Residual stress detection method, device, equipment and medium for large machine tool base

Abstract

The application relates to the technical field of machine tool casting detection, and discloses a residual stress detection method, device, equipment and medium for a large machine tool base piece, which comprises the following steps: obtaining a three-dimensional model of the large machine tool base piece, generating a detection point distribution, and obtaining a detection point sequence in a model coordinate system; converting the detection point sequence in the model coordinate system into target detection point coordinates in a world coordinate system; generating an optimal moving track of a moving chassis, and controlling the moving chassis to carry an industrial robot to move to a preset parking position of each target detection point along the optimal moving track; after the moving chassis is parked, adjusting a stress detector carried by the industrial robot by controlling a joint of the industrial robot, and focusing on the target detection point; and collecting residual stress data of the target detection point by the stress detector according to a stress detection instruction sent by an industrial PC. The technical scheme provides an efficient, accurate, safe and adaptive solution for residual stress detection of a large machine tool base piece.

Description

Technical Field

[0001] This application relates to the field of machine tool casting inspection technology, and in particular to a method, apparatus, equipment and medium for detecting residual stress in large machine tool base components. Background Technology

[0002] In the field of high-end equipment manufacturing, the machining accuracy of large machine tools (such as CNC lathes and machining centers) directly determines the quality of downstream products (such as aerospace components and precision molds). The manufacturing quality of machine tool foundation components (bed, column, beam, etc.) is crucial to ensuring the machine tool's accuracy, as they are the core load-bearing and positioning parts. These foundation components are mostly made of cast iron or cast steel through casting processes. During solidification and cooling, the different cooling rates at different parts of the casting (such as areas with varying wall thicknesses or corners) lead to differences in the timing of the transition from a plastic to an elastic state. This results in the formation of mutually restraining internal stresses, i.e., residual stresses.

[0003] The presence of residual stress has multi-dimensional negative impacts on the performance of machine tool base components. On the one hand, during the machining stage, the release of residual stress can cause workpiece warping and deformation, leading to deviations in dimensional accuracy from design requirements (e.g., bed guide rail flatness error exceeding 0.02mm / m), necessitating additional straightening processes and increasing manufacturing costs. On the other hand, during long-term use of the machine tool, the slow redistribution of residual stress can cause continuous micro-deformation of the base components, resulting in drift in machine tool positioning accuracy (e.g., annual drift reaching 0.05mm), reducing the consistency of machined products, and potentially even causing casting cracks due to stress concentration, thus shortening the machine tool's service life. As the requirements for component machining accuracy in aerospace, new energy, and other fields increase to the micron level, how to accurately control the residual stress of machine tool base components has become one of the core technical bottlenecks in high-end machine tool manufacturing.

[0004] To address the issue of residual stress control, the industry has developed various testing technologies, including destructive testing (such as blind hole method and strip cutting method) and non-destructive testing (such as X-ray diffraction and ultrasonic methods). Among these, X-ray diffraction has become the mainstream technology for residual stress detection in machine tool castings due to its advantages such as non-destructive nature, high detection accuracy (stress measurement error ≤ ±5MPa), and ability to perform on-site testing. Currently, mature X-ray stress testing equipment on the market is mainly portable instruments, with the XL-640 X-ray stress meter being a typical example. This type of equipment has been widely used in the casting quality inspection process of machine tool manufacturers.

[0005] However, in the inspection of large machine tool components (such as machine beds exceeding 3m in length and weighing over 5 tons), existing manual X-ray inspection solutions have significant technical shortcomings, making it difficult to meet the requirements for efficient, accurate, and safe inspection. Specific deficiencies are as follows:

[0006] 1) High radiation safety risks and heavy protective pressure on operators.

[0007] The X-ray diffraction method works by emitting high-energy X-rays (tube voltage typically 20-50kV) that penetrate the surface of a casting, using crystal diffraction to calculate residual stress. However, X-rays pose an ionizing radiation hazard to human tissues (such as bones and the hematopoietic system). In current methods, operators must hold the XL-640 instrument's detection head close to the casting surface (usually ≤300mm) for inspection. While protective gear such as lead aprons and caps is required, prolonged operation (each inspection exceeding 2 hours) can easily lead to fatigue, potentially resulting in improper use of protective equipment (e.g., lead apron collars not closed) or accidental entry into the radiation zone during inspection, posing a radiation exposure risk. Data from a machine tool factory survey shows that, when operating manually, the annual incidence of radiation dose exceeding the standard due to negligence is approximately 0.5 times per 100 personnel, indicating a significant safety hazard.

[0008] 2) Low testing efficiency, making it difficult to adapt to the needs of mass production.

