Automatic thickness measurement method and system based on point cloud matching

Abstract

The present invention belongs to the technical field of thickness detection, and provides an automatic thickness measurement method and system based on point cloud matching. The method includes: establishing a point cloud model file of a standard test piece according to the external shape characteristics of the standard test piece, where the standard test piece is used to represent a standard part of an engine blade; automatically clamping the test piece to be measured, scanning the point cloud data of the test piece to be measured, and performing point cloud matching based on the point cloud model file to calibrate the position information of the target detection point of the test piece to be measured; performing a thickness detection action on the test piece to be measured after calibrating the position information; collecting the wall thickness of the test piece to be measured at the target detection point in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point; making a pass / fail judgment on the collected wall thickness of the target detection point, and displaying the detection data related to the target detection point in real time. The present invention realizes the automation of the detection action, ensures the measurement consistency, and improves the detection efficiency and measurement accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of thickness detection, in particular to the technical field of engine blade thickness detection, and specifically relates to an automatic thickness measurement method and system based on point cloud matching. Background Art

[0002] At present, the thickness detection of domestic blades mainly relies on workers to hold thickness gauges for measurement, and there is no automatic detection system. Turbine engine blades usually have air cooling channels designed inside to enable the blades to work in high-temperature environments. During the processing and production of blades, due to the influence of process levels, etc., the internal thickness of the blades may be lower than a certain safety threshold, which will seriously affect the working temperature of the turbine and pose a certain threat to flight safety. Most traditional blade thickness measurement methods manually mark the blades by hand and then hold a thickness gauge for measurement. This method is not only time-consuming and laborious but also brings artificial random errors. Therefore, it is particularly important to measure the wall thickness of engine blades (such as turbine blades).

[0003] Currently, the main methods for measuring turbine blades include coordinate measuring method, laser measuring method, structured light measuring method, binocular vision measuring method, laser triangulation measuring method, etc. The coordinate measuring method can accurately locate the spatial coordinate positions of each measurement point by adaptively sampling the blade, reducing the position error of the thickness measurement system. However, the equipment cost of this method is relatively high, and the measurement speed is slow, making it difficult to meet the speed requirements of actual production. The laser measuring method projects a large number of laser lines on the blade surface to generate three-dimensional data of the blade. This method has high accuracy and strong anti-interference ability. However, this method requires point-by-point scanning, has high requirements for equipment, and at the same time, the generated laser data volume is large, and the equipment's processing speed of laser data is limited. In addition, laser measuring equipment is relatively expensive, so laser measuring solutions have not been popularized in the market. The structured light measuring method can achieve dynamic, fast, accurate measurement of the blade. However, for turbine blades with a large curvature, the structured light measurement is prone to measurement blind spots, resulting in poor accuracy in the detection area that is not perpendicular to the structured light surface, and the point cloud data is relatively sparse, bringing large errors to the subsequent positioning of the spatial three-dimensional coordinates of the measurement points. The binocular vision measuring method obtains the depth information of the blade through two cameras, calculates the parallax, and then constructs a three-dimensional model of the object. However, this method is easily affected by light, and has a high algorithm complexity and slow response speed. In addition, electromagnetic Hall effect measurement means are also used to measure the wall thickness of the blade, and the electromagnetic Hall effect often causes the phenomenon of steel balls getting stuck and blocked in the inner cavity of the blade, resulting in inaccurate thickness measurement.

[0004] Existing manual inspection methods have the following problems: 1) Due to differences in manual measurement operating habits, as working hours and labor intensity increase, manual measurement will result in poor measurement consistency and low stability. 2) There are no obvious measurement point markings on the blade surface, making it difficult to locate the required measurement position by visual perception alone. Moreover, handheld instrument measurement cannot always ensure that the stylus and the workpiece surface are perpendicular, resulting in large errors in the measurement results of some measurement positions. As a result, the accuracy is low due to inaccurate measurement positions. 3) In addition to increasing employee salary costs, increasing staff turnover also leads to higher training costs. The lack of experience of new employees after joining the company also increases the loss of false measurements, which in turn leads to high measurement costs. 4) Manual recording of measurement data is large and cannot be fully recorded. Paper storage and regular destruction make it impossible to trace back when needed, which makes it difficult to preserve measurement data for a long time and makes it impossible to trace back. 5) Although some protective measures are taken during manual measurement, uneven force often exists, and sweat contact with the workpiece to be measured can cause corrosion to the workpiece. In addition, there is still much room for improvement in many aspects such as automation of detection actions, automation of data analysis, and improving detection efficiency and accuracy.

[0005] Therefore, it is necessary to provide a new automatic thickness measurement method to realize the automation of engine blade thickness detection and solve the above problems. Summary of the Invention

[0006] The present invention aims to provide an automatic thickness measurement method based on point cloud matching to solve the technical problems in the prior art, such as poor consistency of manual measurement, low accuracy and high measurement cost due to inaccurate measurement position, manual recording of measurement data, inability to record completely the large amount of information, untraceability of data, inability to realize automated measurement process, and multiple steps requiring human intervention. The technical problems to be solved by the present invention are achieved through the following technical solutions.

[0007] A first aspect of the present invention proposes an automatic thickness measurement method based on point cloud matching, comprising: establishing a point cloud model file of a standard test piece according to the external features of the standard test piece, the standard test piece being used to characterize a standard part of an engine blade, the point cloud model file comprising point cloud data of the standard test piece; automatically clamping a test piece, scanning the point cloud data of the test piece, and performing point cloud matching based on the point cloud model file to calibrate the position information of a target detection point of the test piece; performing a thickness detection operation on the test piece after the position information is calibrated; collecting the wall thickness of the test piece at the target detection point in a posture in which the detection direction of an ultrasonic thickness gauge is parallel to the normal direction of a tangent plane of the target detection point; judging the collected wall thickness of the target detection point as qualified, and displaying the detection data related to the target detection point in real time.

[0008] According to an alternative embodiment, the method for automatically clamping a test piece, scanning the point cloud data of the test piece, and performing point cloud matching based on the point cloud model file to calibrate the position information of the target detection points of the test piece includes: performing coincidence matching between the point cloud data of the test piece at the current position and the point cloud model file to obtain the point cloud matching degrees of the respective target detection points of the test piece, where the point cloud data includes the three-dimensional position coordinates of the target detection points in a specified area of the test piece; based on the obtained point cloud matching degrees, determining the pose transformation relationship for transforming the target detection points of the test piece from the current position to the target position, where the pose transformation relationship represents the pose transformation relationship from the camera coordinate system to the robot arm base coordinate system.

[0009] According to an alternative embodiment, the reprojection error between the point cloud data of the test piece at the current position and the point cloud data of the standard test piece in the point cloud model file is calculated through the following expression to obtain the point cloud matching degrees of the respective target detection points of the test piece:

[0010]

[0011] where N represents the number of points in the target model corresponding to the point cloud data of the standard test piece; p i represents the i-th point in the target model; represents the position of the corresponding point of the test piece in the scene point cloud in the reconstructed point cloud, i.e., the current position; w i represents the weight coefficient of the i-th point.

[0012] According to an alternative embodiment, project the original three-dimensional position coordinates of the target detection points of the test piece onto the two-dimensional image plane; through grayscale conversion, morphological processing, and feature selection processing on the two-dimensional image plane, extract the blade body area of the test piece, and then convert it to the three-dimensional space to obtain the point cloud data of the blade body area. By calculating the three-dimensional coordinate deviation between the point cloud data of the blade body area and the point cloud data of the standard test piece, the point cloud matching degrees of the respective target detection points of the test piece are obtained.

[0013] According to an alternative embodiment, according to the determined pose transformation relationship, adjust the pose of the test piece so that the position of the test piece is consistent with the position of the standard test piece, and obtain the test piece with calibrated position information.

[0014] According to an alternative embodiment, identify the blade features of the test piece to obtain the pose information of the blade body area in the blade features in the camera coordinate system. According to the obtained pose information and the pose information of the blade body area in the blade features, calibrate the clamping position of the test piece.

[0015] According to an alternative embodiment, the pose information of the test piece to be tested in the camera coordinate system is converted into the pose information of the test piece to be tested in the base coordinate system of the robotic arm, and based on the pose information of the test piece to be tested in the base coordinate system of the robotic arm, the deviation information between the current position and the target position of each target detection point of the test piece to be tested is calculated.

