Robot phased array nondestructive testing defect positioning method based on man-machine cooperation

Abstract

The invention discloses a robot phased array nondestructive testing defect positioning method based on man-machine cooperation. The method comprises the following steps: acquiring a contour image of a workpiece to be scanned; the point cloud data are converted into contour point cloud data, and splicing and filtering of the point cloud data are completed; multiple points in different directions are picked up, and point cloud cutting is completed; fitting to obtain a curve surface, and generating a robot scanning track; normal information at contour feature points on the track is extracted, the tail end posture of the robot is determined, and a nondestructive testing scanning path is generated; the robot scans according to the scanning path, and the phased array scanning probe analyzes damage echo data; and fitting scanning data to form a damage cloud picture, extracting pose information of the robot corresponding to the defect area, judging whether secondary scanning is needed or not, and meanwhile, judging the scanned defect damage type. According to the method, the detection process is simple and convenient to operate, manual coordinate system calibration is not needed, secondary scanning of defect positions is facilitated, and the method can be suitable for flaw detection operation of workpieces made of multiple materials and in multiple models.

Description

Technical Field

[0001] The present invention belongs to the technical field of non-destructive testing of materials, and particularly relates to a method for defect location in robotic phased array non-destructive testing based on human-machine collaboration. Background Art

[0002] During the manufacturing process of advanced composite materials, due to the difficulty in precisely controlling various process parameters, the quality of composite material structures is unstable and has a certain degree of randomness. Static loads, mechanical damage, fatigue, creep, overheating, etc. during use can also cause damage in composite materials. The generation, propagation, and accumulation of damage will exacerbate the environmental and stress corrosion of the materials, accelerate the aging of the materials, cause a serious decline in the thermal performance of the materials, a sharp loss of strength and stiffness, and greatly reduce the service life of the structure. Therefore, it is extremely important to perform non-destructive testing on composite material structures before and during use. In addition, the high cost of spacecraft launch and operation requires the weight of spacecraft structures to be as light as possible. To reduce the weight of aerospace composite material structures and manufacturing costs, it is necessary to adopt damage detection technology to accurately detect and identify various defects and damages in the materials.

[0003] As an effective means of material defect detection, compared with other detection technologies, phased array non-destructive testing can detect microscopic small defects, and has the characteristics of high sensitivity, strong penetration ability, precise positioning, harmless to materials, environmental protection and hygiene. Ultrasonic testing is widely used as an important means to control and identify product quality at home and abroad. However, currently in many industries, the non-destructive testing of products is still carried out manually by workers holding probes to scan parts. A large amount of human resources are consumed during the entire detection process and the scanning efficiency is low. At the same time, manual scanning greatly affects the reliability of measurement results, easily causes missed detection and false detection of defects, and causes losses to subsequent production. Moreover, some large-configured and multi-curved products cannot provide product digital models, or the material components have poor rigidity and large deformations. The traditional operation of generating non-destructive testing paths using theoretical digital models is complex and cannot solve the problem of deviation between the theoretical digital model and the physical part caused by error accumulation during the manufacturing and assembly processes of actual parts. And complicated calibration work is required to ensure the accuracy of machining NC. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for defect location in robotic phased array non-destructive testing based on human-machine collaboration in view of the above-mentioned prior art deficiencies.

[0005] To achieve the above technical objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for defect location in robotic phased array non-destructive testing based on human-machine collaboration, comprising the following steps:

[0007] S1. Take multiple photos through the binocular camera fixed on the end effector of the robot to obtain the contour image of the workpiece to be scanned;

[0008] S2. Convert the contour image into contour point cloud data in the robot base coordinate system, complete the stitching of the point cloud data, and perform filtering processing on the point cloud data;

[0009] S3. Manually pick multiple points in different orientations on the point cloud image, and connect the interactive points in sequence according to the picking order to automatically complete point cloud clipping, obtaining the clipped point cloud contour set;

