degree-of-freedom moving plane of the trolley includes the lower bottom surface, the middle layer surface, and the upper layer surface of the trolley. The first laser tracker is used to measure the coordinates of the target balls at each key point, and the second laser tracker is used to measure the coordinates of the trolley target balls on each degree-of-freedom moving plane of the trolley.

[0011] The number of the first laser trackers is ensured to be able to measure all the target balls at the key points, and the position of each first laser tracker is ensured to be able to measure the most target balls at the key points.

[0012] The number of the second laser trackers is ensured to be able to measure all the target balls on the three-degree-of-freedom moving plane of the trolley, and the position of each second laser tracker is ensured to be able to measure the most target balls on the three-degree-of-freedom moving plane of the trolley.

[0013] Further, the specific method of rough matching in step S2 is as follows:

[0014] (101) Calculate the weights of each key point on the moving total section. Respectively, draw circles with each key point as the center and a radius of 1 / 2 of the deck width of the ship total section. Then, the weight w i of the i-th key point on the moving total section is:

[0015]

[0016] where M is the total number of pipelines on the moving total section, and N i is the number of pipelines whose centers of circles are included in the circle R i with the i-th key point on the moving total section as the center, where i is the number of the i-th key point on the moving total section; the set of weights of all key points on the moving total section is W = {w1,..., w N} and N is the total number of key points on the moving total section;

[0017] (102) From the set of weights of all key points W = {w1,..., w NAmong them, first select three key points with the largest sum of weights and that are not collinear. Connect the three points pairwise head to tail to form a planar triangle, and calculate the centroid of the triangle as g1. According to the concept of random combination, select three key points with the largest sum of weights and that are not collinear except for the above combination. Connect the three points pairwise head to tail to form a planar triangle, and calculate the centroid of this triangle as g2. Similarly, calculate the centroids corresponding to all possible combinations of three key points on all the key points of the moving total segment in this way. Where H is the number of groups where three key points are collinear;

[0018] (103) Calculate the centroids corresponding to the key points on the fixed total segment in the same way. Take the three non-collinear centroids g j , g k , g l on the moving total segment and the corresponding three centroids g j ′, g′ k , g l ′ on the fixed total segment, and perform docking and matching using the three-point alignment method to achieve the rough matching of the moving total segment and the fixed total segment.

[0019] Furthermore, the calculation methods of the precise rotation matrix and the precise translation matrix in step S2 are as follows:

[0020] (201) Establish the docking surface coordinate system of the moving total segment and the docking surface coordinate system of the fixed total segment;

[0021] Take point g j as the coordinate origin, and take the direction vector from g j to g k as the x-axis to construct a unit vector:

[0022]

[0023] Then, determine the y-axis by taking the cross product of the direction vector from g j to g l and the x-axis direction vector, and construct the y-axis unit vector:

[0024]

[0025] Finally, determine the z-axis direction by taking the cross product of e1 and e2, and its unit vector is:

[0026] e3 = e1 × e2

[0027] Then the docking surface coordinate system of the moving total segment is: O g -e1, e2, e3;

[0028] Similarly, determine the docking surface coordinate system of the fixed total segment as: O g′ -e1′, e′2, e3′;

[0029] (202) Calculate the initial rotation matrix and the initial translation matrix;

[0030] By using the three - point alignment method to match the measurement data, the initial rotation transformation matrix R0 of the key point sets on the moving total section docking surface and the fixed total section docking surface is expressed as:

[0031] R0 = [e′1, e′2, e′3] * [e1, e2, e3] T

[0032] According to the initial rotation matrix R0, calculate the initial translation matrix t0 as:

[0033] t0 = g j '- g j R0

[0034] (203) Substitute the initial rotation matrix R0 and the initial translation matrix t0 into the Iterative Closest Point (ICP) algorithm, and through iterative calculation, solve for the accurate rotation matrix R and the accurate translation matrix t.

[0035] Furthermore, in step S3, the docking trajectory planning of the moving total section path adopts the step - by - step posture adjustment method, specifically:

[0036] (31) Adjust the rotation angle α of the moving total section around the x - direction to rotate the moving total section of the ship to a position horizontal to the x - axis, adjust the rotation angle β of the moving total section around the y - direction to rotate the moving total section of the ship to a position parallel to the y - axis, adjust the rotation angle γ of the moving total section around the z - direction to rotate the moving total section of the ship to a position parallel to the z - axis, and simultaneously adjust the translation amounts in the x and z degrees of freedom directions;

[0037] (32) When the distance between the moving total section and the fixed total section reaches the preset distance threshold, adjust the translation in the y - direction to avoid interference during the adjustment process of the moving total section due to the presence of the total section pipeline, and adopt the fifth - order polynomial trajectory method to plan the docking trajectory of the trolley path.

[0038] Furthermore, step S4 is specifically:

[0039] (41) Use a BP neural network to predict the docking errors caused by the plate deformations of each moving total section of different ship types with different weights under the same acceleration and speed. In the same assembly environment, with the same speed and acceleration, and the trolley in the same support position on the total section, with the control variable being the weight of the moving total section of the ship, extract the on - site data, predict the docking errors caused by moving total sections with different weights, and solve the docking errors to obtain the compensation amounts in the three degrees of freedom directions of each trolley, and add the compensation amounts to the docking trajectory of the trolley path planned in step S3 to adjust the docking trajectory of the trolley path;

[0040] (42) Control the weight of the ship's moving block, keep the trolley position and assembly environment unchanged, so as to predict the docking deviation of the block at different accelerations and speeds, and compensate the three-degree-of-freedom motion trajectories of each trolley. The input layer of the BP neural network contains five neurons for selecting weight, speed, acceleration, displacement, and material properties; the hidden layer is set to 3 layers and contains multiple neurons; the output layer has one neuron, which is the displacement compensation amount of each trolley in each direction.

[0041] Further, step S5 is specifically as follows:

[0042] The second laser tracker measures the trolley target balls set on the three-degree-of-freedom moving plane of the trolley. At the same time, a laser rangefinder is used to measure the motion of the three degrees of freedom of the trolley, and the data measured by the second laser tracker and the laser rangefinder are fused. When the second laser tracker measures that a certain trolley target ball is blocked, the data is supplemented by the laser rangefinder; when the trolley has a power failure or a fault, the trolley feedback measurement device records the current position of the trolley; when the trolley is powered on or the fault is eliminated, the trolley path docking trajectory control continues until the moving block and the fixed block are docked.

[0043] A high-precision measurement and automatic docking system suitable for ship's non-margin blocks of the present invention includes:

[0044] A data acquisition and processing module, which is used to calibrate the measurement device according to the set common reference point, calculate the position of the measurement device in the global coordinate system, measure the coordinates of each key point target ball on the docking surfaces of the fixed block and the moving block by using the measurement device, measure the position coordinates of the three-degree-of-freedom moving plane of the trolley at the same time, and convert the coordinates of each key point and the position coordinates of the three-degree-of-freedom moving plane of the trolley to the global coordinate system by using the singular value decomposition method SVD;

