Method for seam finding and adaptive deviation correction in intersecting-line welding of TKY tubular joints

Abstract

A method for seam finding and adaptive deviation correction in intersecting-line welding of TKY tubular joints, comprising the following steps: collecting marine environmental data, establishing a wave slope reflection disturbance potential function on the basis of the marine environmental data, and generating a laser scanning error distribution field by means of calculation; on the basis of welding seam point cloud data generated by means of laser scanning of a robot, in combination with the laser scanning error distribution field, screening for outliers in the welding seam point cloud data to form a point cloud data set to be corrected; defining a point cloud correction objective energy functional, and by incorporating the wave slope reflection disturbance potential function, iteratively optimizing said point cloud data set to obtain corrected welding seam point cloud data; and on the basis of the corrected welding seam point cloud data, controlling the robot to execute automatic welding. The method for seam finding and adaptive deviation correction in intersecting-line welding of TKY tubular joints solves the actual problem of errors occurring in welding seam point clouds caused by sunlight reflection on the sea surface.

Description

A method for TKY pipe joint intersection welding positioning and adaptive correction Technical Field

[0001] This invention relates to the field of automated welding technology in marine engineering, and more specifically, to a method for positioning and adaptive correction of welding intersection lines of TKY pipe nodes. Background Technology

[0002] The TKY pipe node intersection welding positioning and adaptive correction method aims to improve weld identification accuracy and reduce point cloud errors in the marine environment. By constructing a wave slope reflection disturbance potential function and combining it with the laser scanning error distribution field to correct point cloud data, the method controls the weld positioning accuracy and trajectory adjustment stability of the welding robot under seawater wave interference, and realizes adaptive correction of weld point cloud data in a dynamic marine environment.

[0003] Existing methods for TKY pipe node intersection welding positioning and adaptive correction often struggle to distinguish the effects of dynamic tilting of seawater waves on laser scanning errors caused by specular reflection of sunlight. Furthermore, in marine environments, wave undulations create a specific slope that reflects sunlight into the weld seam, and strong sunlight can blur the edge clarity of the weld seam, leading to errors when the welding robot scans the weld seam point cloud in real time. Therefore, this paper proposes a method for TKY pipe node intersection welding positioning and adaptive correction. Summary of the Invention

[0004] The purpose of this invention is to provide a method for TKY pipe node intersection line welding positioning and adaptive correction, in order to solve the problem mentioned in the background art that, in the marine environment, the wave fluctuations form a specific slope that reflects sunlight into the weld seam, and the strong sunlight will dilute the clarity of the weld seam edge, causing errors when the welding robot scans the weld seam point cloud in real time.

[0005] To achieve the above objectives, the present invention aims to provide a method for TKY pipe node intersection line welding positioning and adaptive correction, comprising the following steps:

[0006] S1. Use sensors to collect marine environmental data, establish the wave slope reflection disturbance potential function based on the marine environmental data, and calculate and generate the laser scanning error distribution field;

[0007] S2. Based on the weld point cloud data generated by robot laser scanning, and combined with the laser scanning error distribution field to filter out abnormal points in the weld point cloud data, a point cloud dataset to be corrected is formed.

[0008] S3. Define the target energy functional for point cloud correction, and combine it with the wave slope reflection perturbation potential function to iteratively optimize the point cloud dataset to be corrected, and obtain the corrected weld point cloud data.

[0009] S4. The robot performs automatic welding based on the corrected weld point cloud data.

[0010] As a further improvement to this technical solution, the marine environmental data includes wave height, wave tilt angle, wave period, wave propagation speed, solar azimuth angle, solar intensity, and seawater refractive index.

[0011] The wave slope reflection disturbance potential function is constructed based on the nonlinear wave equation and takes into account the influence of the reflection of sunlight on the welding point by the wave slope. It is used to describe the dynamic interference caused by the reflection of sunlight by the wave slope on the scanning welding point.

[0012] As a further improvement to this technical solution, in step S1, a nonlinear wave equation is used to establish the wave slope reflection disturbance potential function, and the laser scanning error distribution field is calculated. The specific method steps are as follows:

[0013] S1.1. Based on marine environmental data, define and construct the wave slope reflection disturbance potential function;

[0014] S1.2 Based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle;

[0015] The laser scanning error distribution field is used to quantitatively describe the laser scanning error caused by the reflection of sunlight from the inclined surface of seawater waves, and to identify and correct error point data.

[0016] As a further improvement to this technical solution, in S1.1, based on marine environmental data, a wave slope reflection disturbance potential function is defined and constructed, as follows:

[0017] Define the wave slope reflection perturbation potential function as follows: :

[0018] ;

[0019] Where t is time; x is the horizontal coordinate of the wave surface; y is the vertical coordinate of the wave surface perpendicular to the propagation direction; z is the height coordinate of the wave surface; and c is the wave propagation speed. Let be the second-order partial derivative of the wave slope reflection perturbation potential function with respect to time; Here, C is the Laplace operator; C is the wave propagation speed. These are higher-order nonlinear coupling coefficients; The nonlinear damping coefficient; Let be the modulus of the wave potential energy gradient; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time.

[0020] In step S1.2, based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle, as follows:

[0021] ;

[0022] in, This represents the laser scanning error distribution field. Let be the partial derivative of the wave slope reflection disturbance potential function with respect to height; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time; This is the influence coefficient of the wave tilt angle; This represents the time-dynamic influence coefficient.