[0009] Large machine tool components typically have a large number of testing points (e.g., 50-100 testing points are needed for the bed, covering the guide rail surface, load-bearing surface, and stress concentration areas at corners). Current solutions require manual operation of the entire process: "moving equipment - positioning the measuring point - adjusting the posture - starting the test - recording data." First, 2-3 operators need to work together to move the portable instrument (weighing approximately 15kg) around the casting, with each movement taking 5-10 minutes. Second, the testing point is manually aligned using the instrument's built-in laser locator, requiring repeated adjustments to the testing head angle (ensuring an angle of 30°-60° with the casting surface normal), with each measuring point's positioning and calibration taking 2-3 minutes. Finally, the instrument software is manually clicked to start the test, with each measuring point taking 8-10 seconds, and the test data (such as stress values ​​and measuring point coordinates) must be manually recorded. According to this process, it takes 4-6 hours to complete the inspection of a large machine bed, while the daily output of castings in machine tool factories is usually 10-15 units. The inspection efficiency of the existing solution is far from meeting the requirements of batch production, making the inspection process a bottleneck in production.

[0010] 3) Detection accuracy is easily affected by human factors, resulting in poor data reliability.

[0011] Several factors during manual operation can lead to fluctuations in detection accuracy: First, when manually adjusting the orientation of the detection head, the operator's visual judgment deviation (such as an angle error of ±5°) can cause the X-ray incident angle to deviate from the preset range, resulting in a 20%-30% decrease in diffraction signal intensity, which in turn increases the stress calculation error to ±15MPa. Second, fatigue caused by prolonged operation can reduce the repeatability of positioning calibration (such as a positional deviation of ±2mm between two positioning operations at the same measuring point), making it impossible to accurately reflect the stress state of the same area of ​​the casting. Third, when manually recording data, errors or omissions may occur (such as confusing the correspondence between measuring point coordinates and stress values), leading to deviations in subsequent stress distribution analysis (such as drawing stress cloud diagrams), affecting the judgment of casting quality.

[0012] 4) Poor adaptability to large castings, and limited testing coverage.

[0013] The testing head cable of the XL-640 instrument is typically 5m long and needs to be moved by hand. For columns exceeding 2m in height or beds exceeding 5m in length, some areas (such as the top of the column and both ends of the bed) are difficult to reach, requiring the construction of temporary scaffolding or the use of forklifts. This not only increases the complexity of operation but may also prevent the testing head from being adjusted to the optimal position due to space constraints, forcing the abandonment of some key measuring points (such as the stress concentration area at the connection between the column and the crossbeam). This results in incomplete testing coverage and an inability to fully assess the residual stress state of the casting.

[0014] In summary, existing manual X-ray residual stress detection solutions face multiple problems in the detection of large machine tool components, including high radiation safety risks, low efficiency, poor accuracy, and insufficient adaptability. Summary of the Invention

[0015] This application provides a method, apparatus, equipment, and medium for detecting residual stress in large machine tool foundation components, aiming to solve many technical problems existing in related technologies, such as high radiation safety risks of X-ray diffraction methods, high protection pressure on operators, low detection efficiency, difficulty in adapting to the needs of mass production, detection accuracy being easily affected by human factors, poor data reliability, poor adaptability to large castings, and limited detection coverage.

[0016] In a first aspect, embodiments of this application provide a method for detecting residual stress in a large machine tool foundation, including:

[0017] A three-dimensional model of the large machine tool base component is obtained, and a detection point distribution is generated based on the three-dimensional model to obtain a sequence of detection points in the model coordinate system;

[0018] The physical space image of the large machine tool base component is acquired by a binocular vision system. The transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system is established by combining hand-eye calibration technology. The detection point sequence under the model coordinate system is converted into the target detection point coordinates under the world coordinate system.

[0019] Based on the coordinates of the target detection points, an optimal movement trajectory for the mobile chassis is generated, and the mobile chassis carrying the industrial robot is controlled to move along the optimal movement trajectory to the preset stopping position of each target detection point.

[0020] Once the mobile chassis comes to a stop, the joints of the industrial robot are controlled to adjust the stress detector it carries, so that the stress detector is focused on the target detection point.

[0021] Based on the stress detection command received from the industrial PC, the residual stress data of the target detection point is collected by the stress detector.

[0022] In one embodiment, optionally, obtaining a three-dimensional model of the large machine tool base component includes:

[0023] If the large machine tool base component has a preset three-dimensional model, directly call the preset three-dimensional model and verify the consistency between the model size and the actual size;

[0024] If the large machine tool base component does not have a preset three-dimensional model, collect point cloud data of its surface;

[0025] The point cloud data is sequentially subjected to denoising, registration, and simplification processes, and then an initial 3D model is generated using the Poisson surface reconstruction algorithm.

[0026] The initial 3D model is calibrated based on the measured key dimensions of the large machine tool base component to obtain a 3D model that matches the actual object.

[0027] In one embodiment, optionally, generating the detection point distribution based on the 3D model includes:

[0028] Based on the structural features of the three-dimensional model, a stress concentration region identification algorithm is invoked to mark key regions that are prone to residual stress concentration.

[0029] In the critical area, detection points are generated in a grid pattern at a first preset interval, and in the non-critical area, detection points are generated at a second preset interval, wherein the second preset interval is greater than the first preset interval;

[0030] The generated detection points are filtered for validity, and detection points located on non-detection surfaces of the large machine tool base and those exceeding the working radius of the industrial robot are removed, forming a sequence of detection points in the model coordinate system.