[0045] The rough matching and precise matching module is used to draw circles with the key points of each moving total segment as the centers and a preset distance as the radius, calculate the weights of the key points of each moving total segment according to the number of pipelines in each circle; sort the key points of each moving total segment in ascending order of weights, arbitrarily select three key points with the largest weights and not collinear, form a plane, obtain the centroid of this plane, according to the concept of random combination, select three key points with the largest sum of weights and not collinear except the above combination, form a plane, obtain the centroid of this plane, and so on to obtain the centroids corresponding to the key points of the moving total segment; find the centroids corresponding to the key points on the fixed total segment in the same way, select three pairs of corresponding centroids on the moving total segment and the fixed total segment and use the three-point alignment method. According to the three pairs of corresponding centroids selected on the moving total segment and the fixed total segment, establish the docking surface coordinate system of the moving total segment and the docking surface coordinate system of the fixed total segment respectively, solve the docking surface coordinate system of the moving total segment and the docking surface coordinate system of the fixed total segment, obtain the initial rotation matrix and the initial translation matrix, and realize the rough matching of the moving total segment and the fixed total segment; substitute the initial rotation matrix and the initial translation matrix into the Iterative Closest Point (ICP) algorithm to solve the precise rotation matrix and the precise translation matrix, and realize the precise matching;

[0046] The path docking trajectory planning module determines the position of the trolley relative to the moving total segment. The docking method of the moving total segment includes two stages. The first stage: adjust the three-degree-of-freedom xyz displacement and the rotation around the xz axis of the moving total segment; the second stage: adjust the y-direction displacement of the moving total segment; according to the obtained position coordinates of the three-degree-of-freedom moving plane of the trolley in the global coordinate system, as well as the obtained precise rotation matrix and precise translation matrix, use inverse kinematics to solve the distances that the trolley needs to move in each degree of freedom, and use the fifth-order polynomial fitting method for the trolley according to this distance data to plan the path docking trajectory.

[0047] The deformation compensation module is used to predict the deformation error of the bottom plate of the moving total segment generated at the connection between the trolley and the moving total segment during the path docking adjustment by using a BP neural network. The BP neural network is trained through historical data. The BP neural network includes an input layer, three hidden layers, and an output layer. Through training and prediction, the error values of deformation caused by different speeds, accelerations, and different weights under the same decision are obtained, so as to compensate the movement amounts of each degree of freedom of the trolley during the path docking adjustment of the trolley, and obtain the path docking trajectory after deformation compensation;

[0048] The motion feedback module is used to measure the target ball coordinates of each trolley in the three-degree-of-freedom direction of the trolley by using a measuring device when controlling the trolley to adjust, and at the same time use the measuring device to measure the plane distances of each trolley moving in the three-degree-of-freedom direction, fuse the target ball coordinates of each trolley in the three-degree-of-freedom direction measured by the measuring device and the distance data of each trolley moving in the three-degree-of-freedom direction, and give real-time feedback on the adjustment state of the trolley until the actual path docking trajectory of the trolley is consistent with the path docking trajectory after deformation compensation.

[0049] Preferably, the measuring device includes a first laser tracker, a second laser tracker and a laser rangefinder. Target balls are respectively installed at the key point positions and on the three-degree-of-freedom moving plane of the trolley. The three-degree-of-freedom moving plane of the trolley includes the lower bottom surface, the middle layer surface and the upper layer surface. The first laser tracker is used to measure the coordinates of the target balls at each key point. The second laser tracker is used to measure the coordinates of the trolley target ball on the trolley. The laser rangefinder is used to measure the moving distances of the trolley on the three-degree-of-freedom direction planes.

[0050] The number of the first laser trackers ensures that all the target balls at the key points can be measured, and the position of each first laser tracker ensures that the maximum number of target balls at the key points can be measured.

[0051] The number of the second laser trackers ensures that all the target balls on the three-degree-of-freedom moving plane of the trolley can be measured, and the position of each second laser tracker ensures that the maximum number of target balls on the three-degree-of-freedom moving plane of the trolley can be measured.

[0052] A device of the present invention includes a memory and a processor, wherein:

[0053] The memory is used to store a computer program that can run on the processor.

[0054] The processor is used to execute the steps of a high-precision measurement and automatic docking method adapted to the ship's non-margin block as described above when running the computer program.

[0055] Beneficial effects: Compared with the prior art, the remarkable technical effects of the technical solution of the present invention are as follows: Aiming at the problems that the ship pipelines are relatively complex, it is not easy to align and the alignment is less during the docking process, a method of assigning weights is adopted to optimize the three-point alignment method, so that the finally solved attitude of the moving block is more accurate. Aiming at the problem that the bottom plate parts in the contact area between the ship's moving block and the trolley are prone to deformation, a BP-neural network is proposed to predict the error caused by the deformation of the bottom plate of the moving block under different conditions, so as to compensate during the adjustment movement of the trolley. It is beneficial to reduce the influence of the deformation of the bottom plate of the moving block on the block alignment, and further improve the docking accuracy and quality of the two blocks. Description of the Drawings

[0056] Figure 1 is a schematic flow chart of the docking method of the present invention;

[0057] Figure 2 is a detailed flow chart of the docking method of the present invention;

[0058] Figure 3 is a schematic diagram of the docking measurement system of the present invention;

[0059] Figure 4 is a schematic diagram of the establishment of each coordinate system in the docking system of the present invention;

[0060] Figure 5 It is a schematic diagram of the distribution of key points in the docking system of the present invention;

[0061] Figure 6 It is a schematic diagram of the distribution of each common reference point in the top view state of the docking system of the present invention;

[0062] Figure 7 It is a schematic diagram of the weight matching method for each key point in the docking method of the present invention. Detailed implementation manners

[0063] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0064] As Figure 1 and Figure 2 shown, a high-precision measurement and automatic docking method suitable for ship blocks of the present invention is as follows. First, the first to fourth laser trackers are calibrated by measuring the common reference points on their same sides. After calibration, the first laser tracker and the second laser tracker are used to measure the key point target balls of the two blocks, and the third laser tracker and the fourth laser tracker are used to measure the trolley target balls on the four trolleys. Then, according to the key point coordinates, the accurate rotation matrix and the accurate translation matrix between the moving block and the fixed block are solved by using the three-point alignment method combined with weight optimization and the ICP algorithm. At the same time, by measuring the trolley target balls on the four trolleys, the actual position of the trolley before movement is obtained. Then, the block alignment method is determined. Combining the above accurate rotation matrix, accurate translation matrix and the actual trolley position data, the theoretically required movement distance of the three degrees of freedom of the trolley is solved by inverse kinematics. Furthermore, a quintic polynomial is used to plan the trajectory of the trolley. Through the visualization of the upper computer, it is judged whether this trajectory planning will cause interference between the moving block and the fixed block of the ship. If interference occurs, different path points are reselected to plan the path of the trolley. If there is no interference phenomenon, the error caused by the deformation of the bottom plate of the moving block of the ship in different situations is predicted by the BP-neural network, so as to solve the compensation amount required during the movement of the trolley. Furthermore, the trolley compensation is realized through PLC control. Finally, during the movement process, the actual movement position of the trolley is solved by measuring the trolley target balls with the laser tracker and measuring each plane of the trolley with the laser rangefinder, and then the difference between the actual movement position and the theoretical movement position of the trolley is obtained, and the movement position of the trolley is fed back to control the trolley to realize the two-stage adjustment of the moving block, so as to realize the high-precision docking of the moving block.

[0065] Specifically, it includes the following steps:

[0066] S1. Fix the fixed total section in place, and place the movable total section on the adjustable trolley. Set common reference points on the foundation on both sides of the fixed total section and the movable total section. The common reference points are not easily deformed on the foundation. Set corresponding key points on the docking surfaces of the fixed total section and the movable total section. Install measuring equipment, which includes a first laser tracker, a second laser tracker, and a laser rangefinder, and calibrate the measuring equipment through the common reference points (that is, use the first laser tracker, the second laser tracker, and the laser rangefinder to measure the common reference points to calibrate their own positions), calculate the positions of the measuring equipment in the global coordinate system. After the measuring equipment is calibrated, use the first laser tracker to measure the coordinates of each key point on the docking surfaces of the movable total section and the fixed total section, and use the second laser tracker to measure the actual position coordinates of the three-degree-of-freedom moving planes of each trolley before adjustment; convert the coordinates of each key point and the actual position coordinates of the three-degree-of-freedom moving planes of each trolley before adjustment to the global coordinate system.

