Large-size complex curved surface composite multi-mode ultrasonic automatic detection method and system

Abstract

The application provides a large-size complex curved surface composite material multi-mode ultrasonic automatic detection system and method, relates to the field of ultrasonic detection, and comprises the following steps: (1) collecting original point cloud data of a workpiece surface; (2) scanning point cloud data to form a three-dimensional model; (3) fitting point cloud data; (4) interpolating the fitted point cloud; (5) calculating the normal of each point in the point cloud; (6) calculating the pose matrix of the workpiece; and (7) establishing the pose matrix of a manipulator. The scheme provides complete and reliable three-dimensional model support for subsequent trajectory planning of the workpiece, and ensures that the ultrasonic probe is attached to the normal of the workpiece surface. The scheme realizes the organic unity of large-span coverage and complex curved surface adaptation, avoids the trajectory planning limitation of a single motion mechanism, does not need to replace detection equipment to complete the process switching of three-dimensional information collection and ultrasonic detection, improves the overall detection efficiency, and guarantees the repeated positioning accuracy after the replacement of the module through the positioning structure of the quick-change interface.

Description

Technical Field

[0001] This invention relates to the field of ultrasonic testing technology. Background Technology

[0002] In the field of nondestructive testing of composite materials, ultrasonic testing is a commonly used method, including ultrasonic pulse-echo testing, penetration testing, and ultrasonic phased array testing. In existing technologies, automated testing of large-scale composite material specimens typically employs multi-axis motion systems or robotic arm collaboration schemes. The closest existing technologies include:

[0003] Multi-axis CNC systems: such as multi-coordinate measuring machines based on X, Y, and Z axes or customized gantry systems, use programming to control the ultrasonic probe to move along a preset path to achieve scanning of partial areas. These systems typically have a limited number of axes (e.g., 3-6 axes), making it difficult to cover the entire surface of ultra-large specimens or complex curved materials.

[0004] Robotic arm-assisted inspection system: This system uses an industrial robot (such as a 6-axis collaborative robot) to carry an ultrasonic probe for inspection, achieving flexible movement through offline programming or online trajectory planning. However, the working range and load capacity of a single robot are limited. For large specimens exceeding 40 meters in length, segmented inspection or multiple positioning operations are often required, resulting in low efficiency.

[0005] Multi-mode ultrasound integrated systems: such as the integrated system of ultrasound pulse reflection, penetration and phased array methods described in US patent US20180017675A1 or European patent EP2856171B1. However, the mode switching of this technology mostly depends on manual operation or independent equipment, lacks an automated switching mechanism, and increases detection time and human error.

[0006] Communication and Synchronization Control: Common methods include distributed control architectures based on traditional fieldbuses (such as PROFIBUS and CAN bus) or Ethernet, with master-slave structures used to coordinate multi-axis motion. However, these systems have long communication cycles (typically greater than 5ms), low synchronization accuracy, and are susceptible to network latency and jitter, leading to motion asynchrony and distorted detection data.

[0007] In terms of motion error compensation, existing technologies correct errors through calibration or sensor feedback, but these are mostly based on simple linear models and lack a full-process compensation mechanism for large-span motion, making it difficult to cope with changes in the surface contour and mechanical deformation of composite material specimens.

[0008] The main drawbacks or shortcomings of the aforementioned existing technologies can be summarized as follows:

[0009] Limited detection coverage: Due to insufficient number of axes or limited working space of the robotic arm, the existing system cannot complete the full-coverage automated detection of ultra-large composite material specimens (length > 40m, width > 8m, height > 2.5m) in one go. Multiple positioning or segmented operations are required, resulting in low detection efficiency, poor consistency, and easy risk of missed detection.

[0010] Low level of automation in multi-mode testing: Existing ultrasonic testing systems lack the ability to automatically switch between pulse reflection, penetration and phased array testing methods. Mode switching relies on manual intervention, which increases operational complexity, testing time and the probability of human error, making it difficult to meet the needs of efficient and multi-scenario testing.

[0011] Insufficient communication and synchronization performance: Traditional bus architectures suffer from long communication cycles and imperfect synchronization mechanisms, resulting in transmission delays and synchronization jitter. This leads to poor multi-axis motion coordination, affecting the real-time performance and accuracy of ultrasound data acquisition. Especially in systems with 18 or more axes, asynchrony between master and slave stations can cause deviations in motion trajectories.