[0016] According to an alternative embodiment, based on the ultrasonic waveform analysis algorithm, the ultrasonic detection device analyzes and acquires the digital ultrasonic waveform for waveform discrimination to obtain the blade wall thickness of the test piece to be tested at the target detection point in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0017] A second aspect of the present invention provides an automatic thickness measurement system based on point cloud matching, which is used to execute the automatic thickness measurement method described in the first aspect of the present invention. The automatic thickness measurement system includes: a clamping device for clamping the test piece to be tested and driving the test piece to be tested to perform a thickness detection action. Before performing the thickness detection action, point cloud matching is performed based on a pre-established point cloud model file to calibrate the position information of the target detection point of the test piece to be tested; a machine vision device located above the detection position for determining the position information of the target detection point of the test piece to be tested; a positioning device for determining the normal direction of the tangent plane passing through the target detection point; an ultrasonic detection device including an ultrasonic thickness gauge for collecting the wall thickness of the test piece to be tested at the target detection point in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point; a control device electrically connected to the clamping device and the ultrasonic detection device, the control device being used to control the automatic clamping and automatic thickness detection of the test piece to be tested and to process the detection data information of the test piece to be tested; and a display device for displaying the detection process, detection data information, and detection result data information of the test piece to be tested.

[0018] According to an alternative embodiment, the automatic thickness measurement system further includes: a storage and feeding device provided on one side of the working platform; the storage and feeding device includes a station device and a transfer and storage area; a central database for collecting and storing the detection data information of the test piece to be tested fed back by the ultrasonic detection device and the control device, the detection data information including the measurement time, the pose information of the test piece to be tested in the base coordinate system of the robotic arm, and the blade wall thickness value of the target detection point; and the central database for processing the detection data information to generate a data report.

[0019] The embodiments of the present invention include the following advantages:

[0020] Compared with the prior art, the automatic thickness measurement method of the present invention obtains the point cloud model file of the standard test piece by establishing the point cloud 3D model of the standard test piece, and performs point cloud matching between the established point cloud model file of the standard test piece and the point cloud data of the test piece to be measured, so as to calibrate the position information of the target detection point of the test piece to be measured, and can obtain more accurate position information of the target detection point of the test piece to be measured, and can more accurately calibrate the position information of the target detection point of the test piece to be measured. By controlling the clamping device to clamp the test piece to be measured on the automatic production line and transporting it to the detection position for detection, the whole process of detection operation does not require direct human participation to achieve unattended operation, can provide an automated and high-load measurement process, can realize the automation of detection actions in the form of pipeline measurement, can ensure measurement consistency, and can improve detection efficiency and measurement accuracy.

[0021] The automatic thickness measurement system of the present invention establishes an electric control system mainly based on machine vision. The control device controls the clamping device to clamp the test piece to be measured on the automatic production line and transports it to the detection position for detection. The whole process of detection operation does not require direct human participation to achieve unattended operation, can provide an automated and high-load measurement process, can realize the automation of detection actions in the form of pipeline measurement, can ensure measurement consistency, can improve detection efficiency and measurement accuracy, can save human resources, can effectively reduce labor costs, and can also effectively avoid various corrosion and pollution.

[0022] In addition, by measuring the blade wall thickness of the test piece to be measured at the target detection point with an ultrasonic detection device, accurate and efficient actions can be realized, and the error of one thousand actions is guaranteed to be at the micron level, which can effectively ensure the measurement accuracy.

[0023] In addition, by making the ultrasonic thickness measurement probe of the ultrasonic detection device form a vertical angle with the target detection point, the blade wall thickness of the test piece to be measured can be obtained in the pose where the detection direction of the ultrasonic detection device is parallel to the normal direction of the tangent plane of the target detection point, thereby improving the thickness measurement accuracy of the target detection point.

[0024] In addition, by using the laser displacement sensor of the positioning device to determine (find) the normal direction of the tangent plane of the target detection point, non-contact measurement of the blade surface of the test piece to be measured at the micron level can be performed, which has the advantages of high precision and good repeatability, and can effectively avoid the technical problems such as low efficiency and insufficient accuracy caused by manually holding the probe (i.e., the thickness measurement probe) to find the normal direction of the tangent plane of the target detection point during ultrasonic thickness measurement, and even the damage of the thickness measurement probe caused by contact and collision with the test piece to be measured.

[0025] In addition, by establishing a central database, the detection data information of the test piece fed back by the ultrasonic detection device and the control device can be automatically collected and saved to provide data support for the subsequent provision of detection reports and reports. The software system based on the database can realize the automation of the analysis of the test result report, and can realize efficient measurement data storage, data traceability, management, and reporting. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of an example of an automatic thickness measurement method based on point cloud matching of the present invention;

[0027] Figure 2 It is an application Figure 1 A schematic structural diagram of an example of an automatic thickness measurement system of an automatic thickness measurement method;

[0028] Figure 3 This is a flow chart of an example of preprocessing the obtained point cloud data of the piece to be tested to obtain 3D point cloud data in the automatic thickness measurement method based on point cloud matching of the present invention;

[0029] Figure 4 1 is a schematic structural diagram of an example of an automatic thickness measurement system based on point cloud matching according to the present invention;

[0030] Figure 5 yes Figure 4 A schematic diagram of the partial structure of the automatic thickness measurement system from another angle;

[0031] Figure 6 yes Figure 4 A schematic diagram of the partial structure of the automatic thickness measurement system from another angle;

[0032] Figure 7 yes Figure 4 Schematic diagram of the three-dimensional structure of the end effector and the piece to be tested in the clamping device of the automatic thickness measurement system;

[0033] Figure 8 yes Figure 4 A schematic diagram of the three-dimensional structure of a work station in a storage and feeding device of an automatic thickness measurement system;

[0034] Figure 9 is a schematic diagram of an example of position calibration using a machine vision device of the automatic thickness measurement system of the present invention;

[0035] Figure 10 It is a schematic diagram of using the automatic thickness measurement system of the present invention to measure the posture information of the test piece in the camera coordinate system at the target detection point. DETAILED DESCRIPTION

[0036] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0037] Referring to Figure 1 、 Figure 2 and Figure 3 , a first aspect of the present invention provides an automatic thickness measurement method based on point cloud matching.

[0038] Figure 1 is a schematic flow chart of an example of the automatic thickness measurement method based on point cloud matching of the present invention.

[0039] As Figure 1 shown, a first aspect of the present disclosure provides an automatic thickness measurement method based on point cloud matching, and the automatic thickness measurement method includes the following steps.

[0040] Step S101, establish a point cloud model file of the standard test piece according to the shape characteristics of the standard test piece, where the standard test piece is used to represent the standard part of the engine blade, and the point cloud model file includes the point cloud data of the standard test piece.

[0041] Step S102, automatically clamp the test piece to be tested, scan the point cloud data of the test piece to be tested, and perform point cloud matching based on the point cloud model file to calibrate the position information of the target detection point of the test piece to be tested.

[0042] Step S103, perform a thickness detection action on the test piece to be tested after calibrating the position information.

[0043] Step S104, collect the wall thickness of the test piece to be tested at the target detection point in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0044] Step S105, perform a pass / fail judgment on the collected wall thickness of the target detection point, and display the detection data related to the target detection point in real time.

[0045] Figure 2 is an example of the structure diagram of an automatic thickness measurement system applying Figure 1 of the automatic thickness measurement method.

[0046] As Figure 2 shown, the automatic thickness measurement system includes a clamping device 100, a machine vision device 200, a positioning device, an ultrasonic detection device 400, a control device 500, a display device 600, and a storage and feeding device 700. The station device 710 of the storage and feeding device 700 is placed between the clamping device 100 and the ultrasonic detection device 400. The clamping device 100, the ultrasonic detection device 400, and the storage and feeding device 700 are all placed on the operation platform 2000.

[0047] It should be noted that the automatic measurement method of the present invention is particularly applicable to Figure 2 the automatic thickness measurement system, but is not limited thereto, and is also applicable to other thickness measurement systems. The above is only described as an optional example and should not be construed as a limitation to the present invention.

[0048] Next, reference will be made to Figure 2 the application example of

[0049] First, in step S101, a point cloud model file of the standard test piece is established according to the external shape characteristics of the standard test piece. The standard test piece is used to represent the standard part of the engine blade, and the point cloud model file includes the point cloud data of the standard test piece.

[0050] In a specific embodiment, before establishing the point cloud model file of the standard test piece (i.e., the standard part of the engine blade, also known as the standard part), one or more target detection points are manually marked on the upper surface (also called the front surface) and the lower surface (also called the back surface) of the standard test piece, or one or more target detection points in the specified area are marked.