[0031] In one embodiment, optionally, generating the optimal movement trajectory of the mobile chassis based on the coordinates of the target detection point includes:

[0032] Using a path planning algorithm, starting from the current position of the mobile chassis and taking the preset stopping positions of each target detection point as waypoints, an environmental map containing all obstacles is constructed.

[0033] Based on the environmental map, the initial optimized trajectory is calculated with the goal of minimizing the total path length, the number of turns, and ensuring that the distance between the mobile chassis and each obstacle is greater than a preset distance.

[0034] The initial optimized trajectory is smoothed to obtain the optimal movement trajectory.

[0035] In one embodiment, optionally, adjusting the stress detector carried by the industrial robot by controlling its joints to focus the stress detector on the target detection point includes:

[0036] Real-time images of the target detection point area are acquired through a binocular vision system, the contour features of the target detection point are extracted, and its three-dimensional coordinates in the binocular vision coordinate system are calculated.

[0037] Based on the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system, the three-dimensional coordinates of the target detection point are converted into coordinates under the industrial robot coordinate system to determine the initial distance between the detection head of the stress detector and the target detection point.

[0038] The joints of the industrial robot are controlled to adjust the position of the detection head by a preset step size. At the same time, the distance sensor built into the detection head provides feedback on the real-time distance to the target detection point until the real-time distance reaches the focusing range of the stress detector.

[0039] During the adjustment of the detection head, an obstacle avoidance model is constructed using the artificial potential field method, and the motion constraints of each joint of the industrial robot are calculated so that the minimum distance between the detection head and the surface of the large machine tool base is greater than a preset distance value.

[0040] In one embodiment, optionally, the method further includes:

[0041] When adjusting the focus of the stress detector, the relative posture of the detection head and the target detection point is monitored in real time through a binocular vision system. If the angle between the axis of the detection head and the surface normal of the target detection point deviates from the preset angle range, the joint fine adjustment of the industrial robot is triggered until the posture meets the requirements.

[0042] Secondly, embodiments of this application provide a residual stress detection system for large machine tool foundation components, the system further comprising:

[0043] A binocular vision system is installed on one side of the large machine tool base component to acquire physical space images of the large machine tool base component and extract the contour features and three-dimensional coordinates of the detection points.

[0044] The mobile chassis is used to receive movement commands sent by an industrial PC, move along a preset optimal movement trajectory to the preset stopping position of each target detection point, and achieve stable stopping through the braking function;

[0045] The industrial robot is detachably connected to the mounting flange of the mobile chassis via a bottom connecting seat. The robot's end effector is equipped with a clamping mechanism for fixing the stress detector. The robot's working radius covers all target detection points of the casting. After the mobile chassis stops, the position of the end effector is adjusted by joint control to drive the stress detector to focus on the target detection points.

[0046] The stress detector is fixed to the end effector of the industrial robot by a clamping mechanism. The detector head of the analyzer faces the direction of the casting inspection and is equipped with a distance sensor to receive stress detection commands from the industrial PC, emit X-rays and collect diffraction signals to calculate the residual stress data of the target detection point.

[0047] An industrial PC is placed on a safe operating platform outside the casting inspection area. It communicates with a binocular vision system, a mobile chassis, an industrial robot, and a stress detector via an industrial bus. It is used to generate movement commands, posture adjustment commands, and stress detection commands, while storing residual stress data and displaying the operating status of each device.

[0048] Thirdly, embodiments of this application provide a residual stress detection device for a large machine tool foundation component, comprising:

[0049] The acquisition module is used to acquire the three-dimensional model of the large machine tool base component, and generate the detection point distribution based on the three-dimensional model to obtain the detection point sequence in the model coordinate system;

[0050] The coordinate transformation module is used to acquire physical space images of the large machine tool base components through a binocular vision system, establish the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system by combining hand-eye calibration technology, and convert the detection point sequence under the model coordinate system into the target detection point coordinates under the world coordinate system.

[0051] The mobile module is used to generate the optimal movement trajectory of the mobile chassis based on the coordinates of the target detection points, and control the mobile chassis to carry the industrial robot to move along the optimal movement trajectory to the preset stopping position of each target detection point.

[0052] An adjustment module is used to adjust the stress detector carried by the industrial robot by controlling the joints of the mobile chassis after the mobile chassis stops, so that the stress detector is focused on the target detection point.

[0053] The acquisition module is used to acquire residual stress data of the target detection point through the stress detector according to the stress detection command sent by the industrial PC.

[0054] Fourthly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for detecting residual stress in large machine tool foundation components.

[0055] Fifthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described method for detecting residual stress in large machine tool foundation components.

