An automatic detection and defect projection calibration method for complex curved surface components based on ultrasonic-laser cooperation

Abstract

The application discloses a kind of complex curved surface component automatic detection and defect projection calibration method based on ultrasonic-laser cooperation, it is related to the cross technical field of intelligent manufacturing and nondestructive testing;The method comprises the following steps: establishing system coordinate system, integrating ultrasonic detection module and laser scanning projection module, and the scanning unit and the projection unit of laser scanning projection module share pre-calibrated module coordinate system;Obtain workpiece surface data by laser scanning projection module and reconstruct curved surface model, and accordingly plan detection path;Switch laser scanning projection module to ultrasonic detection module, and after switching, the cooperative relationship does not need to be recalibrated;Along the planned path, ultrasonic detection is performed, real-time attitude data is collected to compensate the detection signal;Identify defects and calculate defect three-dimensional information;Based on the conversion relationship between module coordinate system and system coordinate system, the defect three-dimensional information is directly mapped to projection pixel coordinates, and the projection unit is driven to project the defect information in situ to the workpiece surface.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and nondestructive testing, and more specifically to an automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy. Background Technology

[0002] In the interdisciplinary field of high-end intelligent manufacturing and non-destructive testing, large-size complex curved surface components made of materials such as carbon fiber reinforced polymer (CFRP) and titanium alloys are increasingly widely used. The accurate detection and in-situ visualization of their internal defects (such as delamination, cracks, and porosity) directly determine the operational safety and service life of the product. Ultrasonic testing of large-size complex curved surface components is mainly done manually, dividing the large component into multiple areas according to structural differences and adjacent parts, and manually inspecting and calibrating defects. Many industries and fields have gradually tried to use automation to improve the low efficiency, high workload, and large error in defect registration feedback of manual operations. Although some automated ultrasonic testing technologies can meet the needs of defect detection and visualization, the following core industry pain points remain: difficulty in achieving efficient collaborative operation and accurate projection, and inability to meet the stringent requirements of high-end manufacturing.

[0003] Limited operating range: The working radius of traditional six-axis robotic arms is usually no more than 3m. For large components larger than 5m×5m, multiple moves of equipment or the construction of auxiliary tracks are required, which can easily lead to the accumulation of positioning errors. This not only reduces the accuracy of ultrasonic testing, but also disrupts the coordinate consistency of ultrasonic testing and positioning in coordinated operations.

[0004] Low module switching efficiency: Independent tool modules, robotic arm scanning modules and defect location modules have inherent relative positional deviations. Each time the ultrasonic testing module is switched with other modules, the coordinate system and collaborative relationship between the two need to be recalibrated, which greatly prolongs the testing cycle and violates the original intention of efficient collaborative operation.

[0005] Outdated or missing defect location systems: Automated ultrasonic scanning systems often struggle to integrate location and marking systems. The usual practice is to manually determine the location and measure the size of defects on the actual workpiece based on the scanned C-display image. This fails to achieve accurate location and rapid measurement of defects detected by ultrasonic testing. Furthermore, the cumbersome calibration process disrupts the synergy and consistency of automated ultrasonic testing, as well as the integrated operational logic.

[0006] The ultrasonic transducer contact surface lacks the ability to adapt to the curved surface of the workpiece and compensate for its posture: Large-sized components generally undergo shape deformation during the manufacturing process. Traditional path planning schemes that rely on theoretical CAD models are difficult to adapt to the actual curved surface shape of the components, resulting in contact surface misalignment and distortion of the ultrasonic transducer. At the same time, the slight vibration during the robotic arm scanning operation can cause the ultrasonic transducer to shift its posture, reducing the quantitative analysis and visualization of defects.

[0007] Therefore, how to provide an ultrasonic-laser collaborative automated detection method that can achieve wide coverage, efficient collaborative switching, accurate detection and in-situ projection calibration of defects, so as to overcome the problems of limited operating range, repeated calibration required for module switching, lagging defect localization and lack of surface adaptation capability of traditional solutions, is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0008] In view of this, the present invention provides an automated detection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy, aiming to solve one of the above-mentioned technical problems.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: An automated detection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy includes the following steps: S1. Establish a system coordinate system, under which an ultrasonic detection module and a laser scanning projection module are integrated, and the scanning unit and the projection unit of the laser scanning projection module share the same module coordinate system; the module coordinate system has a fixed relative pose transformation relationship between the scanning unit and the projection unit; S2. Obtain workpiece surface data and reconstruct the curved surface model through the laser scanning projection module; S3. Planning detection paths based on reconstructed surface models; S4. Switch the laser scanning projection module to the ultrasonic detection module. After switching, there is no need to recalibrate the coordination relationship. S5. Perform ultrasonic testing along the detection path planned in step S3, and collect real-time attitude data to compensate for the detection signal; S6. Identify defects based on detection signals and calculate the three-dimensional information of the defects; S7. Based on the transformation relationship between the module coordinate system and the system coordinate system, the three-dimensional information of the defect is directly mapped to the projection pixel coordinates, and the projection unit is driven to project the defect information onto the workpiece surface in situ.