[0023] As a further improvement to this technical solution, in step S2, the weld point cloud data generated by robot laser scanning is combined with the laser scanning error distribution field to filter out abnormal points in the weld point cloud data, forming a point cloud dataset to be corrected. The specific method steps are as follows:

[0024] S2.1 The weld point cloud data generated by the robot laser scanning is as follows:

[0025] ;

[0026] Where M is the total number of point data in the weld point cloud data; Index for point data; For point data Horizontal coordinates; Point data Coordinates perpendicular to the direction of propagation; For point data Elevation coordinates; For point data 3D coordinates; The weld point cloud data collected at time t;

[0027] S2.2 Use the laser scanning error distribution field to filter out the abnormal points in the weld point cloud data and form a point cloud dataset to be corrected.

[0028] As a further improvement to this technical solution, in step S2.2, the abnormal points in the weld point cloud data are screened out using the laser scanning error distribution field to form a point cloud dataset to be corrected. The specific method steps are as follows:

[0029] S2.2.1 Calculate the theoretical error of each point in the weld point cloud data based on the laser scanning error distribution field:

[0030] ;

[0031] in, For point data Theoretical error; The effect of changes in wave height on the error; The impact of dynamic changes over time on the error; Point data for the oblique surface of the ocean wave The effect of tilt at the location; Point data of wave disturbances in the time dimension The impact of location;

[0032] S2.2.2 Setting the error threshold The screening error is greater than the error threshold. The point data constitutes the point cloud dataset to be corrected:

[0033] ;

[0034] in, This is a point cloud dataset that needs to be corrected.

[0035] As a further improvement to this technical solution, the point cloud correction target energy functional is constructed based on the wave slope reflection perturbation potential function combined with the point cloud dataset to be corrected, and is used to correct the weld point cloud data error caused by the reflection of sunlight by sea waves.

[0036] As a further improvement to this technical solution, in step S3, a target energy functional for point cloud correction is defined, and combined with the wave slope reflection perturbation potential function, the point cloud dataset to be corrected is iteratively optimized to obtain the corrected weld point cloud data. The specific method steps are as follows:

[0037] S3.1 Define the target energy functional for point cloud correction;

[0038] S3.2. Use the gradient descent algorithm to calculate the gradient descent direction of the point cloud data to be corrected;

[0039] S3.3. Combine the wave slope reflection disturbance potential function to correct the error point and calculate the offset correction amount of the error point;

[0040] S3.4. Based on the offset correction amount of the error points, the final correction formula is obtained;

[0041] S3.5 Iteratively execute S3.1-S3.4 to optimize all point cloud data to be corrected, and finally obtain the corrected weld point cloud data.

[0042] As a further improvement to this technical solution, in step S3.1, a point cloud correction target energy functional is defined, and the specific method is as follows:

[0043] ;

[0044] in, To correct the target energy functional of the point cloud; Q is the point cloud dataset to be corrected, equivalent to... ; The current coordinates of the point data to be corrected; These are standard points in an ideal weld model; Suppress error terms; For smoothing regularization; The deviation weighting coefficient between the point cloud data and the ideal weld model; These are the error field weighting coefficients; The weights for point cloud smoothing regularization; To optimize the number of iterations; K is the total number of error points in the point cloud dataset to be corrected;

[0045] In step S3.2, the gradient descent algorithm is used to calculate the gradient descent direction of the point cloud data to be corrected. The specific method is as follows:

[0046] ;

[0047] in, These are the partial derivatives used in the gradient descent algorithm.

[0048] ;

[0049] in, This represents the number of iteration rounds. For the first Point cloud coordinates in each iteration; For the first Point cloud coordinates in each iteration; This is the learning rate.

[0050] As a further improvement to this technical solution, in step S3.3, the error point is corrected by combining the wave slope reflection disturbance potential function, and the offset correction amount of the error point is calculated. The specific method is as follows:

[0051] ;

[0052] in, For point data The offset correction amount; This represents the influence coefficient of seawater ripple disturbance in the horizontal direction; The wave disturbance impact coefficient over time; The influence of the wave slope reflection disturbance potential function on the horizontal direction of the point cloud; The influence of the wave slope reflection disturbance potential function in the time dimension;

[0053] In step S3.4, the final correction formula is obtained based on the offset correction amount of the error point. The specific method is as follows:

[0054] ;

[0055] In step S3.5, steps S3.1-S3.4 are executed iteratively to optimize all point cloud data to be corrected, ultimately obtaining the corrected weld point cloud data. The specific method is as follows:

[0056] ;

[0057] in, To correct the weld point cloud data

[0058] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0059] 1. In the TKY pipe node intersection welding positioning and adaptive correction method, based on the joint modeling of the wave slope reflection disturbance potential function and the laser scanning error distribution field, the point cloud error caused by the dynamic tilt of seawater waves and the interference of sunlight mirror reflection is identified, and the weld point cloud data is corrected in real time.