[0051] It should be noted that in the present invention, the upper surface of the object in the drawing is the front surface, and the lower surface of the object in the drawing is the back surface. This is only for a clearer description of the detection process and should not be construed as a limitation to the present invention.

[0052] For example, using a laser triangulation scanner, a camera or other devices with scanning functions, the standard test piece clamped by the robotic arm of the clamping device 100 (i.e., the end effector 111 of the six-axis robot 110) is scanned, and the blade information of the standard test piece is collected at a constant speed through the laser triangulation scanner to obtain the point cloud data of all points on the upper surface (i.e., the front surface) of the standard test piece. Then, the robotic arm flips the standard test piece by 180°, and passes it through the laser triangulation scanner again at a constant speed to obtain the point cloud data of all points on the lower surface (i.e., the back surface) of the standard test piece.

[0053] Specifically, through point cloud filtering, point cloud segmentation, and point cloud registration algorithms, the two sets of point cloud data of the standard test piece before and after are processed to obtain the 3D point cloud model of the standard test piece. Among them, for example, through point cloud segmentation, the point cloud data of the standard blade is segmented into the blade body part (i.e., the body area) and the blade tenon part, and the blade tenon part is discarded, and the body area is used as the 3D point cloud model of the standard test piece (i.e., the standard part).

[0054] Specifically, the point cloud model file includes the point cloud data of the upper surface and the lower surface of the standard test piece, the initial position when collecting the point cloud data, the number and position information of the pre-marked target detection points, the information of the specified area containing the target detection points, and the blade number information corresponding to the specified area.

[0055] It should be noted that the robotic arm clamps the standard test piece to the ultrasonic thickness gauge (i.e., the ultrasonic detection device 400), and manually adjusts the position of the robotic arm so that the pre-marked target measurement point on the standard test piece is directly above the ultrasonic thickness gauge (specifically, the probe or ultrasonic probe). When the waveform and numerical value of the ultrasonic thickness gauge are stable, the pose information of the robotic arm is recorded. Each target measurement point has a unique corresponding robotic arm pose information. In other embodiments, the position of the robotic arm can also be manually adjusted to ensure that the target measurement point is directly above the ultrasonic thickness gauge, and the unique robotic arm pose information of each measurement point is recorded.

[0056] By establishing a 3D point cloud model of the standard test piece, a point cloud model file of the standard test piece is obtained for calibrating the position information of the target detection points of the test piece to be tested, and more accurate position information of the target detection points of the test piece to be tested can be obtained, and the position information of the target detection points of the test piece to be tested can be calibrated more accurately.

[0057] In addition, the blade number of the test piece to be tested is identified by constructing a blade recognition model (i.e., a pre-trained OCR recognition model). Specifically, based on the blade depth map scanned by the laser triangulation scanner, operations such as threshold segmentation and morphological processing are performed on the blade depth map to segment the OCR character area (the image is binarized and grayscale processed, and then segmented according to an appropriate pixel threshold to retain the OCR character area), and the image of the OCR character area is transmitted to the pre-trained OCR recognition model for recognition to obtain the blade number information. Among them, the training data includes images marked with specified areas (the specified area includes one or more target detection points).

[0058] For example, when the image of the standard test piece or the test piece to be tested is input into the blade recognition model (i.e., the pre-trained OCR recognition model), the blade number of the standard test piece or the test piece to be tested is output for subsequent use in blade thickness measurement or for identifying blade characteristics.

[0059] It should be noted that since each test blade has a serial number, the respective serial number information needs to be recorded in the database. The blade serial number identification process and the thickness measurement process are processed in parallel in the program without interference. Specifically, when scanning the point cloud data, for example, while a laser triangulation scanner scans, it outputs depth map data for OCR character recognition to extract the blade serial number (or blade serial number information) for storage in the database. In addition, the above is only described as an optional example and should not be construed as a limitation to the present invention.

[0060] Next, in step S102, the test piece to be tested is automatically clamped, the point cloud data of the test piece to be tested is scanned, and point cloud matching is performed based on the point cloud model file to calibrate the position information of the target detection point of the test piece to be tested.

[0061] In a preferred embodiment, when automatically measuring the thickness of the test piece to be tested, the test piece to be tested is automatically clamped, the point cloud data of the test piece to be tested is scanned, and the point cloud data of the test piece to be tested is obtained. This point data is, for example, in the tif file format.

[0062] Specifically, the point cloud data of the test piece to be tested obtained is preprocessed to obtain the 3D point cloud data of the test piece to be tested. The following steps are performed. For details, see Figure 3 .

[0063] Step S310, perform a projection transformation on the obtained point cloud data (i.e., 3D point cloud data or three-dimensional data). First, a 2D plane (i.e., two-dimensional plane) composed of, for example, the x-axis and y-axis under a laser triangulation scanner is determined (or selected) as the projection plane, and the 3D point cloud data is mapped onto this 2D plane. The 3D point cloud data is orthogonally projected along the line of sight direction to form a first image on the 2D plane.

[0064] Step S320, perform feature extraction on the first image. After obtaining the first image, for example, the method of Blob analysis is used to segment the blade body region. First, the first image is grayscale, and most of the background regions are removed by selecting the grayscale threshold range. Then, the connected regions (Blobs) in the image are distinguished by, for example, the connected component analysis method. Among them, a connected region refers to a pixel region in the image composed of adjacent pixels with the same grayscale value or within a certain range. Next, the blade body region is further extracted by feature selection (i.e., extracting the specified region or the region containing the target detection point in the blade body part of the test piece to be tested). By performing threshold selection on the area feature, shape feature, and direction feature of each Blob region, the ROI region, that is, the complete blade point cloud region, is finally obtained. Thus, a second image is obtained.

[0065] Step S330, perform 3D region selection. After obtaining the 2D image (i.e., the second image) of the ROI region, it is necessary to map it back to the 3D space to obtain the corresponding 3D region of the second image. For example, by using the back-projection method, the 2D coordinates of the ROI region are converted into 3D coordinates to obtain the three-dimensional position coordinates of the target detection points of the test piece to be tested.

[0066] In this embodiment, the point cloud data of the test piece at the current position (specifically, the point cloud data of the target detection points) is overlapped and matched with the point cloud model file (the point cloud data of the target detection points of the standard test piece) to obtain the point cloud matching degree of each target detection point of the test piece to be tested. The point cloud data includes the three-dimensional position coordinates representing the target detection points in the specified region of the test piece to be tested.

[0067] It should be noted that the deviation information between the current position and the target position needs to calculate the point cloud matching degree of at least two target detection points of the test piece to be tested. For example, calculate the point cloud matching degree of two target detection points. Specifically, it is achieved by calculating the reprojection error of the two target detection points. Among them, the reprojection error refers to projecting the 3D points of the target detection points of the test piece to be tested from the target model onto the 2D image plane; in the two-dimensional image plane, through grayscale processing, morphological processing, and feature selection, the blade region of the test piece to be tested is extracted, and then converted to the 3D space to obtain the point cloud data of the blade region. By calculating the three-dimensional coordinate deviation between the point cloud data of the blade region and the point cloud data of the standard test piece, the point cloud matching degree of each target detection point of the test piece to be tested is obtained.

[0068] Specifically, the reprojection error between the point cloud data of the test piece to be tested at the current position and the point cloud data of the standard test piece in the point cloud model file is calculated through the following expression to obtain the point cloud matching degree of each target detection point of the test piece to be tested:

[0069]

[0070] Among them, E represents the point cloud matching degree of each target detection point of the test piece to be tested; N represents the number of points in the target model corresponding to the point cloud data of the standard test piece; p i represents the i-th point in the target model; represents the position of the corresponding point of the test piece to be tested in the scene point cloud in the reconstructed point cloud, that is, the current position; w i represents the weight coefficient of the i-th point.

[0071] It should be noted that for the weight coefficient (w i ), it can be adjusted according to factors such as feature matching results and distance metrics. The above is only for illustrative purposes as an optional example and should not be construed as a limitation to the present invention.