[0010] Through the above technical solution, this invention discloses an automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser collaboration. With the integrated laser scanning and laser projection as its core, it constructs a closed-loop, calibration-free operation from surface reconstruction, path planning, ultrasonic detection, to defect projection. This method effectively eliminates the reliance on repetitive multi-coordinate system calibration in traditional solutions by establishing a unified system coordinate system and solidifying the relative pose transformation relationship between the laser scanning unit and the projection unit, effectively avoiding the accumulation of repetitive positioning errors of the robotic arm and hand-eye calibration errors. Based on this, it obtains the true surface morphology of the workpiece through laser scanning and reconstructs the curved surface model, solving the theoretical C... The method addresses the adaptation between the AD model and the actual component deformation; it combines model predictive control algorithms to plan the scanning path, ensuring that the ultrasonic probe always fits along the surface normal, thus guaranteeing the perpendicular incidence of the ultrasonic signal from a motion planning perspective; after switching to the ultrasonic detection module, there is no need to recalibrate the collaborative relationship, and the inertial measurement unit collects attitude data in real time for beam delay compensation, effectively suppressing the beam offset caused by the slight vibration of the robotic arm; finally, based on the pre-calibrated coordinate system transformation relationship, the three-dimensional information of the defect is directly mapped to the projection pixel coordinates, driving the projection unit to project the defect contour and depth values ​​onto the workpiece surface in situ; this method achieves full-process automation of wide coverage, efficient collaborative switching, accurate detection, and defect projection calibration.

[0011] Preferably, in the above-mentioned automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser collaboration, the specific method for establishing the system coordinate system includes: setting up a three-axis gantry, a six-axis industrial robot installed at the end of the three-axis gantry, and an ultrasonic-laser collaborative inspection and projection component. The ultrasonic-laser collaborative inspection and projection component includes a tooling module fixed to the three-axis gantry and moving synchronously with it. The tooling module is provided with two tool stations, which respectively carry the ultrasonic inspection module and the laser scanning projection module. Through the composite architecture of the three-axis gantry and the six-axis robotic arm, full-area coverage of large-size components (0-8m×0-8m) is achieved, and positioning accuracy of ±0.01mm is ensured through the grating ruler and quick-change interface. The collaborative benchmark of ultrasonic inspection and laser scanning and projection is established simultaneously, effectively solving the industry pain point that the operating range and collaborative accuracy of traditional equipment cannot be simultaneously achieved. The tooling module is fixed to the gantry and moves synchronously with it, providing physical guarantee for coordinate consistency after tool switching, and is the underlying hardware support for achieving calibration-free switching.

[0012] Preferably, in the above-mentioned automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser collaboration, the effective stroke of the X-axis crossbeam of the three-axis gantry is 0-8m, the effective stroke of the Z-axis lifting mechanism of the three-axis gantry is 0-2m, the end flange of the six-axis industrial robot integrates an automatic tool switching master disk with a tool disk positioning accuracy of ±0.01mm, a tool switching time ≤10 seconds, and a working radius of 1.5m for the six-axis industrial robot. By limiting the effective stroke of the three-axis gantry to 0-8m and the working radius of the robotic arm to 1.5m, single-position full-coverage inspection of large-sized components (5m×5m and above) can be achieved without multiple moves of equipment or the construction of auxiliary tracks, effectively avoiding the accumulation of positioning errors. The closed-loop control of the grating ruler ensures a positioning accuracy of ±0.02mm, and the quick-change interface ensures a tool switching accuracy of ±0.01mm.

[0013] Preferably, in the above-mentioned automated inspection and defect projection calibration method for complex curved surface components based on ultrasound-laser synergy, the ultrasonic inspection module integrates a 128-element phased array ultrasonic probe and an inertial measurement unit (IMU). The IMU has a sampling frequency of 1 kHz and an attitude accuracy of ±0.05°. The ultrasonic inspection module is also equipped with an automatic coupling agent supply device at its front end. The 128-element phased array probe achieves electronic focusing and deflection of the acoustic beam, adapting to component surfaces with different curvatures without mechanical oscillation. The high-precision IMU senses minute changes in probe attitude in real time, providing a high-frequency, high-precision data source for attitude compensation; the automatic coupling agent supply device ensures continuous and stable ultrasonic signal transmission, guaranteeing the reliability of long-term continuous scanning.