[0060] 2. In the TKY pipe node intersection welding positioning and adaptive correction method, the point cloud is modified by constructing a point cloud target energy functional, and the point cloud data that is greatly affected by the dynamic interference of sea waves is adaptively optimized to ensure the smoothness of the point cloud data in the local range and the overall consistency, so that the welding robot can accurately position itself on the actual weld trajectory. Attached Figure Description

[0061] Figure 1 is a flowchart of the overall method of the present invention. Detailed Implementation

[0062] 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, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. Example

[0063] Please refer to Figure 1. This embodiment provides a method for TKY pipe node intersection line welding positioning and adaptive correction, including the following steps:

[0064] S1. Use sensors to collect marine environmental data, establish the wave slope reflection disturbance potential function based on the marine environmental data, and calculate and generate the laser scanning error distribution field;

[0065] The marine environmental data includes wave height, wave tilt angle, wave period, wave propagation speed, solar azimuth angle, solar intensity, and seawater refractive index.

[0066] The wave slope reflection disturbance potential function is constructed based on the nonlinear wave equation and takes into account the influence of the reflection of sunlight on the welding point by the wave slope. It is used to describe the dynamic interference caused by the reflection of sunlight by the wave slope on the scanning welding point.

[0067] In this embodiment, sensors are used to collect marine environmental data, as detailed below:

[0068] The system uses lidar to acquire weld seam point cloud data and monitors the surface morphology of ocean waves in real time, including wave tilt angle and wave height. An inertial measurement unit is used to detect wave propagation speed. An ultrasonic ranging sensor is used to measure wave height and period. Installed near the robot, the sensor uses ultrasonic echoes to measure water surface height and analyze wave undulation. An ambient light sensor monitors the azimuth angle and light intensity of sunlight. An underwater optical refractive index sensor measures the refractive index of seawater and monitors the optical refractive properties of seawater to calculate laser beam path offset.

[0069] In this embodiment S1, a nonlinear wave equation is used to establish the wave slope reflection disturbance potential function, and the laser scanning error distribution field is calculated. The specific method steps are as follows:

[0070] S1.1. Based on marine environmental data, define and construct the wave slope reflection disturbance potential function;

[0071] S1.2 Based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle;

[0072] The laser scanning error distribution field is used to quantitatively describe the laser scanning error caused by the reflection of sunlight from the inclined surface of seawater waves, and to identify and correct error point data.

[0073] In step S1.1, based on marine environmental data, a wave slope reflection disturbance potential function is defined and constructed as follows:

[0074] The measured marine environmental data are as follows: wave propagation speed 1.5 m / s; wave height 1.2 m; wave period 8 s; solar azimuth angle 35°; solar intensity 600 W / m²; seawater refractive index 1.33; nonlinear damping coefficient 0.1; higher-order nonlinear coupling coefficient 0.02.

[0075] Define the wave slope reflection perturbation potential function as follows: It satisfies the following nonlinear wave equation:

[0076] ;

[0077] Where t is time; x is the horizontal coordinate of the wave surface; y is the vertical coordinate of the wave surface; and z is the height coordinate of the wave surface. Let be the second-order partial derivative of the wave slope reflection perturbation potential function with respect to time; For the Laplace operator; Let be the modulus of the wave potential energy gradient; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time;

[0078] In step S1.2, based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle, as follows:

[0079] ;

[0080] in, This represents the laser scanning error distribution field. Let be the partial derivative of the wave slope reflection disturbance potential function with respect to height; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time; This is the influence coefficient of the wave tilt angle; This represents the time-dynamic influence coefficient.

[0081]

[0082] ;

[0083] ;

[0084] At the moment of maximum wave inclination (t=2s), calculate its specific value:

[0085] ;

[0086] ;

[0087] Wave tilt angle influence coefficient:

[0088] ;

[0089] Time-based dynamic influence coefficient:

[0090] ;

[0091] Substitute specific data into the calculation:

[0092] ;

[0093] ;

[0094] The final calculation yielded a maximum error distribution value of 1.355m.

[0095] Since the error range is as large as 1.355m, it will cause a large-scale shift in the weld point cloud. Therefore, it is necessary to introduce a point cloud correction target energy functional to optimize the data collected by the welding robot and reduce its error to an acceptable range (±0.1mm).

[0096] Error point filtering threshold set to ;

[0097] In this embodiment, in a marine environment, seawater waves form a dynamic tilted surface, causing optical deviations in the laser scanning path. Therefore, the perturbation potential function of the wave surface, i.e., the perturbation potential function of the wave tilted surface reflection, is set as follows: It satisfies a nonlinear wave equation, which describes the temporal evolution of the wave surface morphology and is used to calculate the perturbation of the laser path by the waves. The laser scanning error can be calculated from the wave surface tilt angle and the sunlight reflection angle, yielding the laser scanning error distribution field. ;

[0098] In marine environments, when welding robots use laser scanning to locate weld seams, the dynamic slopes formed by ocean waves cause sunlight reflection to interfere with the laser scanning, resulting in systematic deviations in the point cloud data. In standard hydrodynamics, ocean waves can be considered as a nonlinear wave system controlled by the Laplace equation and boundary conditions. However, due to factors such as turbulence, surface tension, and viscous damping, linear wave equations cannot accurately characterize these complex phenomena. Therefore, it is necessary to introduce nonlinear wave equations. Using nonlinear wave equations can accurately simulate ocean wave dynamics without relying on empirical filtering methods, thus exhibiting greater generalization ability. This can be achieved through the laser scanning error distribution field. By calculating the error distribution, the robot point cloud data can be directly adaptively corrected to improve the positioning accuracy of the weld. Because nonlinear terms are taken into account, the model can adapt to extreme conditions such as strong winds and waves and wave breakage.