[0072] In a specific embodiment, first, starting from the original three-dimensional position coordinates (x, y, z) of the target detection points on the test piece to be tested, these coordinates are projected onto a two-dimensional image plane composed of the (x, y) axes by an orthographic projection method from top to bottom. Then, it is processed by morphological methods in two-dimensional image processing. Morphology is a processing method based on image shape, used to analyze and process the structure of two-dimensional images. It mainly includes operations such as dilation, erosion, opening, and closing. In this example, morphological processing is used to extract the blade body region, making the contour of the blade body region clearer by eliminating noise and connecting broken regions. Next, it is further processed by feature selection methods in two-dimensional image processing. Feature selection is to select some features that contribute to the target task from the original features to reduce the feature dimension, reduce the amount of calculation, and improve the generalization ability of the model. Here, feature selection is mainly used to extract the key features of the blade body region for better matching with the standard test piece. Then, it is converted from two-dimensional image to three-dimensional space. After completing morphological processing and feature selection, the blade body region of the test piece to be tested has been extracted on the two-dimensional image plane. Next, the extracted blade body region needs to be converted back from the two-dimensional image to three-dimensional space. This step requires an inverse operation according to the previous orthographic projection method to remap the two-dimensional coordinates into three-dimensional space, generating the point cloud data of the blade body region. Finally, calculate the point cloud matching degree: After obtaining the three-dimensional point cloud data of the blade body region of the test piece to be tested, the three-dimensional coordinate deviation between the three-dimensional point cloud data of the blade body region of the test piece to be tested and the point cloud data of the standard test piece can be calculated. Through this calculation, the point cloud matching degree of each target detection point of the test piece to be tested can be obtained. A higher matching degree means that the shape and structure of the test piece to be tested are closer to those of the standard test piece, thus it can be used to evaluate the test piece to be tested.

[0073] In another specific embodiment, first, some feature points (e.g., SURF feature points, i.e., target detection points) are extracted from the point cloud 3D model (i.e., the target model) of the point cloud model file of the standard test piece, and descriptors (e.g., SURF descriptors) of the point cloud 3D model are calculated based on these feature points. Then, similar feature points to the point cloud 3D model (i.e., the target model) of the standard test piece are searched in the point cloud data of the test piece to be tested in the detection scene, and descriptor matching is performed on these feature points. For example, the KD-Tree algorithm is used to accelerate the matching process.

[0074] According to the matching results of the feature points, the initial pose of the point cloud 3D model of the test piece to be tested to the scene point cloud is calculated. For example, the MRSAC algorithm is used for pose estimation. The point cloud data on the point cloud 3D model of the test piece to be tested is projected onto the 2D image plane using the calculated pose, and then back-projected into 3D space to obtain the position (i.e., the current position) of the corresponding points of the test piece to be tested in the scene point cloud in the reconstructed point cloud.

[0075] Calculate the reprojection error between the reconstructed point cloud of the test piece to be tested and the point cloud data (i.e., the target detection points) of the point cloud 3D model of the standard test piece. Specifically for each point (i.e., each target detection point of the test piece to be tested), for example, use the Euclidean distance or other distance metrics to calculate the distance between the position of the corresponding point of each target detection point of the test piece to be tested in the scene point cloud and its position in the reconstructed point cloud.

[0076] Then, perform a weighted average of the distances of all points to obtain an overall reprojection error value, which is used as the score for the matching degree of the two point clouds. The influence of different distance metrics and feature matching results can be balanced by adjusting the weight coefficients.

[0077] Based on the obtained point cloud matching degree, determine the pose transformation relationship for transforming the target detection points of the test piece to be tested from the current position to the target position, where the pose transformation relationship represents the pose transformation relationship from the camera coordinate system to the robotic arm base coordinate system.

[0078] Then, according to the determined pose transformation relationship, adjust the pose of the test piece to be tested so that the position of the test piece to be tested is consistent with the position of the standard test piece, and obtain the test piece to be tested after calibrating the position information.

[0079] Performing point cloud matching based on the established point cloud model file of the standard test piece and the point cloud data of the test piece to be tested can more accurately calibrate the position information of the target detection points of the test piece to be tested.

[0080] In an alternative embodiment, by identifying the blade features of the test piece to be tested, obtain the pose information of the blade body region in the camera coordinate system among the blade features, and calibrate the clamping position of the test piece to be tested according to the obtained pose information and the pose information of the blade body region in the blade features.

[0081] Specifically, convert the pose information of the test piece to be tested in the camera coordinate system into the pose information of the test piece to be tested in the robotic arm base coordinate system, and calculate the deviation information between the current position and the target position of each target detection point of the test piece to be tested according to the pose information of the test piece to be tested in the robotic arm base coordinate system, so as to calibrate the position information of the target detection points of the test piece to be tested.

[0082] For example, by obtaining the pose information (position and orientation information) of the robotic arm (i.e., the end effector 111) of the clamping device 100 in real time, since the robotic arm (i.e., the end effector 111) clamps the test piece to be tested and the position relative to the test piece to be tested remains fixed, calibrating the pose of the robotic arm can obtain the pose information of the test piece to be tested relative to the robotic arm base coordinate system. This pose information includes the position and orientation information of the object (i.e., the test piece to be tested) in three-dimensional space. For example, it is represented by a 4x4 transformation matrix, and this transformation matrix (or pose matrix) can be expressed as:

[0083]

[0084] Among them, γ 11 , γ 12 , γ 13 , γ 21 , γ 22 , γ 23 , γ 31 , γ 32 , γ 33 represents the orientation information of the object (test piece to be tested), that is, 9 elements of the transformation matrix; t x , t y , t z represents the position information of the object, that is, the three-dimensional coordinates of the object in the reference coordinate system; the 0 in the last row represents the homogeneous coordinates of the Euclidean coordinate space, and 1 represents a unit Euclidean length. In the present invention, the transformation matrix (or pose matrix) is used to describe the position and orientation information of the test piece to be tested in three-dimensional space.

[0085] Specifically, according to the above transformation matrix (or pose matrix), assist in determining the pose transformation relationship for transforming the target detection point of the test piece to be tested from the current position to the target position.

[0086] It should be noted that the pose information of the test piece to be tested relative to the robotic arm base coordinate system, that is, the pose information in the robotic arm base coordinate system, includes the position information and orientation information of the end effector 111 (i.e., the robotic arm) in the three-dimensional space of the robotic arm base coordinate system. The Euler angle pose information includes homogeneous coordinates in the Euclidean coordinate space, a unit Euclidean length, etc. The transformation matrix (or pose matrix) can be represented by a multi-dimensional matrix, for example, and is used to describe the position information and orientation information of the test piece to be tested in three-dimensional space.

[0087] Next, in step S103, perform a thickness detection operation on the test piece to be tested after calibrating the position information.

[0088] Specifically, automatically perform a thickness detection operation on the test piece to be tested after calibrating the position information, for example, using Figure 2In the clamping device 100, the end effector 111 of the six-axis robot 110 clamps the test piece 900 to be tested and drives the test piece 900 to perform a thickness detection action.

[0089] The control device controls the end effector 111 of the clamping device 100 to clamp the test piece to be tested on the automatic production line and transport it to the detection position for detection.

[0090] In one embodiment, it further includes determining whether the clamping device 100 is connected to the control device 500. When it is determined that the clamping device 100 is connected to the control device 500, "connected" is displayed in the working state of the clamping device 100.

[0091] For example, the main interface of the control device includes the following function menus: Data, Settings, Start, Stop, About, Log, Debug, etc. For example, clicking on "Data" will enter the historical test data interface. For example, clicking on "Settings" will enter the settings interface where the calibration position and test process can be defined. For example, a user with administrator privileges can perform the above-defined operations. For example, clicking on "Start", the six-axis robot will measure the data of the test piece to be tested according to the defined detection process. Among them, the measured data will be displayed on the main interface and also on the data function page. For example, clicking on "Stop", the six-axis robot will stop running and stop at a certain position during operation. For example, clicking on "About" will display information about the designer and design company. Clicking on "Log": will enter the log information of historical operations. Clicking on "Debug": will enter the single-step debugging interface.

[0092] The control device controls the end effector 111 of the clamping device 100 to clamp the test piece to be tested on the automatic production line and transport it to the detection position for detection. The entire process of this detection operation does not require direct human participation to achieve unattended operation, providing an automated and high-load measurement process, and enabling the automation of detection actions in a production line measurement manner.

[0093] Next, in step S104, the wall thickness of the test piece to be tested at the target detection point is collected in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0094] In this example, the normal direction of the tangent plane passing through the target detection point is determined by the positioning device. For example, the target detection point is the target detection point on the test piece to be tested determined by the machine vision device 200.

[0095] Optionally, the target detection point is the target detection point calibrated in step S102, specifically the target detection point on the test piece 900 determined by the machine vision device 200.