[0067] Such as Figure 3As shown in the figure, where: 1 - fixed general section; 2 - moving general section; 3 - first laser tracker; 4 - second laser tracker; 5 - third laser tracker; 6 - fourth laser tracker; 7 - three - dimensional attitude - adjusting trolley; 8 - trolley target ball; 9 - laser rangefinder; 10 - fixed support pier; 11 - common reference point target ball; 12 - key - point target ball. The fixed general section 1 is supported by four support piers 10, and the four support piers are symmetric in pairs; the moving general section 2 is arranged on four three - dimensional attitude - adjusting trolleys, and trolley target balls are installed on the three - degree - of - freedom moving planes of the four trolleys for the laser tracker to measure and feedback the distance for adjusting the trolleys. The three - degree - of - freedom moving plane of the trolley includes the lower bottom surface, the middle layer surface and the upper layer surface; six pairs of corresponding key - point sets are arranged on the docking surface between the fixed general section and the moving general section, and key - point target balls 12 are installed at each key - point position for subsequent measurement by the first laser tracker; on the ground on both sides of the fixed general section and the moving general section of the ship, 8 pairs of common reference points are symmetrically arranged, and common reference point target balls 11 are arranged on each common reference point; set the positions of the first and second laser trackers. The positions of the first and second laser trackers should be such that the laser trackers are placed to measure as many key - point target balls as possible; then the first and second laser trackers respectively measure the common reference point target balls on their same sides, so as to determine the accurate positions of the first and second laser trackers at this time, realize the position calibration of the first and second laser trackers, calculate the positions of the first and second laser trackers in the global coordinate system at this time. After calibration, use the first and second laser trackers to measure the key - point target balls of the fixed general section and the key - point target balls of the moving general section; control the brackets of the third and fourth laser trackers to descend, set the third and fourth laser trackers to be placed at positions where the coordinates of the trolley target balls can be measured without obstruction, fix this position, and then the third and fourth laser trackers respectively measure the common reference point target balls on their same sides, calibrate the positions of the third and fourth laser trackers at this time, calculate the positions of the third and fourth laser trackers in the global coordinate system at this time, and further measure the coordinates of the trolley target balls on the four trolleys; convert the measured coordinates of the key - point target balls of the fixed general section and the moving general section of the ship and the coordinates of the trolley target balls on each trolley to the global coordinate system through singular value decomposition (SVD) to obtain the coordinates of the key - point target balls and the coordinates of the trolley target balls in the global coordinate system.

[0068] As Figure 4 shown, establish a global coordinate system O - XYZ, establish a fixed general - section coordinate system O1 - X1Y1Z1, and a moving general - section coordinate system O2 - X2Y2Z2. Select the key points P1 - P6 to be measured on the docking surface of the fixed general section and the key points Q1 - Q6 on the docking surface of the moving general section according to the design drawings and the key - position coordinates in the design drawings. The key points P1 - P6 and Q1 - Q6 correspond one by one, and the positions and distributions of the key points are as Figure 5As shown in the figure, 16 specific common reference points are set on the overall section assembly foundation, and the first 8 common reference points M1 to M8 and the last 8 common reference points M9 to M 16 are symmetric with respect to the center line in the Y1-axis direction of the fixed overall section of the ship. The positions and distributions of the common reference points are as Figure 6 shown. Four laser trackers are respectively placed at positions I to IV. The first laser tracker is placed at position I, the second laser tracker is placed at position II, the third laser tracker is placed at position III, and the fourth laser tracker is placed at position IV. The first laser tracker measures the common reference points M1 to M8 on the foundation, and obtains the accurate position of the first laser tracker by measuring the common reference points, and establishes the coordinate system O3-X3Y3Z3 of the first laser tracker at position I. Then, the first laser tracker placed at position I is used to measure the key points of the fixed overall section of the ship and the movable overall section of the ship, and obtains the coordinate positions of the measurable key points P1, P2, P3 of the fixed overall section of the ship and the measurable key points Q1, Q2, Q3 of the movable overall section of the ship in the coordinate system of the first laser tracker at position I. Similarly, the second laser tracker measures the common reference points M9 to M 16 to obtain the accurate position of the second laser tracker, and establishes the coordinate system O4-X4Y4Z4 of the second laser tracker at position II. The second laser tracker is used to measure the coordinate positions of the measurable key points P4, P5, P6 of the fixed overall section of the ship and the measurable key points Q4, Q5, Q6 of the movable overall section of the ship in the coordinate system of the second laser tracker at position II. By fusing and resolving the key point coordinates of the movable overall section and the fixed overall section, the singular value decomposition method (SVD) is used to transform the key point coordinates of the key points P1, P2, P3, Q1, Q2, Q3 in the coordinate system O3-X3Y3Z3 of the first laser tracker at position I and the key point coordinates of the key points P4, P5, P6, Q4, Q5, Q6 in the coordinate system O4-X4Y4Z4 of the second laser tracker at position II to the global coordinate system, and obtains the key point coordinates in the global coordinate system.

[0069] Similarly, the third laser tracker and the fourth laser tracker respectively measure the common reference points M1 to M8 and M9 to M on the foundation on their same sides 16 and obtain the accurate positions of the third laser tracker and the fourth laser tracker by measuring the common reference points, and respectively establish the coordinate system O5-X5Y5Z5 of the third laser tracker at position III and the coordinate system O6-X6Y6Z6 of the fourth laser tracker at position IV. Then, the third laser tracker and the fourth laser tracker are used to measure the actual coordinates of the trolley target balls on the trolleys on their same sides respectively, and the obtained trolley target ball coordinates are transformed to the global coordinate system by the singular value decomposition method (SVD), and the actual position before the trolley moves is obtained.

[0070] S2. Coarsely match the fixed section and the moving section, and obtain the initial rotation matrix and the initial translation matrix of the key point sets of the docking surfaces of the fixed section and the moving section. Then, use the Iterative Closest Point (ICP) algorithm to further obtain the accurate rotation matrix and the accurate translation matrix. The method includes: drawing circles with the key points of the fixed section as the centers and a preset distance as the radii, calculating the weights of the key points of the fixed section according to the number of pipelines in each circle; arbitrarily selecting three key points with the largest weights and not collinear to form a plane, and obtaining the centroid of this plane. According to the concept of random combination, select three key points with the largest sum of weights and not collinear except for the above combination to form a plane, and obtain the centroid of this plane. And so on to obtain the centroids corresponding to the key points of the fixed section; calculate the centroids corresponding to the key points on the moving section in the same way. Select three pairs of corresponding and non-collinear centroids on the fixed section and the moving section, and respectively establish the coordinate system of the docking surface of the moving section and the coordinate system of the docking surface of the fixed section through the three-point alignment method. Solve the coordinate systems of the docking surface of the moving section and the docking surface of the fixed section constructed by the three-point alignment method to obtain the initial rotation matrix and the initial translation matrix, and realize the coarse matching of the moving section and the fixed section. Substitute the initial rotation matrix and the initial translation matrix into the Iterative Closest Point (ICP) algorithm to solve the accurate rotation matrix and the accurate translation matrix. This method enables more pipelines to be accurately docked on the basis of the completion of the ship end face docking, and solves the problems of difficult pipeline docking and fewer accurate docking quantities.