[0012] Insufficient motion accuracy and error compensation: Existing error compensation methods are mostly based on local calibration, without considering geometric deformation and cumulative errors in large-span motion, and lack a full-process compensation scheme from acquisition and modeling to motion decomposition, resulting in low positioning accuracy and failing to meet the high-precision testing requirements of composite materials.

[0013] Low system integration: Existing technologies often treat trajectory planning, error compensation, and mode switching as independent modules, lacking unified optimization, resulting in system complexity, maintenance difficulties, and difficulty in adapting to the rapid testing needs of large-scale specimens. Summary of the Invention

[0014] To overcome the problems of limited detection range and low system integration of existing ultrasonic testing equipment, this invention provides a multi-mode ultrasonic automatic testing method and system for large-size complex curved surface composite materials.

[0015] The technical solution adopted by the present invention to achieve the above objectives is: a multi-mode ultrasonic automatic testing system and method for large-size complex curved surface composite materials, comprising the following steps:

[0016] Step 1: Collect raw point cloud data of the workpiece surface;

[0017] Step 2: After scanning the point cloud data, a complete 3D model is formed;

[0018] Step 3: Fit the point cloud data;

[0019] Step 4: Perform interpolation on the fitted data points;

[0020] Step 5: Calculate the normal vector of each data point after interpolation;

[0021] Step 6: Calculate the workpiece pose matrix;

[0022] Step 7: Calculate the pose matrix of the robot arm based on the pose matrix of the workpiece.

[0023] Specifically, in step one, a laser sensor is used to emit a laser beam to collect the original point cloud data of the workpiece surface.

[0024] Specifically, in step three, a B-spline-based surface fitting technique is employed. This involves performing the Kronecker product on the B-spline basis functions in the x and y directions to construct a two-dimensional basis function system. This mathematical processing method effectively establishes a globally smooth mapping relationship between discrete data points (x, y, z). During the function solution process, the least squares optimization method is used to calculate the optimal coefficients, thereby generating a surface model with continuously differentiable properties.

[0025] Specifically, in step four, an arc length interpolation algorithm is first performed based on a uniformly accelerated and decelerated straight path. Then, an arc length interpolation algorithm based on a uniformly accelerated and decelerated model is used for the straight path. Next, the total displacement is initialized and motion constraint parameters are set. Finally, the displacement sequence is periodically generated using the discretized motion equations to obtain the coordinates of the interpolation points.

[0026] Specifically, in step four, the motion constraint parameters include: maximum speed v max , acceleration a and interpolation period T.

[0027] Specifically, in step five, the geometric difference method in geometric modeling is used to solve for the normal vector of each point. That is, the normal vector of the target point is obtained by connecting the target point in the point cloud and its four adjacent points and performing cross product operation. At the same time, in order to enhance the noise resistance, the algorithm introduces a double diagonal cross product strategy, which calculates the normal vector estimate by combining multiple triangles. Each group of triangles generates a normal vector. Finally, the average value of the four normal vectors is taken. For the obtained normal vector, its x and y directions are then adaptively weighted.

[0028] Specifically, in step six, pose is constructed using point cloud data and the normal of each point. The pose is constructed based on the Tait-Bryan angle, aligning the normal vector N = (nx, ny, nz) with the X-axis. The rotation matrix (Rot_y) is calculated using the angle between the normal vector and the X-axis, and then added to the normal direction of each point to form the pose of that point.

[0029] Specifically, in step seven, after obtaining the interpolation point coordinates and corresponding normals, the large-span coordinate changes in the x, y, and z directions are first converted into slider coordinate changes. For the fine parts of the surface, a coordinate system transformation is performed to transform the coordinates of the workpiece points to the coordinates of the robot's end effector. This process involves transforming the workpiece coordinate matrix to the world coordinate system, then to the robot's base coordinate system, and then transforming it to the robot's end effector coordinates based on the six-axis continuous transformation of the robot. This process then reverse-engineers the pose matrix of the robot's end effector coordinates and finally calculates the six-axis rotation angles of the robot.

[0030] Specifically, in step seven, the order of solving the six axes of the robot is to first calculate axes 1, 3, and 2, and then calculate axes 5, 4, and 6. By utilizing the definition of polar angles in polar coordinates and the precise judgment of the limits of the atan2 function, the angle of rotation of the base joint around each axis can be directly solved.

[0031] The present invention provides a multi-mode ultrasonic automatic testing system for large-size complex curved surface composite materials, comprising:

[0032] The data acquisition module collects raw point cloud data of the workpiece surface;

[0033] The data modeling module connects to the sensor acquisition module and is used to scan point cloud data into a 3D model.