[0010] S4. Perform NURBS surface fitting on the clipped point cloud contour set to obtain the curve surface S, set the parameters of the scanning trajectory on the upper computer, and automatically generate the robot scanning trajectory C on the curve surface S according to the set process parameters;

[0011] S5. Extract the normal information at the contour feature point P i on the robot scanning trajectory C, use the contour feature point as the machining path point, and determine the posture of the robot end according to the normal information to generate the robot non-destructive testing scanning path;

[0012] S6. The upper computer directly sends the non-destructive testing scanning path to the robot. The robot executes the motion trajectory. After the robot moves to the area to be scanned, the upper computer sends an instruction to turn on the phased array scanning probe and make scanning preparations;

[0013] S7. Adopt the method of synchronous task fixed-period triggering for data recording, start the task synchronization and data collection threads at the same time. The robot moves according to the scanning path, and at the same time the phased array performs damage scanning. The upper computer synchronously collects the phased array scanning echo data and the robot motion data at the set frequency and records the time stamp;

[0014] S8. After the scanning is completed, the upper computer aligns the transmitted damage echo data information and the robot motion data according to the time stamp, constructs a visual damage cloud map, analyzes the distribution and severity of the damage, and obtains the defect area that needs to be scanned again;

[0015] S9. Use the damage cloud map data obtained from each scanning as a sample set to train and establish a defect classification model, and then automatically identify and display the categories of scanning defects;

[0016] S10. For the defect area that needs to be scanned again, the upper computer automatically extracts the pose information of the robot corresponding to the defect area, and sends the extracted pose information to the robot again for secondary scanning.

[0017] To optimize the above technical solution, the specific measures taken also include:

[0018] In the above-mentioned step S1, the overall detection range of the binocular camera is a square area with 2.3MP pixels, and the camera collects multiple adjacent areas.

[0019] In the above-mentioned step S2, Gaussian filtering is used to filter the point cloud data, and the voxel grid method is used to downsample the point cloud data.

[0020] In the above-mentioned step S3, multiple points in different orientations are picked interactively on the point cloud image, and the interactive points are connected in sequence according to the picking order to automatically complete point cloud cropping, obtaining the cropped point cloud contour set. The specific process includes:

[0021] S31. Pick three or more points in different orientations according to the processing requirements;

[0022] S32. The three-dimensional position information of the picked points is combined with the normal information to form a slicing plane P l , and the point cloud data band K surrounded by the slicing plane P l and the point cloud surface E is the cropped point cloud contour set.

[0023] In the above-mentioned step S4, the scanning trajectory parameters include the scanning trajectory type, the scanning point interval, the robot moving speed, and the turning area radius.

[0024] The scanning preparation in the above-mentioned step S6 means that after the robot moves to the area to be scanned, it continues to execute the motion instruction to press the non-destructive testing end against the area to be scanned, opens the water inlet pump on the end effector, and after the inner cavity of the non-destructive testing end effector is filled, opens the water outlet pump installed on the end effector to achieve the dynamic balance of water inlet and outlet.

[0025] In the above-mentioned step S7, the synchronous task fixed-period triggering method is adopted to start the task synchronization and data collection threads simultaneously. Using the custom synchronous task triggering mechanism, the data acquisition signal is triggered at a fixed time interval, and the phased array scanning echo data and the robot motion data are synchronously acquired based on the TTL pulse, and the time stamp is recorded. The obtained phased array scanning corresponding damage echo signal data set is P = {q1, q2,..., q n} and the current motion pose set of the robot is P F = {p1, p2,..., p n} to realize the matching of the robot pose and the damage echo signal during the scanning process.