[0096] Specifically, the laser displacement sensor of the positioning device detects the test piece 900 to obtain the pose information of the test piece 900 in the camera coordinate system, so as to determine the normal direction of the tangent plane passing through the target detection point.

[0097] By determining (finding) the normal direction of the tangent plane of the target detection point through the laser displacement sensor, non-contact measurement of the blade surface of the test piece can be carried out at the micron level, with the advantages of high precision and good repeatability. It can effectively avoid the technical problems such as low efficiency and insufficient precision caused by manually holding the probe (i.e., the thickness measurement probe) to find the normal direction of the tangent plane of the target detection point during ultrasonic thickness measurement, and even the damage of the thickness measurement probe caused by contact and collision with the test piece.

[0098] Next, in step S105, the wall thickness of the target detection point collected is judged to be qualified, and the detection data related to the target detection point is displayed in real time.

[0099] Based on the determined position information of the target detection point, the wall thickness of the test piece at the target detection point is collected in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0100] Specifically, the ultrasonic thickness gauge of the ultrasonic detection device 400 measures the blade wall thickness of the test piece at the target detection point, specifically, the blade wall thickness of the test piece at the target detection point is collected in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0101] In an embodiment, when the test piece 900 is located at the detection position, the spatial coordinate position of the target detection point on the test piece 900 in the base coordinate system of the robotic arm is detected, so that the normal direction of the tangent plane passing through the target detection point is parallel to the detection direction of the ultrasonic detection device 400 (specifically, the thickness measurement probe of the ultrasonic thickness gauge), so that the detection direction of the ultrasonic thickness gauge forms a perpendicular angle with the tangent plane of the target detection point.

[0102] Based on the ultrasonic waveform analysis algorithm, the ultrasonic detection device analyzes and obtains the digital ultrasonic waveform for waveform discrimination, so as to obtain the blade wall thickness of the test piece at the target detection point in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0103] Specifically, the ultrasonic detection device 400 analyzes and acquires digital ultrasonic waveforms for waveform discrimination. Specifically, a discrimination algorithm is used to discriminate the digital ultrasonic waveforms, filtering out chaotic waves or incorrect waveforms with messy waveforms, and leaving correct waveforms such as those shaped like iron towers. When the detection direction of the thickness measurement probe of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point, the ultrasonic thickness gauge can receive echoes to measure the blade wall thickness of the target detection point on the test piece to be tested. By the above method, it is determined whether the detection direction of the thickness measurement probe of the ultrasonic thickness gauge is parallel (or consistent) with the normal direction of the tangent plane of the target detection point, so as to obtain the blade wall thickness of the test piece to be tested at the target detection point in the pose where the detection direction of the ultrasonic detection device 400 (specifically, the thickness measurement probe of the ultrasonic thickness gauge) is parallel to the normal direction of the tangent plane of the target detection point.

[0104] Next, the wall thickness of the collected target detection point is judged to be qualified, and the detection data related to the target detection point is displayed in real time.

[0105] Specifically, the wall thickness of the collected target detection point is processed. For example, according to the blade thickness numerical information fed back by the ultrasonic detection device to the control device 500 during thickness measurement, it is judged whether the current test piece to be tested has been detected and whether there are defects (such as defects of too thin blade wall thickness or eccentric blade cavity). For another example, the wall thickness and related detection data are sent to the display device for real-time display on the display screen.

[0106] Optionally, the control device 500 gives an audible and visual alarm for unqualified test pieces to be tested and prompts the operator that there are unqualified test pieces to be tested.

[0107] In this example, the display device is used to display the detection process, detection data information, and detection result data information of the test piece to be tested (such as whether the thickness of the target detection point of the test piece to be tested is qualified).

[0108] Compared with the prior art, the automatic thickness measurement method of the present invention, by establishing a point cloud 3D model of a standard test piece, obtaining a point cloud model file of the standard test piece, and performing point cloud matching between the established point cloud model file of the standard test piece and the point cloud data of the test piece to be tested, is used to calibrate the position information of the target detection point of the test piece to be tested, can obtain more accurate position information of the target detection point of the test piece to be tested, and can more accurately calibrate the position information of the target detection point of the test piece to be tested. By controlling the clamping device to clamp the test piece to be tested on the automatic production line and transporting it to the detection position for detection, the entire detection operation process does not require direct human participation to achieve unattended operation, can provide an automated and high-load measurement process, can realize the automation of detection actions in a production line measurement manner, can ensure measurement consistency, and can improve detection efficiency and measurement accuracy.

[0109] The following are system embodiments of the present invention. The method of the first aspect of the present invention is particularly applicable to the automatic thickness measurement system of the present invention. For details not disclosed in the system embodiments of the present invention, please refer to the method embodiments of the present invention.

[0110] Refer to Figure 2 、 Figures 4 to 10 , the automatic thickness measurement system 1000 of the present invention includes a clamping device 100, a machine vision device 200, a positioning device, an ultrasonic detection device 400, a control device 500, and a display device 600.

[0111] In Figure 4 's example, the automatic thickness measurement system 1000 further includes a storage and feeding device 700. The storage and feeding device 700 is arranged on one side of the operation platform 2000. The station device 710 of the storage and feeding device 700 is placed between the clamping device 100 and the ultrasonic detection device 400. The clamping device 100, the ultrasonic detection device 400, and the storage and feeding device 700 are all placed on the operation platform 2000.

[0112] As Figure 2 、 Figure 4 and Figure 8 shown, the storage and feeding device 700 includes a station device 710 and a transfer and storage area 720. Among them, the station device 710 is used to place the test piece 900 to be tested. The station device 710 includes a plurality of stations 711. The transfer and storage area 720 is used to store the station device 710. Specifically, when the test piece 900 in the station device 710 on the operation platform 2000 is measured, the unworked station device 710 in the transfer and storage area 720 is replaced with the station device 710 on the operation platform 2000.

[0113] The automatic thickness measurement system 1000 further includes an outer cover 3000. The outer cover 3000 is an outer cover composed of a safety fence. Its main function is to isolate the working area from the outside world to prevent problems that affect measurement caused by external factors.

[0114] In Figure 4 's example, the clamping device 100 includes a six-axis robot 110. The six-axis robot 110 is provided with an end effector 111. The end effector 111 is used to clamp the test piece 900 to be tested and drive the test piece 900 to perform thickness detection actions.

[0115] In this example, the six-axis robot 110 mimics the basic structure from the waist to the arm of a human. The six-axis robot includes the base (i.e., the fixed support for the bottom and waist) structure of the six-axis robot and the waist joint rotation device, the large arm (i.e., the large arm support frame) structure and the large arm joint rotation device, the small arm (i.e., the small arm support frame) structure and the small arm joint rotation device, the wrist (i.e., the wrist support frame) structure and the wrist joint rotation device, and the end effector (i.e., the gripper part) 111. Among them, the end effector 111, as the gripper part of the robot, has a clamping function and is fixed on the flange 112 of the six-axis robot 110. For details, please refer to Figure 4 and Figure 5 .

[0116] It should be noted that in other examples, but not limited to, the six-axis robot can also be other types of robots with a manipulator and a clamping function. The above is only described as an optional example and should not be construed as a limitation to the present invention.

[0117] Specifically, the end effector 111 is used to, for example, pick up the test piece 900 to be tested from the work station 710 and can firmly hold the test piece 900 to be tested, so that the six-axis robot 110 drives the end effector 111 holding the test piece 900 to be tested to perform detection actions. The detection actions include, for example, driving the end effector 111 holding the test piece 900 to be tested to move to move to the detection position, move to the camera scanning area Q1, move to the ultrasonic thickness measurement area Q2, etc. For details, please refer to Figure 5 and Figure 6 .

[0118] In an embodiment, the test piece 900 to be tested is, for example, an engine blade of a specified model. Twelve test pieces to be tested are placed in the work station 710. These twelve test pieces are placed in the work station 710 in the same specified posture (e.g., the same orientation) so that the posture and orientation of the test piece 900 in the work station 710 are kept consistent. The spacing between two adjacent test pieces is the same. In other words, these twelve test pieces are placed at equal intervals to facilitate the six-axis robot 110 to automatically pick up (or grab) the test pieces to be tested from the work station 710 in sequence each time.

[0119] It should be noted that there are no special restrictions on the test piece to be tested and the work station. The test piece to be tested can also be engine blades of other models. The work station can also place ten, eleven or more test pieces, etc. The above is only described as an optional example and should not be construed as a limitation to the present invention.