[0071] First, solve the measurement data in step S1. The data solution is to perform fitting and solution on the key point coordinates measured for the fixed section and the moving section. The fitting and solution first coarsely match the key point sets measured for the docking surfaces of the moving section and the fixed section of the ship by using the three-point alignment method. The coarse matching method is optimized by combining with pipeline docking. It includes:

[0072] (101) Draw circles with the key points of the moving section as the centers and a preset distance as the radii, calculate the weights of the key points of the fixed section according to the number of pipelines in each circle; sort the key points of the moving section in ascending order of weights, select three key points with the largest weights and not collinear to form a plane, and obtain the centroid of this plane; then select the three non-collinear points with the largest weights on the moving section of the ship except for the above combination, connect them in sequence to form a plane, and obtain the centroid of this plane; and so on to obtain the centroids corresponding to the remaining key points of the moving section; calculate the centroids corresponding to the key points on the fixed section in the same way. Select three pairs of corresponding and non-collinear centroids on the moving section and the fixed section, and respectively establish the coordinate system of the docking surface of the moving section and the coordinate system of the docking surface of the fixed section through the three-point alignment method. Solve the coordinate systems of the docking surface of the moving section and the docking surface of the fixed section constructed by the three-point alignment method to obtain the initial rotation matrix and the initial translation matrix, and realize the coarse matching of the moving section and the fixed section.

[0073] When docking ship sub - assemblies, the accuracy of pipeline docking in the sub - assemblies needs to be considered. Pipeline measurement is carried out manually using tools such as measuring tapes, with the nearest structure as the reference. The reference for ship pipeline measurement is generally the strong structure of the ship sub - assembly docking surface. The key measurement points of ship sub - assembly measurement are generally at the structure joints. Therefore, key - point weights are introduced to better control the docking accuracy of pipelines, enabling more pipe fittings to be completely docked. For example Figure 7 As shown, for the convenience of explaining this method in this embodiment, four key points (Q1, Q2, Q4, Q5) among the six key points of the ship mobile sub - assembly docking surface are used for elaboration, and this method can be analogized for multiple key points more than four. This method assigns weights to each key measurement point by calculating the number of pipelines in each circle. The specific content of weight assignment is as follows: For Q1, with Q1 as the center, preferably with a radius R equal to half of the fixed sub - assembly deck width to draw a circle R1, and the maximum selection of the radius shall not exceed the diagonal distance of the fixed sub - assembly. Similarly, with Q2, Q4, and Q5 as the centers respectively, draw corresponding circles R2, R4, and R5 with a radius of R. The total number of pipelines in the ship mobile sub - assembly is M; the number of pipelines whose centers are included in the circle R i with the i - th key point on the mobile sub - assembly as the center is N i , in this embodiment, i = 1, 2, 4, 5. Then the weight value w i of the i - th key point on the mobile sub - assembly is:

[0074]

[0075] where M is the total number of pipelines in the mobile sub - assembly, and N i is the number of pipelines whose centers are included in the circle R i with the i - th key point on the mobile sub - assembly as the center, where i is the number of the i - th key point on the mobile sub - assembly; the set of weight values of all key points on the mobile sub - assembly is W = {w1, …, w N}, and N is the total number of key points on the mobile sub - assembly.

[0076] (102) From the set of weight values of all key points on the mobile sub - assembly W = {w1, …, w N}, first select three key points with the largest sum of weight values and non - collinear. Connect the three points pairwise head - to - tail to form a planar triangle, and calculate the centroid of the triangle as g1. According to the concept of random combination, select three key points with the largest sum of weight values and non - collinear except the above combination to form a plane, and obtain the centroid g2 of this plane. Similarly, calculate the centroids corresponding to the key points on the mobile sub - assembly according to this method where H is the number of groups where the selected key points are collinear. Randomly select three key points from N key points and use the concept of random combination: Represents the cases of all combinations of three key points; then, let H represent the number of groups where the three-point combination shows collinear phenomenon, and the difference between them is the number of all non-collinear groups of three-point random combinations.

[0077] (103) Calculate the centroid corresponding to the key points on the fixed total segment according to the same method as above. For the three non-collinear centroids g j , g k , g l on the moving total segment, and the corresponding three centroids g j ′, g k ′, g l ′ on the fixed total segment, perform docking and matching using the three-point alignment method to achieve the rough matching of the moving total segment and the fixed total segment.

[0078] Secondly, obtain the initial rotation matrix and the initial translation matrix of the key point sets of the docking surfaces of the moving total segment and the fixed total segment, and solve for the exact rotation matrix and the exact translation matrix.

[0079] (201) Establish the coordinate systems of the docking surfaces of the moving total segment and the fixed total segment.

[0080] Take the g j point as the coordinate origin, and take the direction vector from g j to g k as the x-axis to construct a unit vector:

[0081]

[0082] Then, determine the y-axis by taking the cross product of the direction vector from g j to g l and the x-axis direction vector, and construct the y-axis unit vector:

[0083]

[0084] Finally, determine the z-axis direction by taking the cross product of e1 and e2, and its unit vector is:

[0085] e3 = e1 × e2

[0086] Then the coordinate system of the docking surface of the moving total segment is: O g -e1, e2, e3;

[0087] Similarly, determine the coordinate system of the docking surface of the fixed total segment as: O g′ -e′1, e′2, e′3;

[0088] (202) Calculate the initial rotation matrix and the initial translation matrix.

[0089] By using the three - point alignment method to match the measurement data, the initial rotation transformation matrix \(R_0\) of the key - point sets on the moving total - section docking surface and the fixed total - section docking surface is expressed as:

[0090] \(R_0=[e'_1,e'_2,e'_3]*[e_1,e_2,e_3]\) T

[0091] According to the initial rotation matrix \(R_0\), the initial translation matrix \(t_0\) is calculated as:

[0092] \(t_0 = g\) j \('-g\) j \(R_0\)

[0093] (203) Substitute the initial rotation matrix \(R_0\) and the initial translation matrix \(t_0\) into the Iterative Closest Point (ICP) algorithm. Through iterative calculation, the accurate rotation matrix \(R\) and the accurate translation matrix \(t\) are solved, thus laying a foundation for ship trajectory planning.

[0094] In this embodiment, after obtaining the weights of each point, first select three non - collinear key points with the largest sum of weights. Connect these three points in sequence to form a plane, and calculate the centroid of the plane as \(g_1\). Then, select three non - collinear key points on the moving total - section of the ship with the largest sum of weights except the above - mentioned group. Connect these three points in sequence to form a plane, and calculate the centroid of the plane as \(g_2\). Finally, select three non - collinear key points on the moving total - section of the ship with the largest sum of weights except the above - mentioned two groups. Connect these three points in sequence end - to - end to form a plane, and calculate the centroid of the plane as \(g_3\). Similarly, the corresponding three centroids of the fixed total - section are \(g_1',g_2',g_3'\). Use these three pairs of centroids to achieve rough matching of the three - point alignment method.

[0095] Taking point \(g_1\) as the coordinate origin and the direction vector from \(g_1\) to \(g_2\) as the \(x\) - axis, construct a unit vector:

[0096]

[0097] Then, determine the \(y\) - axis by the cross - product of the direction vector from \(g_1\) to \(g_3\) and the \(x\) - axis direction vector, and construct a unit vector of the \(y\) - axis:

[0098]

[0099] Finally, determine the \(z\) - axis direction by the cross - product of \(e_1\) and \(e_2\), and its unit vector is:

[0100] \(e_3 = e_1\times e_2\)

[0101] Then the coordinate system of the moving total - section docking surface is: \(O\) g \(-e_1,e_2,e_3\)

[0102] Similarly, determine the coordinate system of the fixed total - section docking surface as: \(O\) g′ \(-e_1',e_2',e_3'\)

[0103] The initial rotation matrix of the measurement point sets of the mating surface of the moving block and the mating surface of the fixed block can be expressed as:

[0104] R0 = [e′1, e′2, e′3] * [e1, e2, e3] T

[0105] According to the obtained initial rotation matrix R0, determine the initial translation matrix t0:

[0106] t0 = g1′ - g1R0

[0107] Solve for the initial rotation matrix R0 and the initial translation matrix t0, substitute them into the Iterative Closest Point (ICP) algorithm for iterative calculation, solve for the precise rotation matrix R and the precise translation matrix t, thereby laying a foundation for the ship trajectory planning.