[0014] Preferably, in the above-mentioned automated detection and defect projection calibration method for complex curved surface components based on ultrasound-laser collaboration, the inertial measurement unit acquires the real-time attitude quaternions of the ultrasonic probe at a frequency of 1 kHz. The main control processing unit combines it with the expected pose quaternion of the planned path. Compare and calculate the deviation quaternion. in, This is quaternion multiplication. for The conjugate quaternion; according to The roll, pitch, and yaw angle deviations are analyzed, and the delay correction of each element of the phased array probe is calculated. The beamforming algorithm, superimposed on the beamforming algorithm, compensates for beam offset caused by attitude deviations in real time, ensuring that the sound beam is perpendicularly incident on the workpiece surface. Real-time attitude deviation calculation based on quaternions avoids the gimbal locking problem of Euler angles, ensuring no singularities across the entire attitude range. By calculating the delay correction of each array element and superimposing it on the beamforming algorithm, the beam offset caused by low-frequency shaking and high-frequency flutter during long-stroke scanning of the robotic arm is compensated in real time, ensuring that the sound beam is always perpendicularly incident. After compensation, the signal-to-noise ratio of the C-scan image is improved by approximately 30%, and weak signal features such as tiny pores are no longer drowned out by background noise, providing high-quality input for accurate segmentation of deep learning networks.

[0015] Preferably, in the above-mentioned automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser collaboration, the linear laser scanning accuracy of the laser scanning projection module is ±0.03mm, the laser projection unit uses an RGB laser light source, the projection resolution is 1920×1080, the projection distance is 0.5-3m, and the projection error is ≤0.3mm. The linear laser scanning accuracy of ±0.03mm ensures a high degree of consistency between the reconstructed curved surface model and the actual surface of the workpiece; the RGB laser light source brightness is ≥5000 lumens, ensuring clear projection even under complex workshop ambient light interference. The projection resolution of 1920×1080, the projection distance of 0.5-3m, and the projection error ≤0.3mm achieve large-scale, high-definition, low-distortion in-situ visualization of defect information, meeting the minimum visualization requirements for aerospace-grade defect tolerance.

[0016] Preferably, in the above-mentioned automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy, in step S2, a parameterized mathematical model of the workpiece surface is constructed using the NURBS algorithm, and the expression of the NURBS model is:

[0017] in, for 3D control point array, These are the weighting factors for the corresponding control points; and They are respectively Towards Subsequent Towards The NURBS algorithm uses B-spline basis functions of order B. It accurately describes arbitrarily complex freeform surfaces, solving the core problem of mismatch between theoretical CAD models and actual component deformation. Setting p=3 and q=3 ensures second-order smoothness and continuity of the surface, guaranteeing smooth and abrupt changes in probe posture during subsequent path planning. The reconstructed surface model provides a unified geometric description benchmark for MPC path planning, defect coordinate mapping, and projection distortion correction, serving as a data bridge connecting the three technical links of laser scanning, ultrasonic testing, and defect projection.

[0018] Preferably, in the above-mentioned automated detection and defect projection calibration method for complex curved surface components based on ultrasound-laser synergy, in step S3, a model predictive control algorithm is used to plan the scanning path and construct a cost function:

[0019] in, For the predicted system state vector, Let be the desired trajectory state vector. and This is the weight matrix. To predict the time domain, the scanning trajectory of the ultrasonic probe is planned under the premise of satisfying the dynamic constraints of the robotic arm and the velocity constraints of the gantry, ensuring that the probe always fits along the normal direction of the workpiece surface with a fit gap ≤0.1mm. Model predictive control has the advantage of foresight, proactively planning the optimal scanning trajectory under the premise of satisfying the dynamic constraints of the robotic arm and the velocity constraints of the gantry; ensuring that the probe always fits along the normal direction of the workpiece surface with a fit gap ≤0.1mm, guaranteeing the perpendicular incidence of the ultrasonic signal from the motion planning level; the cost function balances tracking accuracy and motion smoothness, avoiding inertial shocks and attitude changes caused by sudden stops and turns, and extending the service life of the equipment.

[0020] Preferably, in the above-mentioned automated detection and defect projection calibration method for complex curved surface components based on ultrasound-laser collaboration, in step S6, a pre-trained Vision Transformer deep learning network is used to perform pixel-level semantic segmentation on the ultrasonic C-scan image, identify defect regions, and mark defect contours; based on the material anisotropic sound velocity model, the Christoffel equation is solved. The sound wave is obtained along the direction of propagation. group velocity Christoffel tensor For material density, identity matrix The direction vector of acoustic wave polarization; based on the defect echo flight time Calculate the depth of the defect The depth measurement error is ≤0.1mm. The main control processing unit fuses the 3D coordinates, contour, and depth data of the defect with the surface model data obtained in step S2 to generate the high-precision data required for laser projection. Vision Transformer is based on a multi-head self-attention mechanism and has a global receptive field. It is good at recognizing composite material defects with varied shapes and blurred boundaries. The segmentation accuracy and classification accuracy both reach industrial-grade standards. Based on the material anisotropic sound velocity model, Christoffel's equation is solved to accurately calculate the group velocity of sound waves in different propagation directions. This avoids the depth calculation error caused by the traditional isotropic assumption. The depth measurement error is ≤0.1mm, realizing a complete representation of the 3D information (planar position + depth) of the defect.