[0120] In an alternative embodiment, the automatic thickness measurement system 1000 has a periodic self-calibration function. After a certain period of thickness measurement operations by the six-axis robot 110, the six-axis robot 110 will enter an automatic calibration program to calibrate by clamping a test piece with a standard thickness (the thickness can be adjusted), and correct the measurement result drift of the automatic thickness gauge. The automatic calibration function can be carried out in two ways: manual and automatic.

[0121] From Figure 2 and Figure 6 it can be seen that the machine vision device 200 is located above the ultrasonic detection device 400 and above the detection position. The machine vision device 200 is used to determine the position information of the target detection point of the test piece 900 to be tested.

[0122] It should be noted that in this example, the positioning device can be another device independent of the machine vision device 200. However, it can also be a component of the machine vision device 200, or the same device as the machine vision device 200. The positioning device is, for example, a laser sensor, etc.

[0123] In this example, the machine vision device 200 includes a 3D camera 210 (such as a laser triangulation scanner). The 3D camera 210 is used to identify the blade features of the test piece 900 to be tested and obtain the pose information of the blade body region in the camera coordinate system in the blade features.

[0124] Specifically, for example, OCR is used to identify the blade body region of the test piece 900 to be tested, and a 3D point cloud model of the blade body region of the test piece 900 to be tested is scanned as the blade features. The blade features include the blade body contour, the blade surface features, the blade identification (such as the blade identity code 1**3, the blade two-dimensional code, etc.), and so on.

[0125] Further, according to the pose information of the 3D camera 210 in the base coordinate system of the robotic arm (specifically including the position and orientation, for example, represented by (x, y, z, Rx, Ry, Rz, α)) and the pose information of the blade body part (blade body region) in the blade features, the clamping position of the test piece 900 to be tested is calibrated to obtain the three-dimensional coordinates of the clamping position of the test piece 900 to be tested. See Figure 9 .

[0126] It should be noted that the pose information specifically includes the position information (x, y, z) in space, the pose information in space, that is, including the position and orientation. The pose information is represented as (x, y, z, Rx, Ry, Rz, α), where Rx, Ry, and Rz represent the rotation angles of the object around the x-axis, y-axis, and z-axis, and α represents the rotation order, for example, first rotate around the z-axis, then rotate around the y-axis, and finally rotate around the x-axis.

[0127] The relative position between the 3D camera 210 and the six-axis robot 110 is known, that is, the relative position between the calibration plate and the 3D camera 210 is represented by X, and the relative position between the calibration plate and the six-axis robot 110 can be obtained to calibrate the clamping position of the test piece 900.

[0128] The relative position between the 3D camera 210 and the six-axis robot 110, i.e., X, is known. The six-axis robot 110 clamps the piece to be tested 900 and shoots it to obtain the spatial coordinate T2. The coordinate space T2 in the camera coordinate system can be converted into the spatial coordinate T1 in the base coordinate system of the six-axis robot 110 (also called the robot arm base coordinate system), i.e., T1 = X*T2, to obtain the clamping position of the piece to be tested.

[0129] Specifically, the calibration process of calibrating the clamping position of the piece to be tested 900 includes transforming the spatial three-dimensional coordinates of the target point.

[0130] First, determine the relative position between 3D camera 210 and six-axis robot 110 (specifically, the position transformation between the camera coordinate system and the robot arm base coordinate system). This is the robot's hand-eye position relationship. This relationship is represented by the symbol X and can be solved using the equation AX = XB. Here, A represents the transformation relationship between the robot's end joints for two consecutive arbitrary movements; B represents the change in the camera coordinates (i.e., the camera coordinate system) for two consecutive arbitrary movements. X is, for example, a 4x4 homogeneous transformation matrix.

[0131]

[0132] Calculate X based on the calibration plate images (i.e., images on the calibration plate) taken at different times. Figure 9 The calibration plate picture shown is shown in Figure 1. t is the translation function. By inputting the translation amount in the xyz direction, X can be obtained.

[0133] The six-axis robot 110 is photographed while clamping the piece to be tested 900 to obtain the spatial coordinate T2 of the piece to be tested 900 relative to the 3D camera 210. The coordinate T2 of the piece to be tested 900 relative to the 3D camera 210 can be converted into the base coordinate T1 of the piece to be tested 900 relative to the six-axis robot 110, that is, T1 = X*T2, to obtain the clamping position of the piece to be tested.

[0134] It should be noted that in this example, the robot arm base coordinate system is a spatial coordinate system centered on the six-axis robot; the camera coordinate system is a spatial coordinate system centered on the camera of the machine vision device. Furthermore, a tool coordinate system is included. The tool coordinate system is the reference coordinate system of the end effector 111, defined by the position and posture of the end effector 111 and used to describe its motion.

[0135] Specifically, the pose information of the test piece 900 in the camera coordinate system is converted into the pose information of the test piece 900 in the robot arm base coordinate system, and the deviation information between the current position and the target position of the test piece 900 is calculated according to the pose information of the test piece 900 in the robot arm base coordinate system.

[0136] In one embodiment, for example, a point cloud matching method is used to calculate the deviation information between the current position and the target position of the test piece 900.

[0137] It should be noted that since the method of using point cloud matching to calculate the deviation information between the current position and the target position of the test piece 900 in this embodiment is substantially the same as the content of step S102 in the first aspect of the present invention (i.e., performing point cloud matching based on the point cloud model file to calibrate the position information of the target detection points of the test piece), the description of the same part is omitted.

[0138] In another embodiment, for example, the MRSAC algorithm is used to calculate the deviation information between the current position and the target position of the test piece 900.

[0139] Next, according to the calculated deviation information, the pose information of the test piece 900 in the robot arm base coordinate system is calibrated. It is also possible to calibrate the pose information of the blade features (such as a certain area or a certain detection point on the blade body part, etc.) of the test piece 900 in the robot arm base coordinate system to obtain the position coordinates of the target detection points on the test piece 900.

[0140] In an alternative embodiment, according to the deviation information, the pose information of the target detection points on the test piece 900 in the robot arm base coordinate system is calibrated to obtain more accurate position coordinates of the target detection points.

[0141] Next, the position coordinates of the target detection points on the obtained test piece 900 are fed back to the ultrasonic detection device 400 and the control device 500.

[0142] Next, the positioning device and the ultrasonic detection device 400 of the automatic thickness measurement system 1000 of the present invention will be described.

[0143] In this example, the positioning device is used to determine the normal direction of the tangent plane passing through the target detection point, where the target detection point is the target detection point on the test piece determined by the machine vision device 200. The ultrasonic detection device 400 is used to collect the wall thickness (i.e., the blade wall thickness) of the test piece 900 at the target detection point.

[0144] Optionally, the target detection point is a calibrated target detection point, and specifically, the machine vision device 200 is used to determine the target detection point on the test piece 900.

[0145] Specifically, the positioning device includes a laser displacement sensor, which is used to detect the test piece 900 to obtain the pose information of the test piece 900 in the camera coordinate system, so as to determine the normal direction of the tangent plane passing through the target detection point.

[0146] Determining (finding) the normal direction of the tangent plane of the target detection point through the laser displacement sensor can perform micron-level non-contact measurement on the blade surface of the test piece, with the advantages of high precision and good repeatability. It can effectively avoid the technical problems such as low efficiency and insufficient precision caused by manually holding the probe (i.e., the thickness measurement probe) to find the normal direction of the tangent plane of the target detection point when measuring the thickness with ultrasound, and even the damage of the thickness measurement probe caused by contact and collision with the test piece.

[0147] Furthermore, the ultrasonic detection device 400 includes an ultrasonic thickness gauge, and the ultrasonic thickness gauge further includes a thickness measurement probe. The ultrasonic thickness gauge is used to measure the blade wall thickness of the test piece at the target detection point. Specifically, the blade wall thickness of the test piece at the target detection point is collected in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

[0148] In an embodiment, when the test piece 900 is located at the detection position, the spatial coordinate position of the target detection point on the test piece 900 in the base coordinate system of the robotic arm is detected, so that the normal direction of the tangent plane passing through the target detection point is parallel to the detection direction of the ultrasonic detection device 400 (specifically, the thickness measurement probe of the ultrasonic thickness gauge), so that the detection direction of the ultrasonic thickness gauge forms a perpendicular angle with the tangent plane of the target detection point.