[0099] In this embodiment, K is the wave tilt angle influence coefficient, which is calculated as follows: Where n is the refractive index of seawater, and is Optical path offset; The time-dynamic influence coefficient is calculated as follows: ,in λ is the laser wavelength, and c is the wave propagation speed.

[0100] S2. Based on the weld point cloud data generated by robot laser scanning, and combined with the laser scanning error distribution field to filter out abnormal points in the weld point cloud data, a point cloud dataset to be corrected is formed.

[0101] In this embodiment S2, the weld point cloud data generated by robot laser scanning is used as the basis for selecting outliers in the weld point cloud data based on the laser scanning error distribution field, thus forming a point cloud dataset to be corrected. The specific method steps are as follows:

[0102] S2.1 The weld point cloud data generated by the robot laser scanning is as follows:

[0103] ;

[0104] Where M is the total number of point data in the weld point cloud data; Index for point data; For point data Horizontal coordinates; Point data Coordinates perpendicular to the direction of propagation; For point data Elevation coordinates; For point data 3D coordinates; The weld point cloud data collected at time t;

[0105] Robot scanning parameters:

[0106] Scanning method: Laser vision scanning

[0107] Scanning device: LiDAR + visual sensor

[0108] Scan trajectory: Line-by-line scan generated by offline programming software

[0109] Scanning accuracy: 1 point per 1mm

[0110] Scan range:

[0111] Horizontal direction : [−0.5m, 0.5m]; Vertical direction : [−0.5m, 0.5m]; height direction : [−0.1m, 0.1m];

[0112] The robot program triggers the laser vision sensor to scan the weld seam point cloud. Following the scanning trajectory preset by the offline programming software, it acquires laser point cloud data of the weld seam surface point by point, forming a complete three-dimensional weld seam point cloud dataset. .

[0113] In this embodiment, the weld point cloud data generated by robot laser scanning is a commonly used technique in the prior art. The specific method for obtaining weld point cloud data is as follows:

[0114] The robot program triggers the laser vision sensor to start positioning and scans the weld seam according to the laser vision scanning trajectory established by the offline programming software to obtain weld seam point cloud data.

[0115] S2.2 Use the laser scanning error distribution field to filter out the abnormal points in the weld point cloud data and form a point cloud dataset to be corrected.

[0116] In this embodiment S2.2, outliers in the weld point cloud data are filtered out using the laser scanning error distribution field to form a point cloud dataset to be corrected. The specific method steps are as follows:

[0117] S2.2.1 Calculate the theoretical error of each point in the weld point cloud data based on the laser scanning error distribution field:

[0118] ;

[0119] in, For point data Theoretical error; The effect of changes in wave height on the error; The impact of dynamic changes over time on the error; Point data for the oblique surface of the ocean wave The effect of tilt at the location; Point data of wave disturbances in the time dimension The impact of location;

[0120] in, The effect of wave height variation on the error is taken as 0.8; The value is 1.2 to represent the effect of dynamic changes over time on the error.

[0121] S2.2.2 Setting the error threshold The screening error is greater than the error threshold. The point data constitutes the point cloud dataset to be corrected:

[0122] ;

[0123] in, This is a point cloud dataset that needs to be corrected.

[0124] Set error threshold The maximum error is 0.5m; point cloud data with errors exceeding the limit are filtered out; the maximum error is substituted:

[0125] ;

[0126] This point is an anomaly and should be added to the point cloud dataset that needs correction.

[0127] S3. Define the target energy functional for point cloud correction, and combine it with the wave slope reflection perturbation potential function to iteratively optimize the point cloud dataset to be corrected, and obtain the corrected weld point cloud data.

[0128] The target energy functional for point cloud correction is constructed based on the wave slope reflection perturbation potential function combined with the point cloud dataset to be corrected, and is used to correct the error in weld point cloud data caused by the reflection of sunlight by seawater ripples.

[0129] In this embodiment S3, a target energy functional for point cloud correction is defined, and combined with the wave slope reflection perturbation potential function, the point cloud dataset to be corrected is iteratively optimized to obtain the corrected weld point cloud data. The specific method steps are as follows:

[0130] S3.1 Define the target energy functional for point cloud correction;

[0131] S3.2. Use the gradient descent algorithm to calculate the gradient descent direction of the point cloud data to be corrected;

[0132] S3.3. Combine the wave slope reflection disturbance potential function to correct the error point and calculate the offset correction amount of the error point;

[0133] S3.4. Based on the offset correction amount of the error points, the final correction formula is obtained;

[0134] S3.5 Iteratively execute S3.1-S3.4 to optimize all point cloud data to be corrected, and finally obtain the corrected weld point cloud data.

[0135] In this embodiment, the purpose of the point cloud correction target energy functional is to construct an optimization framework so that the error point cloud data scanned by the robot can be automatically corrected and approach the real weld morphology. Due to the wave characteristics of ocean waves, the weld point cloud data may have local offsets, distortions or outliers. The point cloud correction target energy functional quantifies the error and uses gradient descent optimization to make the error points converge toward the real weld trajectory, thereby eliminating laser scanning errors.

[0136] Traditional point cloud correction methods typically use simple interpolation or filtering to remove error points without considering the dynamic disturbance characteristics of ocean waves. This method constructs a point cloud correction target energy functional based on the wave slope reflection perturbation potential function, which can dynamically and adaptively adjust according to the ocean waves, making it more in line with the actual conditions of the marine welding environment.