[0034] The surface fitting module, connected to the data modeling module, is used to fit and interpolate point cloud data.

[0035] The workpiece pose calculation module, connected to the surface fitting module, is used to calculate the normals of each point and establish the workpiece pose matrix.

[0036] The inverse calculation module is connected to the workpiece pose calculation module and is used to inversely solve the workpiece pose matrix to obtain the pose matrix of the robot arm.

[0037] The power supply modules provide power to the data acquisition module, data modeling module, surface fitting module, workpiece pose calculation module, and inverse calculation module.

[0038] A computer program product includes a computer program for performing the steps of the above-described multi-mode ultrasonic automatic detection method for large-size complex curved surface composite materials.

[0039] A computer-readable storage medium comprising computer-readable instructions for executing a program for a multi-mode ultrasonic automatic testing method for large-size complex curved surface composite materials.

[0040] The beneficial effects of this invention are as follows:

[0041] 1. A three-dimensional information acquisition scheme combining laser scanning with fitting interpolation and normal calculation is proposed. Addressing the issues of missing data and uneven density in point cloud data of complex curved surfaces, the scheme uses a fitting interpolation algorithm to fill data gaps while accurately calculating the normal information of each detection point. This scheme not only provides a complete and reliable three-dimensional model of the workpiece for subsequent trajectory planning but also ensures normal alignment between the ultrasonic probe and the workpiece surface, improving the accuracy of the detection data from the source.

[0042] 2. A trajectory planning system integrating slider trajectory and robot inverse kinematics is constructed. First, the linear motion trajectory of the guide rail slider is generated based on the workpiece's three-dimensional coordinates. Then, the position and posture requirements of the detection points are converted into robot joint motion parameters through inverse kinematics algorithm. After collaborative verification to eliminate motion interference, a composite detection trajectory is formed. This design fully utilizes the slider's large-span motion advantage and the robot's flexible posture adjustment capability to achieve an organic unity of large-span coverage and complex surface adaptation, avoiding the limitations of trajectory planning based on a single motion mechanism.

[0043] 3. A dual-detection module integration solution with a quick-change interface design allows the laser scanning module and ultrasonic testing module to be alternately mounted on the end effector of the robotic arm via a standardized quick-change interface with positioning pins. This eliminates the need to change the testing equipment, enabling seamless switching between 3D information acquisition and ultrasonic testing. This significantly reduces preparation time and improves overall testing efficiency. Furthermore, the quick-change interface's positioning structure ensures accurate repeatability after module replacement, avoiding errors caused by equipment disassembly and reassembly. Attached Figure Description

[0044] Figure 1 This is a flowchart of the detection process of this invention.

[0045] Figure 2 This is a hierarchical structural diagram of the three systems of this invention.

[0046] Figure 3 This is a system structure block diagram of the present invention. Detailed Implementation

[0047] The present invention will be further explained and described below with reference to the accompanying drawings and embodiments.

[0048] Example: Figure 1 and Figure 2 As shown, the ultrasonic testing system of the present invention consists of a three-layer architecture, which is progressive and seamlessly connected, ultimately achieving precise coordination between the movement of the robotic arm and data acquisition in ultrasonic testing, and fully adapting to the testing needs of complex curved surface workpieces.

[0049] First layer: Clock synchronization layer (establishes a unified timing reference)

[0050] The main purpose is to eliminate the clock difference between the master and slave stations, build a unified clock at the nanosecond level, and provide a precise time scale for subsequent instruction execution and data feedback.

[0051] Step 1: The master station and all slave stations complete hardware startup and self-test, and establish a stable communication link through the general fieldbus;

[0052] Step 2: The master station activates its own global clock source to generate a stable reference clock signal, which serves as the time reference for the entire system.

[0053] Step 3: The master station periodically broadcasts a synchronization clock signal to all slave stations via the bus. The signal contains the master station's reference clock stamp and transmission delay compensation information.

[0054] Step 4: After receiving the synchronization clock signal, each slave station compares the deviation between its own hardware clock and the master station's reference clock, and corrects the clock drift through a local compensation algorithm to achieve nanosecond-level alignment with the master station's clock.

[0055] Step 5: The slave station feeds back the clock alignment result to the master station. The master station checks the synchronization accuracy of all slave stations. If all meet the standard, it proceeds to the next layer. If there are any unaligned slave stations, the master station triggers secondary synchronization until all slave stations meet the timing requirements.