[0056] In the above-mentioned scheme for residual stress detection of large machine tool foundation components, the following steps are taken: A three-dimensional model of the large machine tool foundation component is obtained, and a detection point distribution is generated based on the three-dimensional model to obtain a detection point sequence in the model coordinate system. A physical space image of the large machine tool foundation component is acquired through a binocular vision system. A transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system is established using hand-eye calibration technology, converting the detection point sequence in the model coordinate system into target detection point coordinates in the world coordinate system. Based on the target detection point coordinates, an optimal movement trajectory for the mobile chassis is generated, and the mobile chassis carrying the industrial robot is controlled to move along the optimal movement trajectory to the preset stopping positions of each target detection point. After the mobile chassis stops, the joints of the industrial robot are controlled to adjust the stress detector it carries, so that the stress detector focuses on the target detection point. Based on the stress detection command received from the industrial PC, the residual stress data of the target detection point is collected through the stress detector. Through the above-mentioned technical solution of the present invention, on the one hand, the traditional manual operation mode is completely changed. Operators can remotely monitor through an industrial PC, avoiding close contact with X-rays and fundamentally eliminating the risk of radiation exposure. At the same time, there is no need for manual handling of equipment and point-by-point calibration, which greatly reduces labor intensity. On the other hand, by standardizing the planning of detection points through a three-dimensional model, combined with binocular vision and hand-eye calibration, a precise mapping from virtual space to physical space is achieved. With the fine adjustment of robot joints, detection errors caused by manual positioning deviation and fatigue are effectively avoided. In addition, the optimal trajectory planning of the mobile chassis and the automated closed loop of movement-positioning-detection greatly shorten the detection time of a single large basic component, meeting the detection needs of batch production. At the same time, the robot's working radius has a wide coverage range and can be adapted to large castings with a height of over 2m and a length of over 5m, solving the problem of incomplete detection coverage in traditional solutions. It provides an efficient, accurate, safe and adaptable integrated technical solution for residual stress detection of large machine tool basic components. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1A A schematic flowchart of a method for detecting residual stress in a large machine tool base component according to an embodiment of this application is shown.

[0059] Figure 1B A schematic diagram of a residual stress detection system for a large machine tool base component according to an embodiment of this application is shown.

[0060] Figure 2 A schematic flowchart of step S101 in a method for detecting residual stress in a large machine tool base component according to an embodiment of this application is shown.

[0061] Figure 3 A schematic flowchart of step S101 in a method for detecting residual stress in a large machine tool base component according to another embodiment of this application is shown.

[0062] Figure 4 A schematic flowchart of step S103 in a method for detecting residual stress in a large machine tool base component according to an embodiment of this application is shown.

[0063] Figure 5 A schematic flowchart of step S104 in a method for detecting residual stress in a large machine tool base component according to an embodiment of this application is shown.

[0064] Figure 6 A block diagram of a residual stress detection system for a large machine tool base component according to an embodiment of this application is shown.

[0065] Figure 7 A block diagram of a residual stress detection device for a large machine tool base component according to an embodiment of this application is shown.

[0066] Figure 8 A block diagram of a computer device according to one embodiment of this application is shown. Detailed Implementation

[0067] To better understand the technical solution of this application, the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0068] It should be understood that the described embodiments are merely some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0069] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0070] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0071] Please see Figure 1A , Figure 1AA schematic flowchart of a method for detecting residual stress in a large machine tool base component according to an embodiment of this application is shown.

[0072] like Figure 1A As shown, the process of a residual stress detection method for a large machine tool foundation component according to an embodiment of this application includes:

[0073] Step S101: Obtain the three-dimensional model of the large machine tool base component, and generate the detection point distribution based on the three-dimensional model to obtain the detection point sequence in the model coordinate system;

[0074] Large machine tool foundation components are the core load-bearing and positioning parts of the machine tool, such as the bed, column, and crossbeam. They are mostly cast structures, and their residual stress directly affects the machining accuracy and service life of the machine tool.

[0075] A 3D model is a digital model built using modeling software (such as SolidWorks or UG) that matches the actual dimensions of the base component. It includes the structural features of the base component (such as holes, corners, and wall thickness) and is used to plan the location of inspection points.

[0076] The distribution of testing points is a set of points on a three-dimensional model that need to be tested, set according to the residual stress distribution law of the basic component, reflecting the spatial coverage of the test.

[0077] The model coordinate system is a three-dimensional Cartesian coordinate system attached to the three-dimensional model. It takes a fixed feature point of the model (such as the center point of the bottom surface) as the origin and is used to define the coordinate position of the detection point to facilitate subsequent coordinate transformation.

[0078] like Figure 2 As shown, in one embodiment, optionally, step S101 includes:

[0079] Step S201: If the large machine tool base component has a preset three-dimensional model, directly call the preset three-dimensional model and verify the consistency between the model size and the actual size;

[0080] A pre-set 3D model is a digital model of the basic components (such as in STL or STEP format) built by the machine tool manufacturer during the design phase. It is usually stored in a product database and has complete structural features and dimensional information. Dimensional consistency verification is performed by actually measuring the key dimensions of the basic components and comparing them with the corresponding dimensions of the pre-set model to verify whether the model matches the actual object and to avoid model failure due to manufacturing errors.

[0081] Step S202: If the large machine tool base component does not have a preset 3D model, collect point cloud data of its surface; the point cloud data is a set of 3D coordinates of a large number of points on the surface of the base component collected by a laser scanner (such as a 3D laser scanner), with a point density typically of 50-100 points / mm. 2It reflects the surface morphology of the actual object.

[0082] Step S203: Denoising, registration and simplification are performed on the point cloud data in sequence, and then an initial three-dimensional model is generated by the Poisson surface reconstruction algorithm.