[0137] In the point cloud correction process, it is not enough to simply make each point converge to its ideal position; the overall consistency of the entire point cloud data also needs to be considered. Therefore, a smoothing regularization term is added to the target energy functional of the point cloud correction. To constrain the changing trend of point cloud data.

[0138] In this embodiment S3.1, the point cloud correction target energy functional is defined, and the specific method is as follows:

[0139] ;

[0140] in, To correct the target energy functional of the point cloud; Q is the point cloud dataset to be corrected, equivalent to... ; The current coordinates of the point data to be corrected; These are standard points in an ideal weld model; Suppress error terms; For smoothing regularization; The deviation weighting coefficient between the point cloud data and the ideal weld model; These are the error field weighting coefficients; The weights for point cloud smoothing regularization; To optimize the number of iterations; K is the total number of error points in the point cloud dataset to be corrected;

[0141] in, The deviation weighting coefficient between the point cloud data and the ideal weld model is set to 1.5. This is the error field weighting coefficient, with a value of 0.8; This is the point cloud smoothing regularization weight coefficient, with a value of 0.5; To optimize the number of iterations, a value of 30 is used;

[0142] In this embodiment S3.2, the gradient descent algorithm is used to calculate the gradient descent direction of the point cloud data to be corrected. The specific method is as follows:

[0143] ;

[0144] This formula is for calculating gradient descent;

[0145] in, These are the partial derivatives used in the gradient descent algorithm.

[0146] The correction direction for each error point is determined by its gradient, and the update formula is as follows:

[0147] ;

[0148] in, This represents the number of iteration rounds. For the first Point cloud coordinates in each iteration; For the first Point cloud coordinates in each iteration; The learning rate is 0.01.

[0149] In this embodiment S3.3, the error point is corrected by combining the wave slope reflection disturbance potential function, and the offset correction amount of the error point is calculated. The specific method is as follows:

[0150] ;

[0151] in, For point data The offset correction amount; The influence coefficient of seawater ripple disturbance in the horizontal direction is 0.7; The wave disturbance impact coefficient over time is set to 0.9. The influence of the wave slope reflection disturbance potential function on the horizontal direction of the point cloud; The influence of the wave slope reflection disturbance potential function in the time dimension;

[0152] In this embodiment S3.4, the final correction formula is obtained based on the offset correction amount of the error point. The specific method is as follows:

[0153] ;

[0154] In this embodiment, steps S3.1-S3.4 are executed iteratively to optimize all point cloud data to be corrected, and finally the corrected weld point cloud data is obtained. The specific method is as follows:

[0155] ;

[0156] in, This is for the corrected weld point cloud data.

[0157] In this embodiment, the final corrected weld point cloud data includes the point cloud data corrected from the point cloud dataset to be corrected plus the original standard-compliant point cloud data.

[0158] S4. The robot performs automatic welding based on the corrected weld point cloud data.

[0159] Application Examples:

[0160] Implementation process

[0161] S1. Use sensors to collect marine environmental data, establish the wave slope reflection disturbance potential function based on the marine environmental data, and calculate and generate the laser scanning error distribution field;

[0162] The marine environmental data includes wave height, wave tilt angle, wave period, wave propagation speed, solar azimuth angle, solar intensity, and seawater refractive index.

[0163] The wave slope reflection disturbance potential function is constructed based on the nonlinear wave equation and takes into account the influence of the reflection of sunlight on the welding point by the wave slope. It is used to describe the dynamic interference caused by the reflection of sunlight by the wave slope on the scanning welding point.

[0164] In this embodiment, sensors are used to collect marine environmental data, as detailed below:

[0165] The system uses lidar to acquire weld seam point cloud data and monitors the surface morphology of ocean waves in real time, including wave tilt angle and wave height. An inertial measurement unit is used to detect wave propagation speed. An ultrasonic ranging sensor is used to measure wave height and period. Installed near the robot, the sensor uses ultrasonic echoes to measure water surface height and analyze wave undulation. An ambient light sensor monitors the azimuth angle and light intensity of sunlight. An underwater optical refractive index sensor measures the refractive index of seawater and monitors the optical refractive properties of seawater to calculate laser beam path offset.

[0166] In this embodiment S1, a nonlinear wave equation is used to establish the wave slope reflection disturbance potential function, and the laser scanning error distribution field is calculated. The specific method steps are as follows:

[0167] S1.1. Based on marine environmental data, define and construct the wave slope reflection disturbance potential function;

[0168] S1.2 Based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle;

[0169] The laser scanning error distribution field is used to quantitatively describe the laser scanning error caused by the reflection of sunlight from the inclined surface of seawater waves, and to identify and correct error point data.