[0108] S3. First, establish the alignment method of the moving block of the ship. The docking of the ship block is different from the docking of other objects. Due to its extremely large area, it needs to be adjusted step by step, and it is necessary to ensure that there is no interference during the docking attitude adjustment. Therefore, the overall docking process of the moving block is divided into two stages: The first stage is to adjust the rotation of the moving block around the three degrees of freedom xyz axes and the translation in the xz directions until the moving block of the ship reaches the target position of the first stage; the second stage is to adjust the translation of the moving block in the y-axis direction; finally, the alignment of the moving block is completed. The rotation of the moving block around the x-axis can be achieved by the coordinated movement of the four trolleys in the yz directions. Similarly, the rotation of the moving block around the y-axis can be achieved by the coordinated movement of the four trolleys in the xz directions, and the rotation of the moving block around the z-axis can be achieved by the coordinated movement of the four trolleys in the xy directions. The translation of the ship block in the xyz directions can be achieved by the simultaneous translation of the four trolleys in the xyz directions. After determining the alignment method of the moving block, based on the actual position of the trolley before movement obtained in the above step S1, combined with the precise rotation matrix and the precise translation matrix solved in the above step S2, calculate the distance that each degree of freedom of the trolley needs to move during the block docking process through inverse kinematics, and finally use the fifth-order polynomial fitting method for the trolley to perform trajectory planning to obtain the trolley motion planning trajectory.

[0109] The trajectory planning of the ship's mobile sub-assembly adopts the step-by-step attitude adjustment method. In the first step, the mobile sub-assembly is rotated by an angle α around the x-axis to move the ship's mobile sub-assembly to a position horizontal to the x-axis. The mobile sub-assembly is rotated by an angle β around the y-axis to move the ship's mobile sub-assembly to a position parallel to the y-axis. The mobile sub-assembly is rotated by an angle γ around the z-axis to move the ship's mobile sub-assembly to a position parallel to the z-axis, and at the same time, the translation amounts in the x and z degrees of freedom directions are adjusted until the mobile sub-assembly reaches the target position in the first stage. In the second step, the translation in the y direction of the mobile sub-assembly is adjusted. According to the actual position of the trolley target ball (i.e., the actual position before the trolley moves) measured above and the specific alignment method of the mobile sub-assembly, the distances that the trolley needs to move in the three degrees of freedom are solved by inverse kinematics, and then the trajectory of the trolley is planned by the five-degree polynomial fitting trajectory method. The two-stage alignment method of the mobile sub-assembly avoids interference due to the presence of the sub-assembly pipeline. After the trajectory planning is completed, through the visualization function of the upper computer software, it is judged whether interference will occur during the adjustment of the mobile sub-assembly. If interference occurs to the mobile sub-assembly, the trolley trajectory planning strategy is changed by reselecting different path points; if no interference occurs, the specific trajectory of the trolley is determined as the trolley trajectory solved by the current five-degree polynomial fitting method, and finally the trolley movement is controlled by the PLC. Solving the trolley trajectory by the five-degree polynomial fitting method makes the trolley speed and acceleration smoother, avoiding problems such as errors caused by sudden changes in acceleration and damage to the motor.

[0110] S4. After determining the motion planning trajectory of the trolley under the premise of no interference for the overall moving section, predict and compensate for the errors caused by the deformation of the bottom plate of the overall moving section, and compensate for the above errors by adjusting the trolley moving distance difference and the motion trajectory. Since the overall masses of different overall moving sections are not the same, when the trolley supports the bottom of the overall moving section, different types of overall moving sections with different weights will cause different degrees of deformation of the ship bottom and the trolley. Different accelerations and speeds of trolley adjustment will also cause slight deformation of the supporting parts of the overall moving section during the adjustment process. The compensation of the deformation amount is realized through a BP-neural network. Firstly, by controlling the adjustment speed and acceleration, the assembly environment, and the material of the bottom plate of the overall moving section, measure, record, and conduct a large number of trainings on the deformation amount data of the bottom plate of the overall moving section caused by different weights of the ship's overall moving section. Secondly, measure, record, and conduct a large number of trainings on the deformation amount data of the bottom plate of the overall moving section caused by different adjustment speeds and accelerations provided by different types of overall sections, and then obtain the errors caused by the deformation of the bottom plate of the overall moving section under different factors for different ship types. The distance that the trolley needs to compensate for the deformation of the overall moving section plate is output through the neural network. By adjusting the actual motion trajectory of the trolley, compensate for the errors caused by the deformation of the bottom plate of the ship's overall moving section. Train the BP-neural network through historical data. The BP-neural network includes an input layer, three hidden layers, and an output layer. Through training and prediction, obtain the error values of deformation caused by different speeds, accelerations, and different weights under the same decision, so as to compensate for the moving amounts of each degree of freedom of the trolley before the trolley motion adjustment, recalculate the trolley motion trajectory in step S3, and obtain the compensated path docking trajectory.

[0111] Specifically, a BP-neural network is used to predict the docking error caused by the plate deformation of ship-shaped mobile sub-assemblies of different weights under the same acceleration and speed. Under the same environment, the same speed and acceleration, and the condition that the support positions of the trolleys on the sub-assembly are the same, the controlled variable is the weight of the ship-shaped mobile sub-assembly. By extracting on-site data, the deformation amount of the bottom plates caused by mobile sub-assemblies of different weights is predicted, and thus the docking error is obtained. The error is resolved by the host computer software to obtain the compensation amounts in the three degrees of freedom directions of each trolley. During the process of controlling the movement of the trolleys, the adjusted values are added to the compensation amounts, thereby reducing the number of alignment times and making the docking of the ship sub-assembly more efficient. Similarly, by keeping the weight of the ship-shaped mobile sub-assembly, the trolley positions, and the docking environment unchanged, the docking deviation of the sub-assembly at different accelerations and speeds is predicted. Through the resolution of the host computer, the movement values of the three degrees of freedom of each trolley are compensated during the process of controlling the movement of the trolleys. Among them, in this error compensation system, there are an input layer, a hidden layer, and an output layer of the BP-neural network. The input layer includes five neurons for selecting weight, speed, acceleration, displacement, and material properties; the hidden layer is set to three layers and includes multiple neurons; the output layer has one neuron, which is the displacement compensation amount in each direction of each trolley.

[0112] S5. Finally, when adjusting and controlling, use the second laser tracker to measure the trolley target balls in the three degrees of freedom directions of each trolley, and at the same time use a laser rangefinder to measure the moving distances in the three degrees of freedom directions of each trolley. After the measurements of the two devices, by processing the measurement data, the adjustment state of the trolley is controlled for real-time feedback until the actual adjustment of the trolley is consistent with the target adjustment data, that is, the actual path docking trajectory of the trolley is consistent with the compensated path docking trajectory in step S4.