[0083] Denoising removal removes noise points from point cloud data (such as isolated points caused by environmental dust reflection or outliers caused by scanning errors) and retains valid points through statistical filtering algorithms.

[0084] Registration processing is required when the size of the base part exceeds the single scan range of the scanner. It involves scanning from multiple perspectives and stitching the point clouds from multiple perspectives into a complete point cloud using the Iterative Closest Point (ICP) algorithm to ensure that the stitching error is ≤0.2mm.

[0085] Simplification involves reducing the number of point clouds and the computational load for subsequent modeling while preserving the structural features of the basic components, and avoiding oversimplification that could lead to feature loss.

[0086] The Poisson surface reconstruction algorithm is an algorithm that constructs a continuous polygonal surface model (such as a triangular patch model) based on the normal vector information of point cloud data, and can generate a smooth initial 3D model that fits the point cloud.

[0087] Step S204: Based on the measured key dimensions of the large machine tool base component, the initial three-dimensional model is calibrated to obtain a three-dimensional model that matches the actual object.

[0088] Critical dimension calibration involves measuring the critical dimensions of the base component (such as length, height, and wall thickness) and adjusting the corresponding dimensions of the initial 3D model to ensure a perfect match between the model and the actual object.

[0089] In this embodiment, it is first determined whether the target basic component has a preset three-dimensional model (such as a design model provided by a machine tool manufacturer): if it exists, it is directly imported into the modeling software and the key dimensions of the basic component (such as length and wall thickness) are measured by a laser rangefinder to verify the consistency between the dimensions of the model and the actual object; if it does not exist, the point cloud data of the surface of the basic component is collected by a laser scanning device and a three-dimensional model is generated after processing.

[0090] Based on the structural features of the 3D model, the test points are planned according to the principle of dense stress concentration areas and sparse non-stress concentration areas: for example, test points are set at 50mm intervals on the bed guide rail mounting surface (stress concentration area) and at 150mm intervals on the non-load-bearing side (non-stress concentration area), thus forming a sequence of test point coordinates in the model coordinate system (such as (x1,y1,z1), (x2,y2,z2)...).

[0091] In this way, the visualization-based point selection based on the 3D model can cover key stress areas, avoid missed or redundant detections, and ensure the comprehensiveness and representativeness of the detection data.

[0092] Step S102, as follows Figure 1B As shown, the physical space image of the large machine tool base component is acquired through a binocular vision system. The transformation relationship between the binocular vision coordinate system and the industrial robot (i.e., robotic arm) coordinate system is established by combining hand-eye calibration technology. The detection point sequence under the model coordinate system is converted into the target detection point coordinates under the world coordinate system.

[0093] A binocular vision system is a vision acquisition device consisting of two industrial cameras with identical parameters, a supplementary lighting module, and an image processing unit. It calculates the three-dimensional coordinates of an object based on the principle of parallax and has non-contact positioning capabilities.

[0094] Hand-eye calibration technology is a technique for establishing the transformation relationship between the binocular vision coordinate system (the coordinate system based on the camera lens) and the industrial robot coordinate system (the coordinate system based on the robot base). It eliminates equipment installation errors through calibration board calibration and ensures coordinate transformation accuracy.

[0095] The world coordinate system is a global coordinate system fixed at the testing site. It takes a fixed point on the ground as the origin and is used to uniformly describe the spatial position of the basic components, mobile chassis, and industrial robots, so as to realize the collaborative positioning of multiple devices.

[0096] In this step, the binocular vision system is fixed 1.5-3m away from the base component, with the lens facing the detection surface of the base component. The supplementary lighting module is activated (ensuring an ambient light level ≥ 500 lux) to acquire physical space images of the base component. During hand-eye calibration, a checkerboard calibration board is fixed to the end effector of the industrial robot. The robot is controlled to move the calibration board to 8-12 different poses within the visual field, acquiring images of the calibration board in each pose. The visual coordinates and robot coordinates of the checkerboard corner points are extracted, and the transformation matrix is ​​solved using the least squares method to establish the association between the two coordinate systems. The sequence of detection points in the model coordinate system is first converted to world coordinate system coordinates through the mapping relationship between the model and the world coordinate system. Then, combined with the transformation matrix from hand-eye calibration, the final target detection point coordinates in the world coordinate system are obtained, ensuring a one-to-one correspondence between virtual and physical points. Through the above technical solution, accurate mapping from virtual space to physical space is achieved, solving the problem of traditional manual positioning and laying the foundation for subsequent precise robot movement. By calibrating a unified coordinate system using hand and eye, deviations caused by binocular vision and the robot's independent positioning are avoided, ensuring that the detection head can accurately align with the target point.

[0097] Step S103: Based on the coordinates of the target detection points, generate the optimal movement trajectory of the mobile chassis, and control the mobile chassis to carry the industrial robot to move along the optimal movement trajectory to the preset stopping position of each target detection point.

[0098] A mobile chassis is a mobile platform that carries industrial robots. It is equipped with omnidirectional wheels with braking function at the bottom, and is equipped with LiDAR and drive motor. It has autonomous navigation and obstacle avoidance capabilities and is used to move the robot to different inspection areas.