[0026] The above-mentioned step S8 specifically includes the following steps:

[0027] S81. The host computer aligns the robot pose data and the damage echo signal data transmitted back during the scanning process according to the time stamp, maps the probe amplitude information arranged in order in the damage echo signal data into the depth of color, and constructs a damage cloud map of the workpiece through color coding;

[0028] S82. When the cloud map shows that there are damage defects in a certain area of the workpiece, retrieve the damage cloud map, number the retrieved defect areas in sequence, and increment the count by 1 for each additional defect area until the final retrieval is completed, realizing defect counting and defect numbering;

[0029] S83. Manually select the area that needs to be scanned again or automatically analyze the damage cloud map by the system according to a preset threshold. When the characteristic value of the damage area exceeds the preset threshold, the area will be automatically marked as the defect area that needs to be scanned again, and the robot pose information of the current area will be extracted.

[0030] The above step S9 specifically includes the following steps:

[0031] S91. Classify the damage situation according to the damage cloud map;

[0032] S92. Divide the defects into pore defects, crack defects, inclusion defects, pit defects, scratch defects and label the defect images to form a damage image data set, and randomly divide the data set into a training set, a validation set and a test set;

[0033] S93. Introduce the GAM attention mechanism. GAM includes two attention modules: channel attention and spatial attention, and uses three-dimensional arrangement to maintain the integrity of information;

[0034] S94. Introduce the normalized Wasserstein distance as the optimization loss function, weight each pixel in the anchor box, and model the anchor box as a two-dimensional Gaussian distribution;

[0035] S95. Construct a defect classification model through the GAM attention mechanism and the normalized Wasserstein distance to output the defect classification of the area for the specific cloud map defect.

[0036] The above step S10 specifically includes the following steps:

[0037] S101. Extract the phased array damage echo signal data set corresponding to the point cloud that needs to be scanned again. If the damage echo signal data to be extracted is Q = {q1, q2,..., q m}, then the set of robot poses corresponding to the defect area during scanning is Q F = {p1, p2,..., p m}, and send the corresponding robot pose information to the robot;

[0038] S102. Use the EtherCAT industrial real-time Ethernet protocol to transmit signals with the robot, read the robot's motion pose data through the TCP / IP protocol, and at the same time compile it into custom xml format robot pose data. The robot reads the xml pose data and moves to the issued position, or the operator manually drags the end of the robot to the position where the workpiece defect is scanned for secondary scanning.

[0039] The present invention has the following beneficial effects:

[0040] The present invention uses the robot as a carrier for non-destructive testing scanning, without the need to obtain the traditional product theoretical digital model, avoiding the problem of difficult model registration caused by the production error between the actual product and the theoretical digital model, effectively eliminating the problem of difficult acquisition of the digital model of some large-configuration and multi-curvature products, and the problem of difficult model registration caused by the production error between the actual product and the theoretical digital model. Without the need to obtain the traditional product theoretical digital model, it solves the problem that the robot processing NC output based on the traditional product theoretical digital model has a scanning path that cannot meet the non-destructive testing requirements.

[0041] The point cloud data of the present invention is visualized. The operator interactively selects the non-destructive testing scanning area in the visualization window according to his own process operation habits, converts the point cloud data into the motion path of the robot for non-destructive testing scanning, completes the path planning of the non-destructive testing scanning, effectively detects the damage and locates the defect, which is convenient for the robot to perform secondary scanning. The detection process is simple and convenient, and it can realize the free placement of the workpiece to be scanned without manual coordinate system calibration;

[0042] The present invention constructs a workpiece defect cloud map according to the scanning information to display the internal scanning situation of the workpiece, converts the scanning situation into a three-dimensional cloud map for display, and can realize the quantitative extraction of defects and the robot pose positioning and extraction of the defect position, which is convenient for the secondary scanning of the defect position.