[0170] In step S1.1, based on marine environmental data, a wave slope reflection disturbance potential function is defined and constructed as follows:

[0171] Measured marine environmental data: Wave propagation speed: 1.5 m / s; Wave height: 1.2 m; Wave period: 8 s; Sun azimuth: 35°; Sunlight intensity: 600 W / m²; Seawater refractive index: 1.33; Nonlinear damping coefficient: 0.1; Higher-order nonlinear coupling coefficient: 0.02;

[0172] Define the wave slope reflection perturbation potential function as follows: It satisfies the following nonlinear wave equation:

[0173] ;

[0174] Where t is time; x is the horizontal coordinate of the wave surface; y is the vertical coordinate of the wave surface; and z is the height coordinate of the wave surface. Let be the second-order partial derivative of the wave slope reflection perturbation potential function with respect to time; Here, C is the Laplace operator; C is the wave propagation speed. These are higher-order nonlinear coupling coefficients; The nonlinear damping coefficient; Let be the modulus of the wave potential energy gradient; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time;

[0175] In step S1.2, based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle, as follows:

[0176] ;

[0177] in, This represents the laser scanning error distribution field. Let be the partial derivative of the wave slope reflection disturbance potential function with respect to height; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time; This is the influence coefficient of the wave tilt angle; This represents the time-dynamic influence coefficient.

[0178] In this embodiment, in a marine environment, seawater waves form a dynamic tilted surface, causing optical deviations in the laser scanning path. Therefore, the perturbation potential function of the wave surface, i.e., the perturbation potential function of the wave tilted surface reflection, is set as follows: It satisfies a nonlinear wave equation, which describes the temporal evolution of the wave surface morphology and is used to calculate the perturbation of the laser path by the waves. The laser scanning error can be calculated from the wave surface tilt angle and the sunlight reflection angle, yielding the laser scanning error distribution field. ;

[0179] In marine environments, when welding robots use laser scanning to locate weld seams, the dynamic slopes formed by ocean waves cause sunlight reflection to interfere with the laser scanning, resulting in systematic deviations in the point cloud data. In standard hydrodynamics, ocean waves can be considered as a nonlinear wave system controlled by the Laplace equation and boundary conditions. However, due to factors such as turbulence, surface tension, and viscous damping, linear wave equations cannot accurately characterize these complex phenomena. Therefore, it is necessary to introduce nonlinear wave equations. Using nonlinear wave equations can accurately simulate ocean wave dynamics without relying on empirical filtering methods, thus exhibiting greater generalization ability. This can be achieved through the laser scanning error distribution field. By calculating the error distribution, the robot point cloud data can be directly adaptively corrected to improve the positioning accuracy of the weld. Because nonlinear terms are taken into account, the model can adapt to extreme conditions such as strong winds and waves and wave breakage.

[0180] In this embodiment, The wave tilt angle influence coefficient is calculated as follows: Where n is the refractive index of seawater, and is Optical path offset; The time-dynamic influence coefficient is calculated as follows: ,in λ is the laser wavelength, and c is the wave propagation speed.

[0181] S2. Based on the weld point cloud data generated by robot laser scanning, and combined with the laser scanning error distribution field to filter out abnormal points in the weld point cloud data, a point cloud dataset to be corrected is formed.

[0182] In this embodiment S2, the weld point cloud data generated by robot laser scanning is used as the basis for selecting outliers in the weld point cloud data based on the laser scanning error distribution field, thus forming a point cloud dataset to be corrected. The specific method steps are as follows:

[0183] S2.1 The weld point cloud data generated by the robot laser scanning is as follows:

[0184] ;

[0185] Where M is the total number of point data in the weld point cloud data; Index for point data; For point data Horizontal coordinates; Point data Coordinates perpendicular to the direction of propagation; For point data Elevation coordinates; For point data 3D coordinates; The weld point cloud data collected at time t;

[0186] In this embodiment, the weld point cloud data generated by robot laser scanning is a commonly used technique in the prior art. The specific method for obtaining weld point cloud data is as follows:

[0187] The robot program triggers the laser vision sensor to start positioning and scans the weld seam according to the laser vision scanning trajectory established by the offline programming software to obtain weld seam point cloud data.

[0188] S2.2 Use the laser scanning error distribution field to filter out the abnormal points in the weld point cloud data and form a point cloud dataset to be corrected.

[0189] In this embodiment S2.2, outliers in the weld point cloud data are filtered out using the laser scanning error distribution field to form a point cloud dataset to be corrected. The specific method steps are as follows:

[0190] S2.2.1 Calculate the theoretical error of each point in the weld point cloud data based on the laser scanning error distribution field:

[0191] ;

[0192] in, For point data Theoretical error; The effect of changes in wave height on the error; The impact of dynamic changes over time on the error; Point data for the oblique surface of the ocean wave The effect of tilt at the location; Point data of wave disturbances in the time dimension The impact of location;

[0193] S2.2.2 Setting the error threshold The screening error is greater than the error threshold. The point data constitutes the point cloud dataset to be corrected:

[0194] ;

[0195] in, This is a point cloud dataset that needs to be corrected.

[0196] S3. Define the target energy functional for point cloud correction, and combine it with the wave slope reflection perturbation potential function to iteratively optimize the point cloud dataset to be corrected, and obtain the corrected weld point cloud data.

[0197] The target energy functional for point cloud correction is constructed based on the wave slope reflection perturbation potential function combined with the point cloud dataset to be corrected, and is used to correct the error in weld point cloud data caused by the reflection of sunlight by seawater ripples.

[0198] In this embodiment S3, a target energy functional for point cloud correction is defined, and combined with the wave slope reflection perturbation potential function, the point cloud dataset to be corrected is iteratively optimized to obtain the corrected weld point cloud data. The specific method steps are as follows:

[0199] S3.1 Define the target energy functional for point cloud correction;

[0200] S3.2. Use the gradient descent algorithm to calculate the gradient descent direction of the point cloud data to be corrected;

[0201] S3.3. Combine the wave slope reflection disturbance potential function to correct the error point and calculate the offset correction amount of the error point;

[0202] S3.4. Based on the offset correction amount of the error points, the final correction formula is obtained;

[0203] S3.5 Iteratively execute S3.1-S3.4 to optimize all point cloud data to be corrected, and finally obtain the corrected weld point cloud data.