[0021] Preferably, in the above-mentioned automated detection and defect projection calibration method for complex curved surface components based on ultrasound-laser collaboration, in step S7, the correspondence between the three-dimensional coordinates of the defect and the projected pixel coordinates is constructed using a pinhole camera model, satisfying the formula... ;in, scale factor Projected pixel coordinates Three-dimensional coordinates of the defect Projector intrinsic parameter matrix The extrinsic parameter matrix is ​​used. Based on the pre-calibrated intrinsic and extrinsic parameter matrices, the three-dimensional coordinates of defects can be directly mapped to projected pixel coordinates without on-site calibration, eliminating the need for repeated multi-coordinate system calibration processes involving the projector, workpiece, ultrasonic testing module, and robotic arm in traditional solutions. This reduces computational load and provides high real-time performance, meeting the stringent requirements for projection delay in online inspection scenarios. This calibration-free mapping is the core invention of this patent, representing the final implementation at the algorithm level. It reduces the projection error from ±2.5mm in traditional solutions to ≤0.3mm, effectively avoiding calibration offsets caused by environmental interference.

[0022] As can be seen from the above technical solution, compared with the prior art, the present invention discloses an automated detection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy, which has the following beneficial effects: 1. This invention integrates the laser scanning unit and the laser projection unit into the same module and shares a pre-calibrated coordinate system. With the tooling module that moves synchronously with the gantry as a unified installation reference, it effectively eliminates the dependence on repeated calibration of multiple coordinate systems between the ultrasonic testing module, laser scanning module, and projection module in traditional solutions from a hardware physical perspective, realizing a calibration-free operation throughout the entire process. On this basis, the composite architecture of a three-axis gantry and a six-axis robotic arm achieves full coverage of large areas from 0 to 8 meters, effectively shortening the tool switching and reference recovery time, as well as the total time for detection and marking, and significantly improving the overall detection efficiency.

[0023] 2. This invention uses the NURBS algorithm to reconstruct the real surface model of the workpiece to solve the adaptation problem between the theoretical CAD model and the actual component deformation. Combined with the model prediction control algorithm, the scanning path is planned to ensure that the probe fits along the normal direction of the surface. The inertial measurement unit is used to collect attitude data in real time for beam delay compensation to suppress the sound beam offset caused by the slight vibration of the robotic arm, thereby improving the signal-to-noise ratio of the C-scan image by about 30%. Based on this, the sound group velocity of composite materials in different directions is accurately calculated using an anisotropic sound velocity model. Combined with the Vision Transformer deep learning network, pixel-level semantic segmentation of defects is achieved with a depth measurement error of ≤0.1mm. Finally, the three-dimensional information of defects is directly mapped to projected pixel coordinates through a pre-calibrated pinhole camera model, improving the spatial defect positioning error from ±1.8mm in the traditional scheme to ±0.1mm, and reducing the in-situ projection visualization error from ±2.5mm to ≤0.3mm. This effectively overcomes the problems of limited operating range, low collaborative efficiency, and lagging defect positioning in the traditional scheme from the perspective of hardware architecture and algorithm collaboration. It is suitable for high-precision non-destructive testing and in-situ defect visualization of large-size complex curved surface components in aerospace panels, wind turbine blades and other fields. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0025] Figure 1 The attached figure is a flowchart of the automated detection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy provided by the present invention; Figure 2 The attached figure is a module logic diagram of the automated detection and defect projection calibration method for complex curved surface components based on ultrasonic-laser synergy provided by the present invention; Figure 3 The attached figure is a three-dimensional structural schematic diagram of the ultrasonic-laser collaborative detection and defect projection calibration provided by the present invention; Figure 4 The attached figure is a schematic diagram of another angle of the ultrasonic-laser collaborative detection and defect projection calibration provided by the present invention; Figure 5 The attached figure is a schematic diagram of the structure of the ultrasonic detection module and the laser scanning projection module provided by the present invention.