[0043] The present invention effectively avoids problems such as low efficiency and high labor intensity of traditional manual detection, realizes the quantitative analysis of defects, locates and extracts the robot pose information corresponding to the defects, drives the robot to perform re-detection scanning, improves the human-machine interaction and scanning efficiency, and without the need for a product digital model and multiple product coordinate system calibrations, it can automatically convert 3D point clouds and process requirements into robot motion instructions, with simple and convenient operation, and can be applied to the flaw detection operations of workpieces with multiple materials and shapes. Description of the Drawings

[0044] Figure 1 It is the system flowchart provided by the embodiment of the present invention;

[0045] Figure 2 It is the hardware configuration diagram provided by the embodiment of the present invention;

[0046] Figure 3 This is the flowchart of point cloud trajectory generation provided by the embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the point cloud contour set obtained by interactive picking in the detection system block diagram provided by the embodiment of the present invention;

[0048] Figure 5 This is the detection system block diagram provided by the embodiment of the present invention. Detailed implementation manners

[0049] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0050] Although the steps in the present invention are arranged with reference numerals, they are not used to limit the order of the steps. Unless the order of the steps is clearly stated or the execution of a certain step requires other steps as a basis, the relative order of the steps can be adjusted. It can be understood that the term "and / or" used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0051] Embodiment 1

[0052] Refer to Figures 1-5 , a method for defect location in phased array non-destructive testing of robots based on human-machine collaboration in the present invention is implemented based on a six-degree-of-freedom robot, a non-destructive testing end effector, a phased array probe, a host computer, etc. The steps included in the specific implementation manners are as follows:

[0053] S1. Multiple photos are taken by a binocular camera fixedly connected to the end effector of the robot to obtain the contour image of the workpiece to be scanned;

[0054] S2. The contour image is converted into contour point cloud data in the robot base coordinate system, the point cloud data is spliced, and the point cloud data is filtered;

[0055] S3. The operator picks multiple points in different orientations on the point cloud image according to the non-destructive testing process habits, and connects the interactive points in sequence according to the picking order, and automatically completes the point cloud clipping to obtain the clipped point cloud contour set;

[0056] S4. The clipped point cloud contour set is subjected to NURBS surface fitting to obtain a curve surface S. The operator sets the parameters of the scanning trajectory on the host computer and automatically generates a robot scanning trajectory C on the curve surface S according to the set process parameters;

[0057] S5. Extract the normal information at the contour feature point P on the upper contour of the robot scanning trajectory C, use the contour feature point as the machining path point, and determine the pose of the robot end according to the normal information to generate the robot non-destructive testing scanning path; i At the place, and determine the pose of the robot end according to the normal information to generate the robot non-destructive testing scanning path;

[0058] S6. The host computer directly sends the non-destructive testing scanning path to the robot. After the robot executes the motion trajectory and moves to the area to be scanned, the host computer sends an instruction to turn on the phased array scanning probe and make scanning preparations;

[0059] S7. Adopt the method of triggering at a fixed period of synchronous tasks to record data. At the same time, start the task synchronization and data collection threads. The robot moves according to the scanning path, and at the same time the phased array conducts damage scanning. The host computer collects the phased array scanning echo data and the robot motion data at the set frequency and records the time stamp;

[0060] S8. After the scanning is completed, the host computer aligns the transmitted damage echo data information and the robot motion data according to the time stamp, constructs a visual damage cloud map, analyzes the distribution and severity of the damage, and obtains the defect area that needs to be scanned again. The damage cloud map represents the severity of the damage through color coding, so that the operator can judge the defect area that needs to be scanned again according to the damage cloud map, or the system automatically identifies the defect area that needs to be scanned again according to the preset threshold;

[0061] S9. Use the damage cloud map data corresponding to each scan as a sample set to train and establish a defect classification model, and then automatically identify and display the types of scanned defects;

[0062] S10. For the defect area that needs to be scanned again, the host computer automatically extracts the pose information of the robot corresponding to the defect area, and sends the extracted robot pose at the defect position to the robot again, driving the robot to move to the defect position of the scanned workpiece according to the pose instruction for secondary non-destructive defect scanning, or the operator manually drags the robot end to move to the defect position of the scanned workpiece for secondary scanning.