[0149] By using the laser displacement sensor to measure the pose information of the test piece in the camera coordinate system at the target detection point, the normal direction of the plane where the tangent line of the target detection point on the blade surface of the test piece is located (i.e., the tangent plane) can be calculated through analytic geometry, so as to obtain the pose information of the target detection point on the test piece when the tangent plane is perpendicular to the detection direction of the thickness measurement probe of the ultrasonic thickness gauge (for example, the central axis of the thickness measurement probe). For specific reference, see Figure 10 。

[0150] Based on the ultrasonic waveform analysis algorithm, the ultrasonic detection device 400 analyzes and acquires the digital ultrasonic waveform for waveform discrimination. Specifically, the digital ultrasonic waveform is discriminated by a discrimination algorithm to filter out the chaotic waves or incorrect waveforms with messy waveforms, and leave the correct waveforms such as those shaped like iron towers. When the detection direction of the thickness measurement probe of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point, the ultrasonic thickness gauge can receive the echo to measure the blade wall thickness of the target detection point on the test piece to be tested. By the above method, it is judged whether the detection direction of the thickness measurement probe of the ultrasonic thickness gauge is parallel (or consistent) with the normal direction of the tangent plane of the target detection point, so as to obtain the blade wall thickness of the test piece at the target detection point in the pose where the detection direction of the ultrasonic detection device 400 (specifically, the thickness measurement probe of the ultrasonic thickness gauge) is parallel to the normal direction of the tangent plane of the target detection point.

[0151] By making the normal direction of the tangent plane passing through the target detection point parallel to the detection direction of the ultrasonic detection device 400 (specifically, the thickness measurement probe of the ultrasonic thickness gauge), so that the detection direction of the ultrasonic detection device 400 forms a perpendicular angle with the tangent plane of the target detection point, it is possible to make the ultrasonic thickness measurement probe of the ultrasonic detection device 400 and the target detection point form a perpendicular angle, so as to obtain the blade wall thickness of the test piece at the target detection point in the pose where the detection direction of the ultrasonic detection device 400 (specifically, the thickness measurement probe of the ultrasonic thickness gauge) is parallel to the normal direction of the tangent plane of the target detection point, and further improve the thickness measurement accuracy of the target detection point.

[0152] In order to realize the automation of the measurement action and the automation of data analysis, and at the same time improve the detection efficiency and accuracy, the automatic thickness measurement system of the present invention controls the automatic clamping and automatic thickness detection of the test piece to be tested through a control device, and is used for processing the detection data information of the test piece to be tested.

[0153] Specifically, the control device 500 is electrically connected to the clamping device 100 and the ultrasonic detection device 400. The control device 500 is used to control the end effector 111 to clamp the test piece 900 and send the test piece 900 to the detection position for thickness detection.

[0154] More specifically, the control device 500 acquires the coordinate space position of the six-axis robot 110 and the Euler angle attitude information of the current state robot in real time, for example, represented by a transformation matrix (i.e., an attitude matrix).

[0155] It should be noted that the spatial position of the coordinate system of the six-axis robot 110 is used to determine the pose information of the test piece to be tested relative to the base coordinate system of the robotic arm (i.e., the spatial coordinate information in the base coordinate system of the robotic arm). This spatial coordinate information includes the position information and orientation information of the end effector 111 (i.e., the robotic arm) in the three-dimensional space under the base coordinate system of the robotic arm. The current state robot Euler angle pose information includes homogeneous coordinates in the Euclidean coordinate space, a unit of Euclidean length, etc. The pose matrix (or attitude matrix), for example, can be represented by a multi-dimensional matrix and is used to describe the position information and orientation information of the test piece to be tested in the three-dimensional space.

[0156] The spatial position of the coordinate system of the six-axis robot 110 includes the spatial coordinate information of the end effector 111 of the six-axis robot 110 under the base coordinate system of the robotic arm. This spatial coordinate information includes the position information and orientation information of the end effector 111 in the three-dimensional space under the base coordinate system of the robotic arm. The current state robot Euler angle pose information includes homogeneous coordinates in the Euclidean coordinate space, a unit of Euclidean length, etc. The pose matrix, for example, can be represented by a multi-dimensional matrix and is used to describe the position information and orientation information of the test piece to be tested in the three-dimensional space.

[0157] According to the detection service process, plan the following detection actions and corresponding movement trajectories for the six-axis robot 110: pick up the test piece to be tested and automatically move the test piece to be tested. Plan the following movement trajectories: first pick up the test piece to be tested and transport the test piece to be tested to the detection position. Then control the six-axis robot 110 to accurately execute the detection action.

[0158] For example, according to the stage of the detection service process, control the end effector 111 of the six-axis robot 110 to pick up the test piece to be tested and accurately send the test piece to be tested to the detection position.

[0159] Furthermore, the control device 500 processes the detection data information of the test piece 900 to be tested. For example, according to the blade thickness numerical information fed back by the ultrasonic detection device to the control device 500 during thickness measurement, determine whether the current test piece to be tested has been detected and whether there are defects (such as the defect of too thin blade wall thickness or eccentric blade cavity). Another example is to send the detection data information to the display device for real-time display on the display screen.

[0160] Optionally, the control device 500 gives an audible and visual alarm for unqualified test pieces to be tested and prompts the operator that there are unqualified test pieces to be tested.

[0161] In this example, the display device is used to display the detection process, detection data information, and detection result data information of the test piece to be tested (such as whether the thickness of the target detection point of the test piece to be tested is qualified).

[0162] Specifically, the display device includes a plurality of display screens located at different positions, such as a first display screen, a second display screen, and a third display screen, enabling the detection results of the test piece 900 to be displayed online in real time (corresponding to the first display screen), displayed at a workstation (corresponding to the second display screen), and displayed as large-screen data (corresponding to the third display screen).

[0163] For example, the thickness measurement data of the current test piece and the measurement data before the current point of the same test piece 900 are simultaneously displayed on the first display screen, the second display screen, and the third display screen. For example, a test piece 900 will be measured at dozens of points, and all the thickness measurement data of the points completed before the current point of the same test piece 900 will be displayed. For example, if the 12th point is currently being measured, the thickness measurement data of the previous 11 points will also be displayed.

[0164] In a preferred embodiment, the automatic thickness measurement system 1000 further includes a central database for collecting and storing the detection data information of the test piece fed back by the ultrasonic detection device and the control device, so as to provide data support for subsequent generation of detection reports and statements. The central database includes a relational database.

[0165] Specifically, the detection data information includes measurement time, the pose information of the test piece in the base coordinate system of the robotic arm, the wall thickness value of the blade at the target detection point, and the blade identification (such as blade barcodes and two-dimensional codes, with visual automatic recognition functions). The detection data information is automatically entered into the relational database based on the blade identification, for example, to facilitate the addition, deletion, modification, and query of data.

[0166] Furthermore, the central database is also used to process the detection data information to generate data statements, such as daily, monthly, and annual detection statements, and prediction statements such as the blade quality trend.

[0167] It should be noted that the drawings are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present invention, rather than for restrictive purposes. It is easy to understand that the processes shown in the drawings do not indicate or limit the chronological order of these processes. Additionally, it is also easy to understand that these processes can be executed synchronously or asynchronously in, for example, multiple modules.

[0168] Compared with the prior art, the automatic thickness measurement system of the present invention establishes an electric control system mainly based on machine vision. The clamping device is controlled by a control device to clamp the test piece on the automatic production line and transport it to the detection position for detection. The entire detection operation process does not require direct human participation to achieve unattended operation, and can provide an automated and high-load measurement process. It can automate the detection action in the form of production line measurement, ensure measurement consistency, improve detection efficiency and measurement accuracy, save human resources, effectively reduce labor costs, and effectively avoid various corrosion and pollution.

[0169] In addition, by measuring the blade wall thickness of the test piece at the target detection point with an ultrasonic detection device, precise and efficient operations can be achieved, and the error of one thousand operations is guaranteed to be at the micron level, effectively ensuring measurement accuracy.

[0170] In addition, by forming a vertical angle between the ultrasonic thickness measurement probe of the ultrasonic detection device and the target detection point, the blade wall thickness of the test piece at the target detection point can be obtained in the pose where the detection direction of the ultrasonic detection device is parallel to the normal direction of the tangent plane of the target detection point, thereby improving the thickness measurement accuracy of the target detection point.