[0026] The components are: 1-Three-axis gantry; 2-Six-axis industrial robot; 3-Tooling module; 4-Ultrasonic-laser collaborative detection and projection assembly; 5-Main control processing unit; 6-Workpiece to be tested; 7-X-axis crossbeam; 8-Ultrasonic detection module; 9-Laser scanning projection module; 10-Tool station; 11-Tool tray; 12-Z-axis lifting mechanism. Detailed Implementation

[0027] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] See Figures 1 to 5 As shown in the figure, this invention discloses an automated inspection and defect projection calibration method for complex curved surface components based on ultrasonic-laser collaboration. This method is implemented using an automated inspection system, which mainly includes a three-axis gantry crane 1, a six-axis industrial robot 2, a tooling module 3, an ultrasonic-laser collaborative inspection and projection component 4, a main control processing unit 5, and a workpiece to be tested 6. This invention effectively avoids repetitive calibration processes between multiple devices through integrated component design, achieving defect projection calibration at the hardware level. The specific structure is as follows: The three-axis gantry frame 1 adopts an integrated structure of crossbeam-slide table-lifting mechanism, including X-axis crossbeam 7 (effective stroke 0-8m) and Z-axis lifting mechanism 12 (effective stroke 0-2m). Each axis is equipped with servo motors and grating rulers, with a positioning accuracy of ±0.02mm. It can drive the robotic arm below and the ultrasonic-laser collaborative component 4 to achieve large-scale planar movement and vertical lifting, ensuring the collaborative operation range of ultrasonic testing and laser scanning and projection, and fully covering the testing area of ​​large-size components.

[0029] Six-axis industrial robot 2: Installed at the lower flange of the Z-axis lifting mechanism 12 of the three-axis gantry. The end flange integrates an automatic tool switching master side, which adopts a quick-change interface design. The positioning accuracy of the tool disk is ±0.01mm, which can realize the rapid grasping and replacement of the laser scanning projection module and the ultrasonic detection module, with a switching time of ≤10 seconds. The robotic arm has a working radius of 1.5m and works in conjunction with the gantry to form a composite motion capability of large-range movement and high-precision flexible operation, providing reliable posture adaptation guarantee for ultrasonic-laser collaborative operation.

[0030] Tooling module 3: It is fixed to the X-axis beam 7 of the three-axis gantry by bolts, located within the working radius of the robotic arm, and moves synchronously with the gantry without the need for additional positioning, ensuring the coordinate consistency between the ultrasonic detection module 8 and the laser scanning projection module 9 when switching. The tooling module has two built-in tool stations 10, which respectively support the laser scanning projection module 9 and the ultrasonic testing module 8. Both modules are equipped with tool sides 11 that are compatible with the main disk at the end of the robotic arm, enabling rapid and precise docking. The ultrasonic testing module 8 has an integrated IMU and phased array probe on its outer shell. Internally, it integrates a 128-element phased array ultrasonic probe (center frequency 5MHz) and a high-precision inertial measurement unit (IMU, sampling frequency 1kHz, attitude accuracy ±0.05°). The probe front end is equipped with an automatic coupling agent supply device, which continuously sprays water during scanning to ensure stable transmission of ultrasonic signals. The laser scanning projection module 9 adopts a line laser scanning method (scanning accuracy ±0.03mm) and has a built-in high-brightness laser projection system (projection resolution 1920×1080, projection distance 0.5-3m), realizing the integration of laser scanning and defect projection functions. It also shares the unified installation benchmark of the tooling module with the ultrasonic testing module, providing core hardware benchmark support for ultrasonic-laser collaborative operation and defect projection calibration.

[0031] Ultrasonic-laser collaborative detection and projection component 4: mainly composed of ultrasonic detection module 8, laser scanning projection module 9 and tool station 10.

[0032] Laser projection module 9: As the core integrated unit of the laser scanning projection module, it shares the same housing, mounting reference, and pre-calibrated coordinate system with the laser scanning unit. This completely eliminates the repetitive calibration process involving multiple coordinate systems of the projector, workpiece 6, ultrasonic testing module 8, and robotic arm in traditional solutions, directly achieving the core goal of defect projection calibration from an architectural design perspective. This projection system uses an RGB laser light source with a brightness ≥5000 lumens. It can project real-time visual information such as defect contours and depth values ​​based on ultrasonic testing results, with a projection error ≤0.3mm. It adapts to the distortion correction requirements of complex curved surfaces, achieving precise collaborative linkage between ultrasonic testing and laser projection, forming a closed loop for integrated testing-projection calibration operations.

[0033] Workpiece 6 to be tested: The complex curved surface component is placed on a high-precision adjustable worktable. The worktable can be adjusted horizontally and tilted to ensure that the workpiece inspection surface is within the optimal working range of the robotic arm and projection system.

[0034] Main Control Processing Unit 5: Set up as a PLC + industrial computer integrated control cabinet, the cabinet surface is equipped with a human-machine interface (HMI). Internally, it integrates a motion control card, data acquisition card, and deep learning inference module, primarily responsible for ultrasonic-laser coordinated control, multimodal data fusion, and defect projection driving functions. The main control processing unit communicates with the three-axis gantry and six-axis robotic arm via EtherCAT bus, synchronously acquires laser scanning data, ultrasonic detection data, and IMU attitude data via USB 3.0 interface, and drives the laser projection system via serial port, realizing coordinated control and real-time data transmission of various components. This constructs an integrated control logic for ultrasonic detection, laser scanning, and defect projection, providing core control assurance for calibrating projection accuracy and ultrasonic-laser coordinated stability.