[0113] Specifically, during the movement of the trolley, the laser rangefinder measures the moving distances in the three degrees of freedom planes of the trolley in real time. After the first-stage adjustment of the mobile sub-assembly is completed, the adjustment process is paused. The third laser tracker and the fourth laser tracker of the sub-assembly are used to measure the trolley target balls set on the three moving planes of the four adjustable trolleys, and then the measured data is fused with the data measured by the laser rangefinder, and the average value is taken to ensure the correctness of the data. The difference is compared with the theoretically required moving distance of the trolley, and this difference is superimposed on the adjustment process of the second stage of the mobile sub-assembly, and then the moving position of the trolley is fed back. Further adjust the rotation of the mobile sub-assembly around xyz and the translation in the xz direction; judge whether the two sub-assemblies reach the target position of the first stage. If not, readjust the rotation of the mobile sub-assembly around xyz and the translation in the xz direction until the two sub-assemblies reach the target position of the first stage; if so, then adjust the translation of the mobile sub-assembly in the y direction.

[0114] The trolley is controlled by a PLC. Since the total section of the ship is large in mass and inertia and has high docking requirements, target balls are set at fixed positions for adjusting directions in each degree of freedom. A laser tracker is used to measure the target balls of the trolley in each degree of freedom direction. Laser rangefinders are set in each degree of freedom direction of the trolley to measure the moving distances in each degree of freedom. The measurement data of the laser tracker and the laser rangefinders are processed. By taking the average value of the two sets of data, the measurement error caused by a single measurement system is reduced. Combining with the theoretically required moving distance of the trolley obtained by the low four-step method, the error value between the actual moving distance and the theoretical distance of the trolley is calculated to provide feedback guidance for the movement of the trolley. When the laser tracker measures that a certain trolley target ball is blocked, the measurement data of the laser rangefinder is used as a supplement to prevent the loss of measurement data, enhancing the robustness of obtaining the data for feedback guidance of the trolley movement. For the strokes of each degree of freedom that have not moved to the determined positions, data feedback and guidance are carried out.

[0115] Although the laser tracker has high precision, there may be occlusion phenomena during measurement. Therefore, when it is difficult to measure a certain trolley target, the data measured by the laser rangefinder is used for supplementation to ensure the integrity of the data. And when the trolley has power-off, failure and other situations, the trolley feedback measurement device can record the current position. The recorded data is stored and calculated by the upper computer. After the trolley is powered on or the failure is eliminated, the movement control, adjustment and detection of the trolley continue until the docking of the total section is completed, thus realizing the high-precision and automatic docking of the total section.

[0116] The laser rangefinder and the laser tracker are used for joint measurement feedback of the trolley's movement in each direction. For situations such as occlusion, power-off, and failure, it can also ensure the correct feedback and recording of data and has good robustness.

[0117] A high-precision measurement and automatic docking system suitable for the ship's total section without allowance of the present invention includes:

[0118] A data acquisition and processing module, which is used to calibrate the measurement equipment according to the set common reference point and calculate the position of the measurement equipment in the global coordinate system. The measurement equipment is used to measure the target ball coordinates of each key point on the docking surfaces of the fixed total section and the moving total section, and at the same time measure the plane position coordinates of the trolley's three degrees of freedom. And the coordinates of each key point and the plane position coordinates of the trolley's three degrees of freedom are both converted to the global coordinate system by the singular value decomposition method (SVD).

[0119] The rough matching and precise matching module is used to draw circles with the key points of each moving section as the centers and a preset distance as the radius, calculate the weights of the key points of each moving section according to the number of pipelines in each circle; sort the key points of each moving section in ascending order of weights, randomly select three key points with the largest weights and not collinear to form a plane, find the centroid of this plane, according to the idea of random combination, select three key points with the largest weight sum and not collinear except the above combination to form a plane, find the centroid of this plane, and so on to find the centroids corresponding to the key points of the moving section; find the centroids corresponding to the key points on the fixed section in the same way, select three pairs of corresponding centroids on the moving section and the fixed section and use the three-point alignment method. According to the three pairs of corresponding centroids selected on the moving section and the fixed section, establish the docking surface coordinate system of the moving section and the docking surface coordinate system of the fixed section respectively, solve the docking surface coordinate system of the moving section and the docking surface coordinate system of the fixed section, find the initial rotation matrix and the initial translation matrix, and realize the rough matching of the moving section and the fixed section; substitute the initial rotation matrix and the initial translation matrix into the Iterative Closest Point (ICP) algorithm to solve the precise rotation matrix and the precise translation matrix, and realize the precise matching;

[0120] The path docking trajectory planning module determines the position of the trolley relative to the moving section. The docking method of the moving section includes two stages. The first stage: adjust the three-degree-of-freedom xyz displacement and the rotation around the xz axis of the moving section; the second stage: adjust the y-direction displacement of the moving section; according to the obtained position coordinates of the trolley's three-degree-of-freedom moving plane in the global coordinate system, as well as the obtained precise rotation matrix and precise translation matrix, use inverse kinematics to solve the distances that the trolley needs to move in each degree of freedom, and use the five-degree polynomial fitting method to plan the path docking trajectory of the trolley according to this distance data.

[0121] The deformation compensation module is used to predict the deformation error of the bottom plate of the moving section generated at the connection between the trolley and the moving section during the path docking adjustment by using a BP neural network. The BP neural network is trained through historical data. The BP neural network includes an input layer, three hidden layers, and an output layer. Through training and prediction, the error values of deformation caused by different speeds, accelerations, and different weights under the same decision are obtained, so as to compensate the moving amounts of each degree of freedom of the trolley during the path docking adjustment of the trolley, and obtain the path docking trajectory after deformation compensation;

[0122] The motion feedback module is used to measure the trolley target ball coordinates in the three-degree-of-freedom directions of each trolley using a measuring device when controlling the trolley to adjust, and at the same time use the measuring device to measure the plane distances of the three-degree-of-freedom directions of each trolley's movement. The target ball coordinates in the three-degree-of-freedom directions of each trolley measured by the measuring device and the distance data of the three-degree-of-freedom directions of each trolley's movement are fused and processed to provide real-time feedback on the adjustment state of the trolley until the actual path docking trajectory of the trolley is consistent with the path docking trajectory after deformation compensation.

[0123] Preferably, the measuring device includes a first laser tracker, a second laser tracker, and a laser rangefinder. Target balls are respectively installed at the key point positions and on the three-degree-of-freedom moving plane of the trolley. The three-degree-of-freedom moving plane of the trolley includes the lower bottom surface, the middle layer surface, and the upper layer surface. The first laser tracker is used to measure the coordinates of the target balls at each key point, and the second laser tracker is used to measure the coordinates of the trolley target ball on the trolley. The laser rangefinder is used to measure the moving distances of the trolley on the three-degree-of-freedom direction planes.

[0124] The number of the first laser trackers is ensured to be able to measure all the target balls at the key points, and the position of each first laser tracker is ensured to be able to measure the most target balls at the key points.

[0125] The number of the second laser trackers is ensured to be able to measure all the target balls on the three-degree-of-freedom moving plane of the trolley, and the position of each second laser tracker is ensured to be able to measure the most target balls on the three-degree-of-freedom moving plane of the trolley.

[0126] A device of the present invention includes a memory and a processor, wherein:

[0127] The memory is used to store a computer program that can run on the processor.

[0128] The processor is used to execute the steps of a high-precision measurement and automatic docking method adapted to the ship's zero-tolerance block as described above when running the computer program.