[0171] In addition, by determining (searching for) the normal direction of the tangent plane of the target detection point with the laser displacement sensor of the positioning device, non-contact measurement of the blade surface of the test piece at the micron level can be performed, with the advantages of high precision and good repeatability. It can effectively avoid the technical problems such as low efficiency and insufficient accuracy caused by manually holding the probe (i.e., the thickness measurement probe) to search for the normal direction of the tangent plane of the target detection point during ultrasonic thickness measurement, and even the damage of the thickness measurement probe caused by contact and collision with the test piece.

[0172] In addition, by establishing a central database, the detection data information of the test piece fed back by the ultrasonic detection device and the control device can be automatically collected and saved, providing data support for subsequent detection reports and statements. Through the software system mainly based on the database, the automation of the analysis of the detection result statements can be realized, and efficient measurement data entry, data traceability, data management, and reporting can be achieved.

[0173] It should be noted that the above detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0174] It should be noted that the terms used herein are for the purpose of describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly dictates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they specify the presence of the features, steps, operations, devices, components, and / or combinations thereof.

[0175] It should be noted that the terms "first", "second", etc. in the description, claims and drawings of the present application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that these terms can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order different from those illustrated or described herein.

[0176] In addition, the terms "comprise" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that comprises a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0177] For ease of description, spatial relative terms such as "above", "over", "on the upper surface", "above" etc. may be used herein to describe the spatial positional relationship of one device or feature to another device or feature as shown in the figures. It should be understood that the spatial relative terms are intended to encompass different orientations in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is inverted, the device described as "above" or "over" another device or structure will then be positioned "below" or "beneath" the other device or structure. Thus, the exemplary term "above" can include both the orientations of "above" and "below". The device may also be positioned in other different ways, such as rotated 90 degrees or in other orientations, and the corresponding interpretations of the spatial relative descriptions used herein will be made.

[0178] In the detailed description above, reference has been made to the accompanying drawings, which form a part hereof. In the drawings, like symbols typically identify like components, unless the context indicates otherwise. The illustrated embodiments described in the detailed description, drawings and claims are not meant to be limiting. Other embodiments may be used and other changes may be made without departing from the spirit or scope of the subject matter presented herein.

[0179] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An automatic thickness measurement method based on point cloud matching, characterized in that, Including: Establish a point cloud model file of the standard test piece according to the external shape characteristics of the standard test piece, where the standard test piece is used to characterize the standard part of the engine blade, and the point cloud model file includes the point cloud data of the standard test piece; Automatically clamp the test piece to be tested, scan the point cloud data of the test piece to be tested, and perform point cloud matching based on the point cloud model file to calibrate the position information of the target detection points of the test piece to be tested; Perform a thickness detection operation on the test piece to be tested after calibrating the position information; Collect the wall thickness of the test piece to be tested at the target detection point in a pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point; Judge whether the collected wall thickness of the target detection point is qualified, and display the detection data related to the target detection point in real time; where The step of automatically clamping the test piece to be tested, scanning the point cloud data of the test piece to be tested, and performing point cloud matching based on the point cloud model file to calibrate the position information of the target detection points of the test piece to be tested includes: performing coincidence matching on the point cloud data of the test piece to be tested at the current position and the point cloud model file to obtain the point cloud matching degree of each target detection point of the test piece to be tested. The point cloud data includes the three-dimensional position coordinates of the target detection points in the specified area of the test piece to be tested; based on the obtained point cloud matching degree, determine the pose transformation relationship for transforming the target detection points of the test piece to be tested from the current position to the target position, and the pose transformation relationship represents the pose transformation relationship from the camera coordinate system to the robot arm base coordinate system; Calculate the reprojection error between the point cloud data of the test piece to be tested at the current position and the point cloud data of the standard test piece in the point cloud model file through the following expression to obtain the point cloud matching degree of each target detection point of the test piece to be tested: Among them, N represents the number of points in the target model corresponding to the point cloud data of the standard test piece; p i represents the i-th point in the target model; represents the position of the corresponding point of the test piece to be tested in the scene point cloud in the reconstructed point cloud, that is, the current position; w i represents the weight coefficient of the i-th point.

2. The automatic thickness measurement method according to claim 1, characterized in that Project from the original three-dimensional position coordinates of the target detection points of the test piece to be tested to the two-dimensional image plane; In the two-dimensional image plane, through grayscale processing, morphological processing and feature selection, extract the blade region of the test piece to be tested, and then convert it to the three-dimensional space to obtain the point cloud data of the blade region. By calculating the three-dimensional coordinate deviation between the point cloud data of the blade region and the point cloud data of the standard test piece, obtain the point cloud matching degree of each target detection point of the test piece to be tested.

3. The automatic thickness measurement method according to claim 1, wherein According to the determined pose transformation relationship, adjust the pose of the test piece to be tested so that the position of the test piece to be tested is consistent with the position of the standard test piece, and obtain the test piece to be tested after calibrating the position information.

4. The automatic thickness measurement method according to claim 1, wherein Identify the blade features of the test piece to be tested, obtain the pose information of the blade region in the camera coordinate system in the blade features, and calibrate the clamping position of the test piece to be tested according to the obtained spatial position information and the pose information of the blade region in the blade features.

5. The automatic thickness measurement method according to claim 4, wherein Convert the pose information of the to-be-tested piece in the camera coordinate system into the pose information of the to-be-tested piece in the robot arm base coordinate system, and calculate the deviation information between the current position and the target position of each target detection point of the to-be-tested piece according to the pose information of the to-be-tested piece in the robot arm base coordinate system.

6. The automatic thickness measurement method according to claim 5, wherein Based on the ultrasonic waveform analysis algorithm, the ultrasonic detection device analyzes and acquires the digital ultrasonic waveform for waveform discrimination to obtain the blade wall thickness of the to-be-tested piece at the target detection point in the pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point.

7. An automatic thickness measurement system based on point cloud matching, which is used to execute the automatic thickness measurement method described in any one of claims 1 to 6, characterized in that, The automatic thickness measurement system includes: A clamping device for clamping the to-be-tested piece and driving the to-be-tested piece to perform the thickness detection action. Before performing the thickness detection action, point cloud matching is performed based on a pre-established point cloud model file to calibrate the position information of the target detection point of the to-be-tested piece; A machine vision device located above the detection position for determining the position information of the target detection point of the to-be-tested piece; A positioning device for determining the normal direction of the tangent plane passing through the target detection point; An ultrasonic detection device including an ultrasonic thickness gauge for collecting the wall thickness of the to-be-tested piece at the target detection point in the pose where the detection direction of the ultrasonic thickness gauge is parallel to the normal direction of the tangent plane of the target detection point; A control device electrically connected to the clamping device and the ultrasonic detection device. The control device is used to control the automatic clamping and automatic thickness detection of the to-be-tested piece and to process the detection data information of the to-be-tested piece; A display device for displaying the detection process, detection data information, and detection result data information of the to-be-tested piece; wherein Automatically clamp the to-be-tested piece, scan the point cloud data of the to-be-tested piece, and perform point cloud matching based on the point cloud model file to calibrate the position information of the target detection point of the to-be-tested piece, including: performing coincidence matching between the point cloud data of the to-be-tested piece at the current position and the point cloud model file to obtain the point cloud matching degree of each target detection point of the to-be-tested piece. The point cloud data includes the three-dimensional position coordinates of the target detection points in the specified area representing the to-be-tested piece; based on the obtained point cloud matching degree, determine the pose transformation relationship for transforming the target detection point of the to-be-tested piece from the current position to the target position, and the pose transformation relationship represents the pose transformation relationship from the camera coordinate system to the robot arm base coordinate system; Calculate the reprojection error between the point cloud data of the to-be-tested piece at the current position and the point cloud data of the standard test piece in the point cloud model file through the following expression to obtain the point cloud matching degree of each target detection point of the to-be-tested piece: Among them, N represents the number of points in the target model corresponding to the point cloud data of the standard test piece; p i represents the i-th point in the target model; represents the position of the corresponding point of the test piece to be tested in the scene point cloud in the reconstructed point cloud, that is, the current position; w i represents the weight coefficient of the i-th point.

8. The automatic thickness measurement system according to claim 7, characterized in that, The automatic thickness measurement system further includes: A storage and feeding device provided on one side of the operation platform; the storage and feeding device includes a work station and a transfer and storage area; A central database, which is used to collect and save the detection data information of the test piece to be tested fed back by the ultrasonic detection device and the control device, and the detection data information includes the measurement time, the pose information of the test piece to be tested in the base coordinate system of the robotic arm, and the blade wall thickness value of the target detection point; the central database is used to process the detection data information to generate a data report.

Images