[0099] The optimal movement trajectory is a chassis movement path that is short, efficient, safe, and collision-free. It is obtained through algorithm optimization, taking into account factors such as distance, number of turns, and obstacles.

[0100] The preset stopping position is the chassis stopping position within the working radius of the industrial robot, which is a distance from the target detection point. It is necessary to ensure that the chassis is stable and without tilting to avoid affecting the robot's posture adjustment accuracy.

[0101] In this step, based on the target detection point coordinates in the world coordinate system, the path planning module of the chassis control system sets the waypoints according to the spatial order of the detection point distribution, starting from the current position of the chassis, and imports obstacle information (such as equipment brackets and tool cabinets) at the detection site to construct an environmental map.

[0102] Specifically, the initial trajectory can be calculated using either the A or RRT algorithm, with the optimization objectives being the shortest total path length, ≤3 turns, and a distance ≥300mm from obstacles. After obtaining the initial optimized trajectory, it is smoothed using B-spline curve fitting (to eliminate sharp angles) to ensure that the rate of change of chassis movement speed is ≤0.1m / s. 2 To avoid sudden stops and turns that could cause the robot to shake.

[0103] The chassis moves along the optimal trajectory and, upon reaching the preset stopping position at each target detection point, triggers the braking function to ensure stopping stability.

[0104] In the above technical solutions, compared with manually propelling the equipment, automated trajectory planning shortens the time required for chassis movement, and obstacle avoidance function prevents collisions with base components or surrounding equipment, reducing the risk of equipment damage. Smooth stopping and moving can reduce vibration at the robot's end effector, providing a stable foundation for accurate focusing of the detection head and avoiding detection deviations caused by chassis shaking.

[0105] Step S104: After the mobile chassis stops, the stress detector carried by the industrial robot is adjusted by controlling the joints of the robot so that the stress detector is focused on the target detection point.

[0106] Industrial robots are multi-jointed robotic arms with high-precision motion control capabilities. Their end effectors can be equipped with stress detectors to adjust the position and orientation of the analyzers.

[0107] Stress detectors are non-destructive testing devices based on the principle of X-ray diffraction (such as μ-X360s). They calculate residual stress by emitting X-rays to collect crystal diffraction signals. The core component is the detection head (including X-ray emission and reception modules).

[0108] Focusing involves adjusting the distance between the detector head and the target detection point to the effective detection range of the analyzer, ensuring a clear diffraction signal, which is a prerequisite for obtaining accurate stress data.

[0109] In this step, after the chassis comes to a stop, the robot control system receives a focusing command. First, it acquires an image of the target detection point area using a binocular vision system, extracts the detection point's outline, and calculates its three-dimensional coordinates. Combining this with the hand-eye calibration transformation matrix, the visual coordinates are converted to robot coordinates, determining the initial distance between the detection head and the detection point. The robot's waist, upper arm, and forearm joints are then controlled to move the detection head closer to the detection point in preset steps, while the detection head's built-in laser distance sensor provides real-time distance data.

[0110] When the feedback distance enters the preset focus range, the joint movement is paused to complete focusing. Throughout the process, a collision detection algorithm calculates the distance between the detection head and the surface of the base component in real time to avoid collisions.

[0111] Step S105: Based on the stress detection command sent by the industrial PC, the residual stress data of the target detection point is collected by the stress detector.

[0112] The stress testing command is a trigger signal sent from an industrial PC (control center) to the stress testing instrument. It contains testing parameters (such as X-ray tube voltage, current, and testing duration) and is used to start the testing process.

[0113] Residual stress data is the stress value (in MPa) of the target detection point obtained by the analyzer through processing diffraction signals. It is divided into tensile stress (positive value) and compressive stress (negative value), reflecting the stress state inside the base component.

[0114] In this step, after focusing is complete, the robot sends a ready signal to the industrial PC. Upon receiving the signal, the PC sends a stress detection command to the analyzer via an industrial bus (such as EtherCAT). This command includes parameters matching the base material (e.g., for 45# steel, pipe voltage 30kV, current 5mA, and detection time 8 seconds). After receiving the command, the analyzer activates the X-ray emission module to continuously acquire diffraction signals. After filtering (Gaussian filtering) and noise reduction, the residual stress data is calculated using Bragg's law. Once the detection is complete, the data is automatically uploaded to the PC's database and stored with the detection point coordinates and timestamp.

[0115] In this way, no manual intervention is required from focusing to data acquisition, avoiding the lag of manual detection. Unified detection parameters and automatic data storage avoid errors caused by manual parameter setting, and the data is bound to coordinates and time, which facilitates subsequent stress distribution analysis and problem tracing.

[0116] Through the above technical solutions, the traditional manual operation mode is completely changed. Operators can remotely monitor via an industrial PC, avoiding close contact with X-rays and fundamentally eliminating the risk of radiation exposure. At the same time, there is no need for manual equipment handling and point-by-point calibration, which greatly reduces labor intensity. On the other hand, by standardizing the planning of detection points through a three-dimensional model, combined with binocular vision and hand-eye calibration, a precise mapping from virtual space to physical space is achieved. With the fine adjustment of robot joints, detection errors caused by manual positioning deviation and fatigue are effectively avoided. In addition, the optimal trajectory planning of the mobile chassis and the automated closed loop of movement-positioning-detection greatly shorten the detection time of a single large basic component, meeting the needs of batch production detection. At the same time, the robot's working radius has a wide coverage range and can be adapted to large castings with a height of over 2m and a length of over 5m, solving the problem of incomplete detection coverage in traditional solutions. This provides an efficient, accurate, safe, and adaptable integrated technical solution for residual stress detection of large machine tool basic components.