Claims

1. A high-precision measurement and automatic docking method adapted to a ship's zero-tolerance sub-assembly, characterized in that, It includes the following steps: S1. Fix the fixed total section, place the movable total section on the adjustable trolley, set common reference points on the foundations on both sides of the fixed total section and the movable total section, set corresponding key points on the mating surfaces of the fixed total section and the movable total section, install measuring equipment. The measuring equipment includes a first laser tracker, a second laser tracker and a laser rangefinder, calibrate the measuring equipment through the common reference points, calculate the positions of the measuring equipment in the global coordinate system, measure the coordinates of each key point with the first laser tracker, and measure the position coordinates of the three-degree-of-freedom moving plane of the trolley with the second laser tracker; convert the coordinates of each key point and the position coordinates of the three-degree-of-freedom moving plane of the trolley into the global coordinate system; S2. Coarsely match the fixed total section and the movable total section, and obtain the initial rotation matrix and the initial translation matrix of the key point sets of the mating surfaces of the fixed total section and the movable total section. Further obtain the accurate rotation matrix and the accurate translation matrix by using the iterative closest point (ICP) algorithm; including: draw circles with the key points of the movable total section as the centers and a preset distance as the radii, calculate the weights of the key points of the movable total section according to the number of pipelines in each circle; select three key points with the largest sum of weights and non-collinear to form a plane, obtain the centroid of this plane, according to the idea of random combination, select three key points with the largest sum of weights and non-collinear except the above combination to form a plane, obtain the centroid of this plane, and so on to obtain the centroids corresponding to the key points of the movable total section; calculate the centroids corresponding to the key points on the fixed total section in the same way, select three pairs of corresponding and non-collinear centroids on the fixed total section and the movable total section, respectively establish the coordinate system of the mating surface of the movable total section and the coordinate system of the mating surface of the fixed total section by the three-point alignment method, and solve the two coordinate systems to obtain the initial rotation matrix and the initial translation matrix, realizing the coarse matching of the movable total section and the fixed total section; substitute the initial rotation matrix and the initial translation matrix into the iterative closest point (ICP) algorithm to solve the accurate rotation matrix and the accurate translation matrix; S3. Determine the alignment method of the movable total section. The alignment method of the movable total section includes two stages. The first stage: adjust the three-degree-of-freedom xyz displacement and the rotation direction around xz of the movable total section; the second stage: adjust the y-direction displacement of the movable total section; then, according to the position coordinates of the three-degree-of-freedom moving plane of the trolley in the global coordinate system obtained in step S1, determine the position of the trolley relative to the movable total section, combine the accurate rotation matrix and the accurate translation matrix obtained in step S2, solve the distances that each degree of freedom of the trolley needs to move through inverse kinematics, and perform path docking trajectory planning on the trolley by using the fifth-order polynomial fitting method to obtain the trolley motion planning trajectory; S4. Use a BP neural network to predict the deformation error generated at the connection between the trolley and the moving overall section during the path docking adjustment process. Train the BP neural network with historical data. The BP neural network includes an input layer, three hidden layers, and an output layer. Through training and prediction, obtain the error values of the deformation caused to the bottom structure of the moving overall section under different speeds, accelerations, and different weights under the same decision, so as to compensate the distances that each degree of freedom of the trolley needs to move obtained in step S3 during the path docking adjustment process of the trolley, recalculate the motion planning trajectory of the trolley in step S3, and thus obtain the compensated path docking trajectory; S5. When controlling the adjustment of the trolley, the second laser tracker measures the coordinates of the trolley target balls on each trolley. At the same time, the laser rangefinder measures the moving distances of each trolley in the three-degree-of-freedom direction. The data measured by the second laser tracker and the laser rangefinder are fused and processed to provide real-time feedback on the adjustment state of the trolley until the actual path docking trajectory of the trolley is consistent with the compensated path docking trajectory in step S4.

2. The high-precision measurement and automatic docking method adapted to the ship block without allowance according to claim 1, wherein In step S1, target balls are installed at each key point position and on the three-degree-of-freedom moving plane of the trolley. The three-degree-of-freedom moving plane of the trolley includes the lower bottom surface, the middle layer surface, and the upper layer surface; the first laser tracker is used to measure the coordinates of the target balls at each key point, and the second laser tracker is used to measure the coordinates of the trolley target balls on the three-degree-of-freedom moving plane of the trolley; The number of the first laser trackers is ensured to be able to measure all the target balls at the key points, and the position of each first laser tracker is ensured to be able to measure the most target balls at the key points; The number of the second laser trackers is ensured to be able to measure all the target balls on the three-degree-of-freedom moving plane of the trolley, and the position of each second laser tracker is ensured to be able to measure the most target balls on the three-degree-of-freedom moving plane of the trolley.

3. A high-precision measurement and automatic docking method adapted to a ship's zero-tolerance sub-assembly, characterized in that The specific method of rough matching in step S2 is: (101) Calculate the weights of each key point on the moving total section. Draw circles with the distance of half of the deck width of the ship's total section as the radius, centered at each key point. Then the weight \(w\) of the \(i\)-th key point on the moving total section is i as follows: Among them, M is the total number of mobile total-section pipelines, N i is the circle R with the i-th key point on the mobile total section as the center i The number of pipelines containing the center of the pipeline, where i is the number of the i-th key point on the mobile total section; the set of weights of all key points on the mobile total section is W = {w1, …, w N}, N is the total number of key points on the mobile total section; (102) From the set of weights \(W = \{w_1, \ldots, w\}\) of all key points on the mobile segment N}, first select three key points with the largest sum of weights and non - collinear. Connect the three points pairwise head - to - tail to form a planar triangle, and calculate the centroid of the triangle as \(g_1\). According to the concept of random combination, select three key points with the largest sum of weights and non - collinear except the above combination. Connect the three points pairwise head - to - tail to form a planar triangle, and calculate the centroid of this triangle as \(g_2\). Similarly, calculate the centroids corresponding to the three key points of all possible combinations of all key points on the mobile segment in this way where \(H\) is the number of groups where the three selected key points are collinear; (103) Calculate the centroids corresponding to the key points on the fixed total segment in the same way. The three non - collinear centroids g j , g k , g l on the moving total segment and the corresponding three centroids g j ′, g′ k , g l ′ on the fixed total segment are docked and matched using the three - point alignment method to achieve the rough matching of the moving total segment and the fixed total segment.

4. A high-precision measurement and automatic docking method adapted to a ship's zero-tolerance block as claimed in claim 3, characterized in that, The calculation methods of the accurate rotation matrix and the accurate translation matrix in step S2 are: (201) Establish the docking surface coordinate system of the moving overall section and the docking surface coordinate system of the fixed overall section; Take g j as the coordinate origin, and take the direction vector from g j to g k as the x-axis to construct a unit vector: Then, determine the y-axis by taking the cross product of the direction vector from g j to g l with the x-axis direction vector, and construct the unit vector of the y-axis: Finally, determine the z-axis direction by taking the cross product of e1 and e2, and its unit vector is: e3 = e1 × e2 Then the coordinate system of the moving total segment docking surface is: O g -e1, e2, e3; Similarly, the fixed total segment docking surface coordinate system is determined as: O g′ -e1′, e′2, e3′; (202) Calculate the initial rotation matrix and the initial translation matrix; Through the three-point alignment method for matching the measured data, the initial rotation transformation matrix R0 of the key point sets on the docking surface of the moving overall section and the docking surface of the fixed overall section is expressed as: R0 = [e′1, e′2, e′3] * [e1, e2, e3] T According to the initial rotation matrix R0, calculate the initial translation matrix t0 as: t0 = g j '-g j R0 (203) Substitute the initial rotation matrix R0 and the initial translation matrix t0 into the iterative closest point ICP algorithm, and through iterative calculation, solve the accurate rotation matrix R and the accurate translation matrix t.