[0063] In the embodiment, in step S1, the part to be scanned is fixed on the fixture. After the camera calibration is completed, the 3D camera fixedly connected to the robot non-destructive testing end effector takes pictures of the workpiece to be scanned (the area to be processed). Move the robot end to make the binocular camera take multiple pictures of this area. The detection range of the binocular camera is: the overall detection range of the binocular camera is a square area with a pixel number of 2.3MP, and the camera performs multiple acquisitions of adjacent areas.

[0064] In step S2, Gaussian filtering is used to filter the point cloud data, and the voxel grid method is used to streamline the point cloud data.

[0065] In step S2, the information collected by the camera is converted into a three-dimensional point cloud model. Multiple sets of point cloud data collected by the camera are stitched together, and the point cloud is transformed into the robot's base coordinate system (that is, the obtained multiple sets of point cloud data are rotated into the robot's base coordinate system to form the scanned point cloud data), realizing the stitching of the point cloud. By stitching the point cloud, the non-destructive testing area is further expanded, and the obtained three-dimensional point cloud data is filtered. The specific process includes the following steps:

[0066] S21. Convert the images of adjacent areas collected by the camera multiple times into contour point cloud data in the robot's base coordinate system to complete the stitching of the point cloud data;

[0067] S22. Set a smoothing window of size Z×Z, and preset a convolution kernel for this window. Set the size x as the Euclidean distance from a certain position in the convolution kernel to the central pixel, f(x) as the value at this position, and e as the standard deviation of the Gaussian function.

[0068] S23. Traverse each pixel in the image with the convolution kernel to complete the smoothing process of Gaussian filtering.

[0069] S24. Use the voxel grid method to downsample the point cloud data.

[0070] In S24, the specific process includes the following steps:

[0071] S241. Establish three-dimensional voxels in the point cloud, and approximately represent the data points in the three-dimensional voxels using the centroid of the voxels.

[0072] S242. Set n grid (i) as the number of points in a voxel. For the three-dimensional point P ij (x ij ,y ij ,z ij ) in the voxel, the centroid P grid (i) is calculated as follows:

[0073]

[0074] In step S3, based on the stitched point cloud, the stitched point cloud is cropped in the human-machine interaction interface. The operator picks the corner points of the area to be subjected to non-destructive testing according to the detection / actual scanning requirements, and crops the stitched point cloud with the picked points as the boundary. The specific process includes the following steps:

[0075] S31. The operator interactively picks three or more interactive points according to the processing requirements.

[0076] S32. The three-dimensional position information of the picked surface feature interaction points is combined with the normal information to form a slice plane P. l , the slice plane P l and the point cloud data band K surrounded by the segmentation of the slice plane P and the point cloud surface E is the point cloud contour set picked by interaction.

[0077] In step S4, based on the cropped point cloud data, the point cloud data is fitted into a NURBS surface and the scanning trajectory parameters are set according to the scanning requirements, and the running trajectory / scanning processing that can be directly executed by the robot is automatically generated.

[0078] The manually set scanning trajectory parameters include but are not limited to the setting of the scanning trajectory type (Z-shaped, loop-shaped, etc.), the interval between scanning points, the moving speed of the robot, the radius of the turning area, etc.

[0079] In step S5, the specific process includes the following steps:

[0080] S51. Using the tangent vector n y , the normal vector n z and the vector n x are respectively the Y-axis direction, the Z-axis direction and the X-axis direction of the robot end attitude coordinate system.

[0081] S52. Suppose the direction vectors obtained after normalizing the rotation vectors of the X, Y, and Z axes of the robot end attitude coordinate system are respectively

[0082] S53. Calculate the rotation Euler angles according to the direction vectors , which are the attitudes of the robot end, and the expression of the rotation Euler angles is:

[0083]

[0084] In the above formula, are respectively the direction vectors obtained after normalizing the rotation vectors of the X, Y, and Z axes of the robot end attitude coordinate system.