[0117] like Figure 3 As shown, in one embodiment, optionally, generating the detection point distribution based on the three-dimensional model includes:

[0118] Step S301: Based on the structural features of the three-dimensional model, call the stress concentration region identification algorithm to mark the key regions that are prone to residual stress concentration;

[0119] Structural features are the geometric features of the basic components, such as right-angle corners, the edges of round holes, abrupt changes in wall thickness, and welded joints. These areas are prone to stress concentration during casting or machining.

[0120] The stress concentration region identification algorithm is a geometric curvature analysis algorithm based on a three-dimensional model. By calculating the curvature value of each point on the model surface, the larger the absolute value of the curvature, the more likely it is to be a stress concentration region, eliminating the need for manual judgment based on experience.

[0121] In this step, the 3D model is imported into stress analysis software (such as ANSYS), and a stress concentration region identification algorithm is run. The software automatically traverses the model surface, calculates the curvature value of each triangular facet, and marks areas with an absolute curvature value greater than a preset value as red (stress concentration areas), such as the connection corner between the bed guide rail and the column, and the edge of the worktable mounting hole; areas with an absolute curvature value less than or equal to the preset value are marked as blue (non-concentration areas), such as the side plane of the bed. After marking, a region boundary file is generated for subsequent inspection point planning. In this way, by using a unified curvature threshold, differences in judgment among different operators are avoided, ensuring consistency in inspection point planning and facilitating batch inspection of multiple basic components.

[0122] Step S302: In the critical area, detection points are generated in a grid pattern at a first preset interval, and in the non-critical area, detection points are generated at a second preset interval, wherein the second preset interval is greater than the first preset interval;

[0123] The first preset spacing is the spacing between detection points in the stress concentration area. For example, it can be 40-80mm. It is necessary to ensure dense coverage in order to accurately reflect the stress distribution gradient (such as the change from tensile stress to compressive stress).

[0124] The second preset spacing is the spacing between detection points in non-stress concentration areas. For example, it can be 120-180mm, which is greater than the first preset spacing. This reduces redundant detection and workload while ensuring the representativeness of the detection.

[0125] For marked stress concentration areas, a mesh generation tool can be used to generate a uniform mesh at 60mm (first preset spacing), with the mesh nodes serving as detection points. For non-stress concentration areas, mesh nodes are generated at 150mm (second preset spacing) as detection points. For example, the guide rail mounting surface (stress concentration area) of a certain machine bed has an area of ​​2m × 0.5m, and approximately 278 detection points are generated at a spacing of 60mm; the non-stress concentration area on the side has an area of ​​2m × 0.3m, and approximately 27 detection points are generated at a spacing of 150mm, balancing detection accuracy and efficiency.

[0126] Step S303: Validity screening of the generated detection points is performed, and detection points located on non-detection surfaces of the large machine tool base and those exceeding the working radius of the industrial robot are removed, thus forming a sequence of detection points in the model coordinate system.

[0127] Non-test surfaces are areas on the surface of the base component that are not suitable for stress testing, such as surfaces with oxide scale, oil stains, burrs, or high-precision mating surfaces that have been machined (testing may damage the surface precision). Effective diffraction signals cannot be obtained in such areas, and the corresponding test points must be eliminated.

[0128] The working radius of an industrial robot is the maximum spatial distance that the end effector (equipped with an analyzer) of the industrial robot can reach (based on the robot's base). Detection points that exceed this range cannot be focused by adjusting the robot's posture and must be rejected.

[0129] In this step, the generated inspection points are first screened for non-inspection surfaces: images of the base component surface are acquired using a vision system to identify areas with oxide scale or oil stains (such as areas with dark surface color or uneven reflection), and inspection points in these areas are marked and removed. Simultaneously, high-precision mating surfaces that have already been machined are checked, and inspection points within the mating surface range marked on the process drawings are removed. Next, the working radius is screened: the working radius parameters of the industrial robot (e.g., 1.5m) are imported, and the straight-line distance from each inspection point to the robot base is calculated by combining the coordinates of the inspection point in the world coordinate system with the coordinates of the robot base. If the distance exceeds the working radius, the inspection point is removed. The final retained inspection points form a sequence of inspection points in the model coordinate system, ensuring that each inspection point meets the conditions of being located on a qualified inspection surface and within the robot's working radius.

[0130] In the above technical solution, eliminating non-detection surfaces and detection points outside the detection range can reduce the amount of invalid detection work (such as avoiding invalid signals and rework caused by detection on oily surfaces), and improve the overall detection efficiency. Retaining only the qualified detection points ensures that the subsequently acquired diffraction signals are clear and accurate, avoiding data deviations caused by improper selection of detection points.