[0056] Second layer: Instruction synchronization layer

[0057] The main function of this layer is to accurately send out and execute the master station's instructions synchronously based on a unified clock, ensuring that the timing of the robot's movement and the ultrasonic detection action are matched, and achieving precise linkage from movement to acquisition.

[0058] Step 1: Based on the workpiece scanning path planning results (such as the curved surface trajectory of an aero-engine blade), the main station generates two types of core instructions: robot motion instructions (joint angles, movement speed, etc.) and ultrasonic data acquisition trigger instructions (transmission / reception timing, etc.). All instructions are bound to a unified timestamp.

[0059] Step 2: The master station synchronously sends timestamped instructions to the corresponding slave stations (motion instructions to the robot drive, acquisition instructions to the ultrasonic acquisition card) via the industrial bus, ensuring that the instruction transmission delay is controllable;

[0060] Step 3: After receiving the instruction from the slave station, verify the timestamp based on the aligned synchronization clock to determine the precise time of instruction execution;

[0061] Step 4: When the synchronization clock reaches the instruction timestamp, each slave station executes the instruction synchronously, the robot moves along the preset trajectory, and the ultrasonic probe synchronously transmits or receives ultrasonic waves.

[0062] Third layer: Feedback synchronization layer

[0063] The main function of this layer is to provide real-time feedback on execution data. The main station corrects deviations through closed-loop control to ensure that the scanning trajectory is consistent with the preset path.

[0064] Step 1: Each slave station collects execution data in real time, namely, the robot arm actuator collects joint position and probe posture data; the ultrasonic acquisition card collects echo signals and acquisition timing data.

[0065] Step 2: The slave station attaches a synchronization clock tag to the collected data and transmits it back to the master station in real time via the bus to ensure that the data corresponds one-to-one with the command execution timing.

[0066] Step 3: The main station compares the feedback data with preset parameters (such as the planned path and standard collection timing) to analyze whether there is a trajectory deviation or timing misalignment.

[0067] Step 4: If a deviation is detected (such as the robot arm position deviating from the preset value), the master station generates an adjustment command based on the synchronous clock and sends it to the corresponding slave station to correct the motion parameters or acquisition timing.

[0068] Step 5: The adjusted execution data is sent back to the main station through the feedback link to form a closed loop, continuously ensuring synchronization accuracy until all scanning tasks are completed.

[0069] In this embodiment, the device of the present invention includes two parallel slide rails, with a sliding column slidably connected to the slide rails. The sliding column can move back and forth on the slide rails. A slider is slidably connected to the sliding column, and the slider can move in both left and right and up and down directions. A six-axis manipulator is connected to the inner side of the slider via a base joint, and an ultrasonic probe is fixed at the front end of the six-axis manipulator. The composite material workpiece is located between the two slide rails.

[0070] Inverse kinematics (IK) solving for a robotic arm is a core step in achieving precise control of the end effector. Essentially, it involves deriving the angular parameters of each joint based on the desired pose of the end effector. For multi-DOF robotic arms, the "wrist-arm separation method" has emerged to simplify the solution process. This method, based on the structural characteristics of the robotic arm, decomposes the overall kinematic chain into two relatively independent subsystems: the "arm" and the "wrist." It solves for the position and attitude-related joint variables separately, making it suitable for robotic arms with ball-arm structures. This method has become a classic approach in industrial robot IK solving that combines theoretical simplicity with engineering practicality.

[0071] The core logic of the "wrist-arm separation method" lies in: first, isolating the wrist's influence on position, and then using the kinematic relationship of the arm joints (from the base to the wrist center point) to solve for the joint variables corresponding to the end effector position, transforming the three-dimensional position problem into a sub-problem that can be simplified using geometric methods (such as the law of cosines, spherical trigonometry, etc.); then, based on the attitude coupling relationship of the wrist joints (from the wrist center point to the end effector), solving for the joint variables corresponding to the attitude through rotation matrix or Euler angle decomposition, thus achieving decoupled calculation of position and attitude. The specific implementation process of the wrist-arm separation method will be described in detail below.

[0072] First, by establishing the wrist coordinate system {W}, the end pose matrix T is decomposed as shown in equation (1).

[0073] (1) ;

[0074] The first axis is the rotational joint of the robot's base. It only controls the overall orientation of the robot in the X and Y planes, while the vertical height Z coordinate of the wrist point and the degree of arm extension are independent of the first axis.

[0075] Then, based on this characteristic, the first axis angle θ1 can be directly obtained from the wrist point position, as shown in the two sets of solutions generated by equation (2), which correspond to the left and right configurations.