[0035] Connection relationships of each component: The main control processing unit 5 uses dashed lines with arrows to indicate the signal flow. The arrows point from the main control processing unit to the three-axis gantry 1, the six-axis industrial robot 2, the laser scanning projection module 9, and the ultrasonic detection module 8, respectively. This enables the issuance of motion control commands, multimodal data acquisition commands, and projection drive commands. At the same time, it receives feedback data from each component (such as position signals, attitude data, detection data, and scanning data). The ultrasonic-laser coordination relationship is calibrated in real time through data fusion algorithms. Relying on the unified benchmark integrated by the hardware, the projection system can be driven to work accurately without any additional calibration operations.

[0036] Based on the above system, the detection method of the present invention includes the following steps: Step S1: System Initialization and Collaborative Reference Establishment: The main control processing unit 5 controls the three-axis gantry 1 to move the robotic arm to the initial position, completes the coordinate calibration of the robotic arm, gantry, and tooling module, establishes a unified coordinate system, and simultaneously establishes the initial collaborative reference of the ultrasonic testing module 8 and the laser scanning projection module 9. This reference is fixed based on the unified installation reference of the tooling module 3, and there is no need to separately calibrate the relative positions of the two, providing a prerequisite for subsequent full-process calibration; the complex curved surface component to be tested is placed on the worktable, and the attitude of the worktable is adjusted so that the component detection area is directly facing the collaborative operation range of the laser scanning projection module 9 and the ultrasonic testing module 8; the workpiece material parameters (such as the elastic stiffness matrix and density of carbon fiber composite materials) and collaborative detection parameters (such as ultrasonic probe frequency, laser scanning step distance, and projection distortion correction coefficient) are input through the HMI.

[0037] Step S2: Laser Scanning and Surface Reconstruction: The main control processing unit 5 instructs the laser scanning projection module 9 on the tooling module at the end of the robotic arm to grasp the tooling, and controls the three-axis gantry 1 to move the robotic arm along the planned area on the workpiece surface. The laser scanning projection module 9 performs line laser scanning on the workpiece surface to acquire discrete point cloud data (point cloud density ≥ 100 points / mm²). After denoising (using Gaussian filtering algorithm) and registration processing of the acquired point cloud data, the parameterized mathematical model of the workpiece surface is constructed using the NURBS algorithm. The expression is:

[0038] in, for 3D control point array, The weighting factor for the corresponding control point (value range 0.1-10). and They are respectively Towards Subsequent Towards B-spline basis functions of degree (set) (to ensure second-order smoothness of the surface) These are the surface parameter domain variables; the reconstructed surface model is used for subsequent path planning to solve the problem of adapting the theoretical CAD model to the deformation of the actual workpiece.

[0039] Step S3: Scan Path Planning: The main control processing unit 5 inputs the NURBS surface model into the Model Predictive Control (MPC) algorithm to construct the cost function:

[0040] in, The predicted system state vector (including robot joint angles, angular velocities, and gantry position). Let be the desired trajectory state vector. and This is the weight matrix. To predict the time domain, under the premise of satisfying the constraints of the robot arm dynamics and the gantry movement speed (X / Y axis ≤ 0.5 m / s, Z axis ≤ 0.2 m / s), the scanning trajectory of the ultrasonic probe is planned to ensure that the probe is always in contact with the normal direction of the workpiece surface and the contact gap is ≤ 0.1 mm.

[0041] Step S4: Automatic Tool Switching and Ultrasonic Detection (Maintaining Coordinate Consistency): The main control processing unit 5 controls the three-axis gantry 1 to drive the robotic arm back to the position of the tooling module 3, and places the laser scanning projection module 9 back to the corresponding workstation (relying on the positioning reference of the tooling module to maintain constant coordinates). After the main disk at the end of the robotic arm separates from the tool disk, it grabs the ultrasonic detection module. Since the tooling module 3 and the gantry move synchronously and the positioning reference is unified, the relative positions of the ultrasonic detection module 8 and the laser scanning projection module 9 have been pre-established, and there is no need to recalibrate the coordination relationship after switching. Along the trajectory planned in step S3, the gantry and the robotic arm move in coordination, the ultrasonic detection module 8 starts the scanning operation, and the automatic coupling agent supply device continuously supplies coupling agent to ensure the stability of ultrasonic signal transmission.