[0085] An inlet water pump and an outlet water pump are also arranged on the non-destructive testing end effector. In step S6, the upper computer directly sends the non-destructive testing scanning path to the robot, and the robot executes the motion trajectory. After the robot moves to the area to be scanned, the upper computer sends an instruction to turn on the phased array scanning probe and make scanning preparations. The scanning preparation means that after the robot moves to the non-destructive testing area and the non-destructive testing end compacts the area to be scanned, the inlet water pump on the end effector is turned on. After the inner cavity of the non-destructive testing end effector is filled, the outlet water pump installed on the end effector is turned on to achieve the dynamic balance of water inlet and outlet.

[0086] In step S7, the host computer software starts the task synchronization and data collection threads. First, it sends start signals to both the robot and the phased array controller simultaneously to ensure the synchronization of data acquisition. Subsequently, it uses a custom synchronization task triggering mechanism to trigger data acquisition signals at fixed time intervals, and realizes microsecond-level synchronous acquisition of phased array scan echo data and robot motion data through TTL pulses, and records timestamps. To ensure low latency and stability in communication with the device, the EtherCAT industrial real-time Ethernet protocol is used to transmit signals with the robot, and the robot motion pose data is read through the TCP / IP protocol. At the same time, a custom xml format for robot pose data is defined to streamline the data size and reduce latency. Similarly, in the communication with the phased array controller, USB3.0 is used for high-speed data transmission. Based on the existing UDP protocol of the controller, a custom data transmission format is defined to reduce the amount of interactive data and ensure the real-time performance and integrity of the scan echo signal.

[0087] After the scan is completed, the host computer obtains the damage echo signal data set P = {q1, q2, ……, q n} and the robot motion pose set P F = {p1, p2,......, p n}, where the damage echo signal is the B-scan echo amplitude information of the phased array probe. Aligning the robot pose data with the damage echo signal data according to the timestamp can achieve precise matching of the pose and signal during the scan process.

[0088] In step S8, after reading the robot pose information and phased array defect information at a fixed period and realizing the registration of the robot pose information and the scan data, a workpiece defect cloud map is constructed according to the scan information, and defect counting is performed to realize the robot pose positioning and extraction of the defect position. The corresponding robot pose located according to the damage information displayed on the damage cloud map can conveniently drive the robot for secondary damage detection. Specifically, it includes the following steps:

[0089] S81. The host computer aligns the robot pose data and the damage echo signal data (including the reflection signal intensity and position information of the scan) transmitted during the scan process according to the timestamp, maps the probe amplitude information arranged in order in the damage echo signal data to the depth of color, where red represents a higher signal amplitude and blue or light color represents a lower signal amplitude. Through color coding, a damage cloud map of the workpiece is constructed and displayed on the host computer interface;

[0090] S82. When the cloud map shows that there are damage defects in a certain area of the workpiece, the scan cloud map is retrieved, and the retrieved defect areas are numbered in sequence. Each time a defect area is added, the count is incremented by 1 until the final retrieval is completed, realizing defect counting and defect numbering;

[0091] S83. For the area that requires secondary scanning selected manually, the system automatically analyzes the damage cloud map according to a preset threshold. The threshold is set based on the severity of the damage (such as the reflection signal intensity, damage area, etc.), and the threshold can be adjusted manually. When the characteristic value of the damage area exceeds the preset threshold, the area will be automatically marked as the defect area that requires secondary scanning, and the robot pose information of the current area will be extracted.