[0076] (2) ;

[0077] In the formula, y W ,x W This indicates the coordinates of the wrist point in the X and Y planes;

[0078] arm joint angle θ 2， The geometric constraint equations for θ3 are established using the parameters of the manipulator link as shown in equation (3).

[0079] (3) ;

[0080] In the formula, y w ,x w The z-coordinate of the wrist point position. w The vertical height coordinate of the wrist point is represented by l, which represents the length of the link.

[0081] Equation (4) can be derived using the Law of Cosines and algebraic operations.

[0082] (4) ;

[0083] In the formula, y w ,x w The z-coordinate of the wrist point position. w The vertical height coordinate of the wrist point is represented by l, and the length of the link is represented by l.

[0084] By using the "wrist-arm separation method" and structural decomposition, the complexity of solving the inverse kinematics of a multi-degree-of-freedom manipulator is effectively solved.

[0085] This invention has been described through embodiments. Those skilled in the art will understand that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of this invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, this invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of this invention.

Claims

1. A multi-mode automatic ultrasonic testing method for large-size complex curved surface composite materials, characterized in that: Includes the following steps: Step 1: Collect raw point cloud data of the workpiece surface; Step 2: After scanning the point cloud data, a complete 3D model is formed; Step 3: Fit the point cloud data; Step 4: Interpolate the fitted data points; Step 5: Calculate the normal vector of each data point after interpolation; Step 6: Calculate the workpiece pose matrix; Step 7: Calculate the pose matrix of the robot arm based on the pose matrix of the workpiece.

2. The automatic ultrasonic testing method according to claim 1, characterized in that: In step three, a surface fitting technique based on B-splines is used to construct a two-dimensional basis function system by performing Kronecker product operations on the B-spline basis functions in the x and y directions. During the function solution process, the least squares optimization method is used to calculate the optimal coefficients, generating a surface model with continuously differentiable properties.

3. The automatic ultrasonic testing method according to claim 1, characterized in that: In step four, a straight path arc length interpolation algorithm based on a uniform acceleration and deceleration model is first used. Then, the total displacement is initialized and motion constraint parameters are set. Finally, the displacement sequence is periodically generated using the discretized motion equations to obtain the interpolation point coordinates.

4. The automatic ultrasonic testing method according to claim 1, characterized in that: In step four, the motion constraint parameters include: maximum speed v max , acceleration a and interpolation period T.

5. The automatic ultrasonic testing method according to claim 1, characterized in that: In step five, the geometric difference method in geometric modeling is used to solve for the normals of each point; The double diagonal cross product method is used to calculate the estimated value of the normal vector by combining multiple triangles. Specifically, each triangle generates a normal vector, and the average value of the four normal vectors is taken. For the obtained normal vector, an adaptive weighting is then applied to its x and y directions.

6. The automatic ultrasonic testing method according to claim 1, characterized in that: In step seven, after obtaining the interpolation point coordinates and corresponding normals, the large-span coordinate changes in the x, y, and z directions are first converted into the coordinate changes of the slider; then, the coordinate system is transformed on the fine part of the surface to transform the coordinates of the workpiece points to the coordinates of the robot end effector; then, the pose matrix of the robot end effector coordinates is solved in reverse, and finally the six-axis rotation angle of the robot is obtained.

7. The automatic ultrasonic testing method according to claim 6, characterized in that: In step seven, the order of solving the six axes of the robot is to first calculate axes 1, 3, and 2, and then calculate axes 5, 4, and 6. By using the definition of polar angles in polar coordinates and the precise judgment of the limits of the atan2 function, the angle of rotation of the base joint around each axis is directly solved.

8. A multi-mode automatic ultrasonic testing system for large-size complex curved surface composite materials, characterized in that: include: The data acquisition module collects raw point cloud data of the workpiece surface; The data modeling module connects to the sensor acquisition module and is used to scan point cloud data into a 3D model. The surface fitting module connects to the data modeling module and is used to fit point cloud data and interpolate based on the fitted curve. The workpiece pose calculation module, connected to the surface fitting module, is used to calculate the normals of each point and establish the workpiece pose matrix. The inverse calculation module is connected to the workpiece pose calculation module and is used to inversely solve the workpiece pose matrix to obtain the pose matrix of the robot arm. The power supply modules provide power to each of the above modules.

9. A computer program product, comprising a computer program, characterized in that: The computer program is used to perform the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, A program storing computer-readable instructions for performing the method of any one of claims 1-7.

Images