[0042] Step S5: Attitude Compensation and Signal Processing: During the scan, the IMU acquires the real-time attitude quaternions of the ultrasound probe at a frequency of 1 kHz. The main control processing unit combines it with the expected pose quaternion of the planned path. Compare and calculate the deviation quaternion. ( This is quaternion multiplication. for (conjugate quaternions); according to Analysis of roll angle, pitch angle, and yaw angle deviations Calculate the delay correction amount of each element of the phased array probe. This is superimposed on the beamforming law to compensate for beam offset caused by attitude deviation in real time, ensuring that the sound beam is incident perpendicularly on the workpiece surface.

[0043] Step S6: Defect Identification and Depth Calculation: The compensated ultrasonic echo signal is converted into a C-scan image and input into a pre-trained Vision Transformer (ViT) deep learning network. Pixel-level semantic segmentation is achieved through a multi-head self-attention mechanism to identify defect regions (such as delamination and cracks) and mark defect contours. Combining workpiece material parameters and based on a material anisotropic sound velocity model, the Christoffel equation is solved. The sound wave is obtained along the direction of propagation. group velocity ( For Christoffel tensors, For material density, It is the identity matrix. (where the sound wave polarization direction vector is used); based on the defect echo flight time. Calculate the depth of the defect (Coefficient 1 / 2 is the ultrasonic echo round-trip path correction factor), depth measurement error ≤ 0.1mm. The main control processing unit fuses the defect's three-dimensional coordinates, contour and depth data with the surface model data obtained in step S2 to generate the high-precision data required for laser projection.

[0044] Step S7: Defect Projection Calibration and Data Storage: Based on the unified pre-calibrated coordinate system built into the laser scanning projection module 9, the main control processing unit, combined with the surface model from step S2 and the attitude compensation data from step S5, directly constructs the correspondence between the three-dimensional coordinates of the defect and the projected pixel coordinates through the pinhole camera model without any additional coordinate system mapping calibration, satisfying the formula... ( As a scale factor, These are the projected pixel coordinates. For the three-dimensional coordinates of the defect, This is the intrinsic parameter matrix of the projector. (External parameter matrix); drives the projection system integrated in the laser profilometer to project the defect contour (red line) and depth value (blue text) onto the workpiece surface in situ. The deviation between the projected position and the actual defect position is ≤0.3mm; accurately implements the defect projection calibration function, forming an integrated calibration operation closed loop of ultrasonic testing and laser projection; at the same time, the main control processing unit stores ultrasonic testing data, laser scanning data, defect information and projection records, and generates an integrated testing report, which can be previewed or exported through HMI.

[0045] Step S8: Inspection Completion and Reset: After the scanning operation is completed, the main control processing unit 5 controls the robotic arm to put the ultrasonic inspection module 8 back into the tooling module 3, the gantry moves the robotic arm back to the initial position, the system shuts down all execution components, and the inspection process is completed.

[0046] In this embodiment, a comparative experiment was conducted based on a 5m×5m complex curved surface composite material component. The experiment was divided into two groups. The first group adopted a traditional independent module scheme, in which the ultrasonic probe, laser scanner, and laser projector were arranged independently or separately. Each time the tool was switched or multiple devices were linked, hand-eye calibration needed to be re-performed to solve the spatial mapping relationship between the projector, sensor, and workpiece. The second group adopted the scheme of this invention, in which laser scanning and laser projection were integrated into a design and processing assembly with a follow-up reference. The experimental data comparison is as follows:

[0047] The experimental data above show that the present invention is significantly superior to traditional solutions in terms of detection efficiency, tool switching speed, defect location accuracy, and projection visualization accuracy. In traditional solutions, the robotic arm must re-perform tedious hand-eye calibration each time it grasps a different tool or performs multi-device linkage. Multi-coordinate system transformations easily introduce the accumulation of repeated positioning errors and hand-eye calibration errors of the robotic arm. In contrast, the present invention starts from the hardware physical architecture, integrating the laser scanning and projection systems into a single housing. Both share a unique pre-calibrated internal coordinate system. At the same time, the tooling module 3 rigidly follows the three-axis gantry 1, giving the robotic arm a unique physical reference when grasping. This breaks the dependence of traditional solutions on temporary coordinate system transformations at the software level and realizes calibration-free operation throughout the entire process.

[0048] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0049] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for automated inspection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, characterized in that, Includes the following steps: S1. Establish a system coordinate system, under which an ultrasonic detection module (8) and a laser scanning projection module (9) are integrated, and the scanning unit and the projection unit of the laser scanning projection module (9) share the same module coordinate system; the module coordinate system has a fixed relative pose transformation relationship between the scanning unit and the projection unit; S2. Obtain workpiece surface data and reconstruct surface model through the laser scanning projection module (9); S3. Planning detection paths based on reconstructed surface models; S4. Switch the laser scanning projection module (9) to the ultrasonic detection module (8). After switching, there is no need to recalibrate the coordination relationship. S5. Perform ultrasonic testing along the detection path planned in step S3, and collect real-time attitude data to compensate for the detection signal; S6. Identify defects based on detection signals and calculate the three-dimensional information of the defects; S7. Based on the transformation relationship between the module coordinate system and the system coordinate system, the three-dimensional information of the defect is directly mapped to the projection pixel coordinates, and the projection unit is driven to project the defect information onto the workpiece surface in situ.