[0204] In this embodiment, the purpose of the point cloud correction target energy functional is to construct an optimization framework so that the error point cloud data scanned by the robot can be automatically corrected and approach the real weld morphology. Due to the wave characteristics of ocean waves, the weld point cloud data may have local offsets, distortions or outliers. The point cloud correction target energy functional quantifies the error and uses gradient descent optimization to make the error points converge toward the real weld trajectory, thereby eliminating laser scanning errors.

[0205] Traditional point cloud correction methods typically use simple interpolation or filtering to remove error points without considering the dynamic disturbance characteristics of ocean waves. This method constructs a point cloud correction target energy functional based on the wave slope reflection perturbation potential function, which can dynamically and adaptively adjust according to the ocean waves, making it more in line with the actual conditions of the marine welding environment.

[0206] In the point cloud correction process, it is not enough to simply make each point converge to its ideal position; the overall consistency of the entire point cloud data also needs to be considered. Therefore, a smoothing regularization term is added to the target energy functional of the point cloud correction. To constrain the changing trend of point cloud data.

[0207] In this embodiment S3.1, the point cloud correction target energy functional is defined, and the specific method is as follows:

[0208] ;

[0209] in, To correct the target energy functional of the point cloud; Q is the point cloud dataset to be corrected, equivalent to... ; The current coordinates of the point data to be corrected; These are standard points in an ideal weld model; Suppress error terms; For smoothing regularization; The deviation weighting coefficient between the point cloud data and the ideal weld model; These are the error field weighting coefficients; The weights for point cloud smoothing regularization; To optimize the number of iterations; K is the total number of error points in the point cloud dataset to be corrected;

[0210] In this embodiment S3.2, the gradient descent algorithm is used to calculate the gradient descent direction of the point cloud data to be corrected. The specific method is as follows:

[0211] ;

[0212] This formula is for calculating gradient descent;

[0213] in, These are the partial derivatives used in the gradient descent algorithm.

[0214] The correction direction for each error point is determined by its gradient, and the update formula is as follows:

[0215] ;

[0216] in, This represents the number of iteration rounds. For the first Point cloud coordinates in each iteration; For the first Point cloud coordinates in each iteration; This is the learning rate.

[0217] In this embodiment S3.3, the error point is corrected by combining the wave slope reflection disturbance potential function, and the offset correction amount of the error point is calculated. The specific method is as follows:

[0218] ;

[0219] in, For point data The offset correction amount; This represents the influence coefficient of seawater ripple disturbance in the horizontal direction; The wave disturbance impact coefficient over time; The influence of the wave slope reflection disturbance potential function on the horizontal direction of the point cloud; The influence of the wave slope reflection disturbance potential function in the time dimension;

[0220] In this embodiment S3.4, the final correction formula is obtained based on the offset correction amount of the error point. The specific method is as follows:

[0221] ;

[0222] In this embodiment, steps S3.1-S3.4 are executed iteratively to optimize all point cloud data to be corrected, and finally the corrected weld point cloud data is obtained. The specific method is as follows:

[0223] ;

[0224] in, This is for the corrected weld point cloud data.

[0225] In this embodiment, the final corrected weld point cloud data includes the point cloud data corrected from the point cloud dataset to be corrected plus the original standard-compliant point cloud data.

[0226] S4. The robot performs automatic welding based on the corrected weld point cloud data.

[0227] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A method for locating and adaptively correcting welded intersection lines of TKY pipe nodes, characterized in that, Includes the following steps: S1. Use sensors to collect marine environmental data, establish the wave slope reflection disturbance potential function based on the marine environmental data, and calculate and generate the laser scanning error distribution field; S2. Based on the weld point cloud data generated by robot laser scanning, and combined with the laser scanning error distribution field to filter out abnormal points in the weld point cloud data, a point cloud dataset to be corrected is formed. S3. Define the target energy functional for point cloud correction, and combine it with the wave slope reflection perturbation potential function to iteratively optimize the point cloud dataset to be corrected, and obtain the corrected weld point cloud data. S4. The robot performs automatic welding based on the corrected weld point cloud data.

2. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 1, characterized in that: The marine environmental data includes wave height, wave tilt angle, wave period, wave propagation speed, solar azimuth angle, solar intensity, and seawater refractive index. The wave slope reflection disturbance potential function is constructed based on the nonlinear wave equation and takes into account the influence of the reflection of sunlight from the seawater wave slope on the welding point. It is used to describe the dynamic interference caused by the reflection of sunlight from the seawater wave during the scanning of the welding point.

3. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 2, characterized in that: In step S1, a wave slope reflection disturbance potential function is established based on marine environmental data, and the laser scanning error distribution field is calculated. The specific steps are as follows: S1.

1. Based on marine environmental data, define and construct the wave slope reflection disturbance potential function; S1.2 Based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle; The laser scanning error distribution field is used to quantitatively describe the laser scanning error caused by the reflection of sunlight from the inclined surface of seawater waves, and to identify and correct error point data.

4. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 3, characterized in that: In step S1.1, based on marine environmental data, a wave slope reflection disturbance potential function is defined and constructed as follows: Define the wave slope reflection perturbation potential function as follows: ： ； Where t is time; x is the horizontal coordinate of the wave surface; y is the vertical coordinate of the wave surface; and z is the height coordinate of the wave surface. Let be the second-order partial derivative of the wave slope reflection perturbation potential function with respect to time; Here, C is the Laplace operator; C is the wave propagation speed. These are higher-order nonlinear coupling coefficients; The nonlinear damping coefficient; Let be the modulus of the wave potential energy gradient; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time; In step S1.2, based on the wave slope reflection perturbation potential function, the laser scanning error distribution field is obtained by calculating the seawater wave tilt angle and the sunlight reflection angle, as follows: ; in, This represents the laser scanning error distribution field. Let be the partial derivative of the wave slope reflection disturbance potential function with respect to height; Let be the partial derivative of the wave slope reflection perturbation potential function with respect to time; This is the influence coefficient of the wave tilt angle; This represents the time-dynamic influence coefficient.

5. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 4, characterized in that: In step S2, the weld point cloud data generated by the robot laser scanning is used to filter out outliers in the weld point cloud data by combining the laser scanning error distribution field, thus forming a point cloud dataset to be corrected. The specific method steps are as follows: S2.1 The weld point cloud data generated by the robot laser scanning is as follows: ; Where M is the total number of point data in the weld point cloud data; Index for point data; For point data Horizontal coordinates; Point data Coordinates perpendicular to the direction of propagation; For point data Elevation coordinates; For point data 3D coordinates; The weld point cloud data collected at time t; S2.2 Use the laser scanning error distribution field to filter out the abnormal points in the weld point cloud data and form a point cloud dataset to be corrected.

6. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 5, characterized in that: In step S2.2, outliers in the weld point cloud data are filtered out using the laser scanning error distribution field to form a point cloud dataset to be corrected. The specific steps are as follows: S2.2.1 Calculate the theoretical error of each point in the weld point cloud data based on the laser scanning error distribution field: ； in, For point data Theoretical error; The effect of changes in wave height on the error; The impact of dynamic changes over time on the error; Point data for the oblique surface of the ocean wave The effect of tilt at the location; Point data of wave disturbances in the time dimension The impact of location; S2.2.2 Setting the error threshold The screening error is greater than the error threshold. The point data constitutes the point cloud dataset to be corrected: ； in, This is a point cloud dataset that needs to be corrected.

7. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 6, characterized in that: The target energy functional for point cloud correction is constructed based on the wave slope reflection perturbation potential function combined with the point cloud dataset to be corrected, and is used to correct the error in weld point cloud data caused by the reflection of sunlight by seawater ripples.

8. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 7, characterized in that: In step S3, a target energy functional for point cloud correction is defined, and combined with the wave slope reflection perturbation potential function, the point cloud dataset to be corrected is iteratively optimized to obtain the corrected weld point cloud data. The specific steps are as follows: S3.1 Define the target energy functional for point cloud correction; S3.

2. Use the gradient descent algorithm to calculate the gradient descent direction of the point cloud data to be corrected; S3.

3. Combine the wave slope reflection disturbance potential function to correct the error point and calculate the offset correction amount of the error point; S3.

4. Based on the offset correction amount of the error points, the final correction formula is obtained; S3.5 Iteratively execute S3.1-S3.4 to optimize all point cloud data to be corrected, and finally obtain the corrected weld point cloud data.

9. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 8, characterized in that: In step S3.1, the point cloud correction target energy functional is defined, and the specific method is as follows: ； in, To correct the target energy functional of the point cloud; Q is the point cloud dataset to be corrected, equivalent to... ； The current coordinates of the point data to be corrected; These are standard points in an ideal weld model; Suppress error terms; For smoothing regularization; The deviation weighting coefficient between the point cloud data and the ideal weld model; These are the error field weighting coefficients; These are the weighting coefficients for point cloud smoothing regularization. To optimize the number of iterations; K is the total number of error points in the point cloud dataset to be corrected; In step S3.2, the gradient descent algorithm is used to calculate the gradient descent direction of the point cloud data to be corrected. The specific method is as follows: ； in, These are the partial derivatives used in the gradient descent algorithm. ； in, This represents the number of iteration rounds. For the first Point cloud coordinates in each iteration; For the first Point cloud coordinates in each iteration; This is the learning rate.

10. The method for TKY pipe node intersection line welding positioning and adaptive correction according to claim 9, characterized in that: In step S3.3, the error point is corrected by combining the wave slope reflection disturbance potential function, and the offset correction amount of the error point is calculated. The specific method is as follows: ； in, For point data The offset correction amount; This represents the influence coefficient of seawater ripple disturbance in the horizontal direction; The wave disturbance impact coefficient over time; The influence of the wave slope reflection disturbance potential function on the horizontal direction of the point cloud; The influence of the wave slope reflection disturbance potential function in the time dimension; In step S3.4, the final correction formula is obtained based on the offset correction amount of the error point. The specific method is as follows: ； In step S3.5, steps S3.1-S3.4 are executed iteratively to optimize all point cloud data to be corrected, ultimately obtaining the corrected weld point cloud data. The specific method is as follows: ； in, This is for the corrected weld point cloud data.

Images