[0092] Step S9 specifically includes the following steps:

[0093] S91. Data collection: The operator classifies the damage situation according to the damage cloud map;

[0094] S92. Data annotation: The defects are classified into spherical (porosity defect), crack (crack defect), inclusion (inclusion defect), ptis (pit defect), scratch (scratch defect), etc., and the defect images are annotated using the image annotation tool Labelimg to form a damage image dataset. The dataset is randomly divided into a training set, a validation set, and a test set according to the ratio of 6:2:2;

[0095] S93. Introduce the GAM attention mechanism. GAM includes two key attention modules: channel attention and spatial attention, and uses three-dimensional arrangement to maintain the integrity of information;

[0096] S94. Introduce the Normalized Wasserstein Distance (NWD) as the optimization loss function, weight each pixel in the anchor box, and model the anchor box as a two-dimensional Gaussian distribution;

[0097] S95. Construct a defect classification model through the GAM attention mechanism and the normalized Wasserstein distance. When the initial sample accumulation is completed, the defect classification of the specific cloud map defect area can be output without manual classification.

[0098] In step S10, drive the robot to perform secondary scanning of the defect location, which specifically includes the following steps:

[0099] S101. Extract the phased array damage echo signal data set corresponding to the point cloud that requires secondary scanning. If the damage echo signal data to be extracted is Q = {q1, q2,......, q m}, then the set of robot poses corresponding to the defect area during scanning is Q F = {p1, p2,......, p m}, and the corresponding robot pose information is sent to the robot;

[0100] S102. Use the EtherCAT industrial real-time Ethernet protocol to transmit signals with the robot, read the robot's motion pose data through the TCP / IP protocol, and compile it into custom xml format robot pose data. The robot reads the xml pose data and moves to the assigned position, or the operator manually drags the end of the robot to the position where the workpiece defect is scanned for secondary scanning.

[0101] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claimed invention.

[0102] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A robot phased array nondestructive testing defect location method based on human-machine collaboration, characterized in that: The following steps are involved: S1, taking multiple photos with a binocular camera fixed on the end effector of the robot to obtain the contour image of the workpiece to be scanned; S2, converting the contour image into contour point cloud data in the robot base coordinate system, completing the splicing of the point cloud data, and filtering the point cloud data; S3, picking multiple points in different directions on the point cloud image through human-computer interaction, and connecting the interactive points in sequence according to the picking order, automatically completing the point cloud cropping, and obtaining the cropped point cloud contour set; S4, performing NURBS surface fitting on the cropped point cloud contour set to obtain a curve surface S, setting the parameters of the scanning trajectory on the host computer, and automatically generating a robot scanning trajectory C on the curve surface S according to the set process parameters; S5. Extract contour feature points P on the robot scanning trajectory C i The normal information at the position is used as the processing path point, and the posture of the robot end is determined according to the normal information to generate the robot non-destructive testing scanning path; S6. The host computer sends the nondestructive testing scanning path directly to the robot. The robot executes the motion trajectory. After the robot moves to the area to be scanned, the host computer sends a command to open the phased array scanning probe and prepare for scanning. S7, using the synchronous task periodic triggering method to record data, and starting the task synchronization and data collection threads at the same time, the robot moves along the scanning path, and the phased array performs damage scanning at the same time. The host computer synchronously collects the phased array scanning echo data and the robot motion data at the set frequency, and records the timestamp; S8. After the scan is completed, the host computer aligns the returned damage echo data information with the robot motion data according to the timestamp and constructs a visual damage cloud map to analyze the distribution and severity of the damage and derive the defective area that needs a second scan; S9. Using the damage cloud map data corresponding to each scan as a sample set to train and establish a defect classification model, and then automatically identify and display the categories of the scanned defects; S10. For the defective area that needs a second scan, the host computer automatically extracts the posture information of the robot corresponding to the defective area, and sends the extracted posture information to the robot again for a second scan.

2. The method for nondestructive testing defect positioning by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: In step S1, the overall detection range of the binocular camera is a square area, the number of pixels is 2.3MP, and the camera performs multiple acquisitions of adjacent areas.

3. The method for nondestructive testing defect positioning by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: In step S2, the point cloud data is filtered using a Gaussian filter, and the point cloud data is simplified using a voxel grid method.

4. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: Step S3 picks multiple points in different directions on the point cloud image through human-computer interaction, and connects the interactive points in sequence according to the picking order, automatically completes the point cloud cropping, and obtains the cropped point cloud contour set. The specific process includes: S31, picking up three or more points in different directions according to processing requirements; S32, the three-dimensional position information of the picked point is combined with the normal information to form a slice plane P l , slice plane P l The point cloud data band K separated and surrounded by the point cloud surface E is the point cloud contour set after clipping.

5. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: In step S4, the scanning trajectory parameters include scanning trajectory type, scanning point interval, robot moving speed, and turning area radius.

6. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: The scanning preparation in step S6 means that after the robot moves to the area to be scanned, it continues to execute the motion command to compact the area to be scanned by the non-destructive testing end effector, turns on the water inlet pump on the end effector, and after the inner cavity of the non-destructive testing end effector is filled, turns on the water outlet pump installed on the end effector to achieve a dynamic balance between water inlet and water outlet.

7. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: In step S7, the synchronous task is triggered periodically, and the task synchronization and data collection threads are started at the same time. The customized synchronous task trigger mechanism is used to trigger the data acquisition signal at a fixed time interval. The phased array scanning echo data and the robot motion data are synchronously collected based on the TTL pulse, and the timestamp is recorded. The acquired phased array scanning corresponding damage echo signal data set is P = {q1, q2, ..., q n }, the robot’s current motion pose set is P F ={p1, p2, ..., p n }, to achieve the matching of the robot posture and damage echo signal during the scanning process.

8. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1 is characterized in that: Step S8 specifically includes the following steps: S81, the host computer aligns the robot posture data and the damage echo signal data transmitted back during the scanning process according to the timestamp, maps the probe amplitude information arranged in sequence in the damage echo signal data to the depth of color, and constructs a damage cloud map of the workpiece through color coding; S82. When the cloud map shows that there is a damage defect in a certain area of ​​the workpiece, the damage cloud map is searched, and the searched defect areas are numbered in sequence. Each time a defect area is added, the count is increased by 1 until the search is finally completed, thereby realizing defect counting and defect numbering. S83. Manually select the area that needs to be scanned again or the system automatically analyzes the damage cloud map according to a preset threshold. When the characteristic value of the damaged area exceeds the preset threshold, the area is automatically marked as a defect area that needs a second scan, and the robot posture information of the current area is extracted.

9. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1, characterized in that: Step S9 specifically includes the following steps: S91. Classify the damage situation according to the damage cloud map; S92, classifying defects into pore defects, crack defects, inclusion defects, pit defects, and scratch defects, and annotating defect images to form a damage image data set, and randomly dividing the data set into a training set, a validation set, and a test set; S93, introduce the GAM attention mechanism. GAM includes two attention modules: channel attention and spatial attention, which uses three-dimensional arrangement to maintain the integrity of information. S94, introduce the normalized Wasserstein distance as the optimization loss function, weight each pixel in the anchor frame, and model the anchor frame into a two-dimensional Gaussian distribution; S95. A defect classification model based on YOLOv7 is constructed through the GAM attention mechanism and normalized Wasserstein distance to output the defect classification of the area for the specific cloud map defect.

10. The method for nondestructive testing defect location by a robot phased array based on human-machine collaboration according to claim 1, characterized in that: Step S10 specifically includes the following steps: S101, extracting the phased array damage echo signal data set corresponding to the point cloud that needs to be scanned again. If the damage echo signal data to be extracted is Q = {q1, q2, ..., q m }, then the robot pose set corresponding to the defect area during scanning is Q F ={p1, p2, ..., p m }, and send the corresponding robot posture information to the robot; S102. Use the EtherCAT industrial real-time Ethernet protocol to transmit signals with the robot, read the robot motion posture data through the TCP / IP protocol, and compile it into a custom XML format robot posture data. The robot reads the XML posture data and moves to the sent position, or the operator manually drags the end of the robot to the defect position of the workpiece for secondary scanning.

Images