2. The method for automated detection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, as described in claim 1, is characterized in that... The specific method for establishing the system coordinate system includes: setting up a three-axis gantry (1), a six-axis industrial robot (2) installed at the execution end of the three-axis gantry (1), and an ultrasonic-laser collaborative detection and projection component (4). The ultrasonic-laser collaborative detection and projection component (4) includes a tooling module (3) fixed to the three-axis gantry and moving synchronously with it. The tooling module (3) is provided with two tool stations (10), and the two tool stations (10) respectively carry the ultrasonic detection module (8) and the laser scanning projection module (9).

3. The method for automated detection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, as described in claim 2, is characterized in that... The effective stroke of the X-axis crossbeam (7) of the three-axis gantry (1) is 0-8m, the effective stroke of the Z-axis lifting mechanism (12) of the three-axis gantry is 0-2m, the end flange of the six-axis industrial robot (2) integrates the tool automatic switching master disk, the positioning accuracy of the tool disk (11) is ±0.01mm, the tool switching time is ≤10 seconds, and the working radius of the six-axis industrial robot (2) is 1.5m.

4. The method for automated detection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, as described in claim 1, is characterized in that... The ultrasonic testing module (8) integrates a 128-element phased array ultrasonic probe and an inertial measurement unit. The sampling frequency of the inertial measurement unit is 1kHz and the attitude accuracy is ±0.05°. The ultrasonic testing module (8) is also equipped with an automatic coupling agent supply device at the front end.

5. The method for automated detection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, as described in claim 4, is characterized in that... The inertial measurement unit acquires the real-time attitude quaternions of the ultrasonic probe at a frequency of 1 kHz. The main control processing unit (5) compares it with the expected pose quaternion of the planned path. Compare and calculate the deviation quaternion. in, This is quaternion multiplication. for The conjugate quaternion; according to The roll, pitch, and yaw angle deviations are analyzed, and the delay correction of each element of the phased array probe is calculated. This is superimposed on the beamforming law to compensate for beam offset caused by attitude deviation in real time, ensuring that the sound beam is incident perpendicularly on the workpiece surface.

6. The method for automated inspection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, as described in claim 1, is characterized in that... The laser scanning projection module (9) has a line laser scanning accuracy of ±0.03mm, the laser projection unit uses an RGB laser light source, the projection resolution is 1920×1080, the projection distance is 0.5-3m, and the projection error is ≤0.3mm.

7. The method for automated inspection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy according to claim 1, wherein in step S2, a parameterized mathematical model of the workpiece surface is constructed using the NURBS algorithm, and the expression of the NURBS model is: in, for 3D control point array, These are the weighting factors for the corresponding control points; and They are respectively Towards Subsequent Towards B-spline basis functions of degree 1.

8. The method for automated detection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy, as described in claim 1, is characterized in that... In step S3, a model predictive control algorithm is used to plan the scanning path and construct the cost function: in, For the predicted system state vector, Let be the desired trajectory state vector. and This is the weight matrix. To predict the time domain, under the premise of satisfying the dynamic constraints of the robotic arm and the speed constraints of the gantry movement, the scanning trajectory of the ultrasonic probe is planned to ensure that the probe is always in contact with the normal direction of the workpiece surface and the contact gap is ≤0.1mm.

9. The method for automated detection and defect projection calibration of complex curved surface components based on ultrasound-laser synergy, as described in claim 1, wherein in step S6, a pre-trained Vision Transformer deep learning network is used to perform pixel-level semantic segmentation on the ultrasonic C-scan image, identify defect regions, and mark defect contours; based on the material anisotropic sound velocity model, the Christoffel equation is solved. The sound wave is obtained along the direction of propagation. group velocity Christoffel tensor For material density, identity matrix The direction vector of acoustic wave polarization; based on the defect echo flight time Calculate the depth of the defect ; Depth measurement error ≤ 0.1mm; The main control processing unit (5) fuses the three-dimensional coordinates, contour and depth data of the defect with the surface model data obtained in step S2 to generate the high-precision data required for laser projection.

10. The method for automated inspection and defect projection calibration of complex curved surface components based on ultrasonic-laser synergy according to claim 1, characterized in that, In step S7, the correspondence between the three-dimensional coordinates of the defect and the projected pixel coordinates is constructed using the pinhole camera model, satisfying the formula... ;in, scale factor Projected pixel coordinates Three-dimensional coordinates of the defect Projector intrinsic parameter matrix This is the extrinsic parameter matrix.

Images