5. A high-precision measurement and automatic docking method adapted to a ship's zero-tolerance sub-assembly, characterized in that, In step S3, the path docking trajectory planning of the moving overall section adopts the step-by-step posture adjustment method, specifically: (31) Adjust the rotation angle of the moving overall section around the x direction by rotating α, so that the ship's moving overall section moves to a position horizontal with the x axis, adjust the rotation angle of the moving overall section around the y direction by rotating β, so that the ship's moving overall section moves to a position parallel to the y axis, adjust the rotation angle of the moving overall section around the z direction by rotating γ, so that the ship's moving overall section moves to a position parallel to the z axis, and at the same time adjust the translation amounts in the x and z two-degree-of-freedom directions; When the distance between the mobile sub-section and the fixed sub-section reaches the preset distance threshold, adjust the translation in the y direction to avoid interference during the adjustment process of the mobile sub-section due to the presence of the sub-section pipeline. The quintic polynomial trajectory method is used to plan the docking trajectory of the trolley path.

6. The high-precision measurement and automatic docking method for a ship block without allowance according to claim 1, characterized in that Step S4 is specifically as follows: (41) Use a BP neural network to predict the docking error caused by the plate deformation caused by each mobile sub-section of different ship types with different weights under the same acceleration and speed. In the same assembly environment, with the same speed and acceleration, and the trolley in the same support position of the sub-section, the control variable is the weight of the ship's mobile sub-section. By extracting on-site data, predict the docking error caused by mobile sub-sections of different weights, calculate the docking error, obtain the compensation amounts in the three degrees of freedom directions of each trolley, and add the compensation amounts to the docking trajectory of the trolley path planned in step S3 to adjust the docking trajectory of the trolley path; (42) Control the weight of the ship's mobile sub-section, keep the trolley position and assembly environment unchanged, so as to predict the sub-section docking deviation at different accelerations and speeds, and compensate the three-degree-of-freedom motion trajectories of each trolley. The input layer of the BP neural network includes five neurons for selecting weight, speed, acceleration, displacement, and material properties; the hidden layer is set to 3 layers and includes multiple neurons; the output layer has one neuron, which is the displacement compensation amount in each direction of each trolley.

7. A high-precision measurement and automatic docking method adapted to a ship's zero-tolerance subassembly, characterized in that, Step S5 is specifically as follows: The second laser tracker measures the trolley target ball set on the three-degree-of-freedom movement plane of the trolley. At the same time, a laser rangefinder is used to measure the movement of the trolley in three degrees of freedom, and the data measured by the second laser tracker and the laser rangefinder are fused. When the second laser tracker measures that a certain trolley target ball is blocked, data is supplemented by the laser rangefinder; when the trolley has a power failure or a fault, the trolley feedback measurement device records the current position of the trolley; when the trolley is powered on or the fault is eliminated, continue to control the docking trajectory of the trolley path until the docking of the mobile sub-section and the fixed sub-section is completed.

8. A high-precision measurement and automatic docking system adapted to ship block without allowance, characterized in that, It includes: A data acquisition and processing module, which is used to calibrate the measurement device according to the set common reference point, calculate the position of the measurement device in the global coordinate system, measure the target ball coordinates of each key point on the docking surface of the fixed sub-section and the mobile sub-section, and measure the position coordinates of the three-degree-of-freedom movement plane of the trolley at the same time. And use the singular value decomposition method (SVD) to convert the coordinates of each key point and the position coordinates of the three-degree-of-freedom movement plane of the trolley into the global coordinate system; The rough matching and precise matching module is used to draw circles with the key points of each moving total segment as the centers and a preset distance as the radius, calculate the weights of the key points of each moving total segment according to the number of pipelines in each circle; sort the key points of each moving total segment in ascending order of weight, randomly select three key points with the largest weights and not collinear to form a plane, calculate the centroid of this plane, according to the concept of random combination, select three key points with the largest weight sum and not collinear except the above combination to form a plane, calculate the centroid of this plane, and so on to calculate the centroids corresponding to the key points of the moving total segment; calculate the centroids corresponding to the key points on the fixed total segment in the same way, select three pairs of corresponding centroids on the moving total segment and the fixed total segment and use the three-point alignment method. According to the three pairs of corresponding centroids selected on the moving total segment and the fixed total segment, establish the docking surface coordinate system of the moving total segment and the docking surface coordinate system of the fixed total segment respectively, solve the moving total segment docking surface coordinate system and the fixed total segment docking surface coordinate system to obtain the initial rotation matrix and the initial translation matrix, and realize the rough matching of the moving total segment and the fixed total segment; substitute the initial rotation matrix and the initial translation matrix into the Iterative Closest Point (ICP) algorithm to solve the precise rotation matrix and the precise translation matrix, and realize the precise matching; The path docking trajectory planning module determines the position of the trolley relative to the moving total segment. The docking method of the moving total segment includes two stages. The first stage: adjust the three-degree-of-freedom xyz displacement and the rotation direction around the xz axis of the moving total segment; the second stage: adjust the y-direction displacement of the moving total segment; according to the obtained position coordinates of the trolley's three-degree-of-freedom moving plane in the global coordinate system, as well as the obtained precise rotation matrix and precise translation matrix, use inverse kinematics to solve the distances that the trolley needs to move in each degree of freedom, and use the five-degree polynomial fitting method to plan the path docking trajectory of the trolley according to this distance data; The deformation compensation module is used to predict the deformation error of the bottom plate of the moving total segment generated at the connection between the trolley and the moving total segment during the path docking adjustment by using a BP neural network. The BP neural network is trained with historical data. The BP neural network includes an input layer, three hidden layers, and an output layer. Through training and prediction, the error values of deformation caused by different speeds, accelerations, and different weights under the same decision are obtained, so as to compensate the moving amounts of each degree of freedom of the trolley during the path docking adjustment of the trolley, and obtain the path docking trajectory after deformation compensation; The motion feedback module is used to measure the target ball coordinates of the trolley in the three-degree-of-freedom directions of each trolley using a measuring device when controlling the adjustment of the trolley, and at the same time use the measuring device to measure the plane distances of the three-degree-of-freedom directions of each trolley. The target ball coordinates of the three-degree-of-freedom directions of each trolley measured by the measuring device and the distance data of the three-degree-of-freedom directions of each trolley are fused and processed to provide real-time feedback on the adjustment state of the trolley until the actual path docking trajectory of the trolley is consistent with the path docking trajectory after deformation compensation.

9. The high-precision measurement and automatic docking system adapted to the ship's zero-tolerance block according to claim 8, characterized in that The measuring device includes a first laser tracker, a second laser tracker and a laser rangefinder. Target balls are respectively installed at the key point positions and on the three-degree-of-freedom moving plane of the trolley. The three-degree-of-freedom moving plane of the trolley includes the lower bottom surface, the middle layer surface and the upper layer surface. The first laser tracker is used to measure the coordinates of the target balls at each key point, and the second laser tracker is used to measure the coordinates of the trolley target ball on the trolley. The laser rangefinder is used to measure the moving distances of the trolley on the three-degree-of-freedom direction planes. The number of the first laser trackers ensures that all the target balls at the key points can be measured, and the position of each first laser tracker ensures that the maximum number of target balls at the key points can be measured. The number of the second laser trackers ensures that all the target balls on the three-degree-of-freedom moving plane of the trolley can be measured, and the position of each second laser tracker ensures that the maximum number of target balls on the three-degree-of-freedom moving plane of the trolley can be measured.

10. A device, characterized in that, It includes a memory and a processor, wherein: The memory is used to store a computer program that can run on the processor. The processor is used to execute the steps of a high-precision measurement and automatic docking method adapted to the ship's zero-tolerance block as described in any one of claims 1-7 when running the computer program.

Images