[0131] like Figure 4 As shown, in one embodiment, optionally, step S103 includes:

[0132] Step S401: Using a path planning algorithm, with the current position of the mobile chassis as the starting point and the preset stopping positions of each target detection point as the waypoints, an environmental map containing each obstacle is constructed.

[0133] Path planning algorithms are used to calculate the optimal path for a mobile chassis from the starting point to the destination. Commonly used algorithms include Algorithm A (based on heuristic search, highly efficient) and RRT algorithm (suitable for complex obstacle environments, robust). The core is to optimize path parameters while avoiding obstacles.

[0134] Obstacles refer to objects at the testing site that may impede the movement of the chassis, such as machine tool accessories, tool cabinets, cable trays, etc. Their spatial locations (coordinates and dimensions) need to be accurately marked on the environmental map to avoid chassis collisions.

[0135] The environmental map is a digital map built based on the actual layout of the testing site, including information such as the starting point of the chassis, the stopping position of the testing point, and the location of obstacles.

[0136] In this step, the site is first scanned using LiDAR to acquire 3D point cloud data of the environment. This data is then imported into path planning software to identify and label obstacles: stationary objects in the point cloud data with a height greater than a preset height are identified as obstacles, and their minimum bounding box coordinates in the world coordinate system are recorded (e.g., the range from x1-y1-z1 to x2-y2-z2 for a tool cabinet). Next, the path start and waypoints are determined: the initial parking position of the mobile chassis (e.g., at the entrance to the inspection site) is used as the start point, and the preset parking positions of each target inspection point (e.g., 500-800mm away from the inspection point, within the chassis's stable support range) are used as waypoints. The waypoints are arranged according to the spatial distribution order of the inspection points (e.g., from the left end to the right end of the base component) to avoid path intersections and backtracking. Finally, an environmental map containing the start point, waypoints, and obstacle locations is generated, providing complete spatial information for subsequent trajectory calculations.

[0137] Step S402: Based on the environment map, the initial optimized trajectory is calculated with the goal of minimizing the total path length, the number of turns, and the distance between the mobile chassis and each obstacle being greater than a preset distance.

[0138] The trajectory optimization objective is the core indicator for measuring the quality of a trajectory, including the shortest total path length (reducing travel time), the fewest turns (reducing chassis steering time and energy consumption), and the distance to obstacles being greater than the preset distance (ensuring safety). It is necessary to simultaneously satisfy multiple objectives for optimization.

[0139] For example, the initial optimization cost is calculated using Algorithm A. The steps are as follows: A heuristic function (e.g., Manhattan distance, estimating the straight-line distance from the current point to the stopping position) is set to guide the search towards the target direction and reduce the invalid search range. For each candidate path node, the path length cost and obstacle distance cost are calculated: the path length cost is positively correlated with the distance from the node to the starting point, and the obstacle distance cost is negatively correlated with the distance from the node to the obstacle (the closer the distance, the higher the cost). The node with the lowest total cost is selected first as the next path point, and the path from the starting point to the first stopping position is gradually generated. Then, using this stopping position as the new starting point, the path to the next stopping position is calculated until all stopping positions are covered. If the total path length increases by ≤10% compared to the straight-line distance, the number of turns is ≤3 times / segment (between adjacent stopping positions), and the distance to the obstacle is ≥300mm, it is determined to be the initial optimization trajectory. If these conditions are not met (e.g., 4 turns), the path nodes are readjusted until the requirements are met.

[0140] In this way, the initial optimized trajectory shortens the movement time while ensuring a safe distance from obstacles, avoiding the pursuit of efficiency at the expense of collision risks. The initial trajectory already meets the core optimization objectives, and subsequent processing only requires smoothing without major modifications, thus improving trajectory planning efficiency.

[0141] Step S403: Smooth the initial optimized trajectory to obtain the optimal movement trajectory.

[0142] Smoothing is achieved by using mathematical algorithms (such as B-spline curves and Bézier curves) to adjust the node positions of the initial trajectory, eliminating sharp angles in the trajectory (such as 90° right-angle turns), making the chassis movement speed change continuous, and avoiding sudden acceleration and deceleration.

[0143] Specifically, B-spline curves can be used to smooth the initial optimized trajectory. The specific steps include: extracting key nodes of the initial trajectory (such as the starting point, turning point, and stopping position) as control vertices of the B-spline curve; setting the curve order and calculating the smoothed trajectory curve; calculating the rate of curvature change of the smoothed trajectory, while verifying that the distance between the trajectory and obstacles is still ≥300mm and the total path length increases by ≤5%; discretizing the smoothed trajectory into motion commands that can be executed by the chassis (such as one position point every 0.1 seconds) and sending them to the chassis drive system.

[0144] In this way, the smooth trajectory avoids sharp turns of the chassis caused by sharp corners, reduces vibration at the robot's end effector, and the continuous speed change reduces the number of motor starts and stops, thereby reducing motor energy consumption and wheel wear and extending the service life of the equipment.

[0145] like Figure 5 As shown, in one embodiment, optionally, step S104 includes:

[0146] Step S501: Acquire real-time images of the target detection point area through a binocular vision system, extract the contour features of the target detection point and calculate its three-dimensional coordinates in the binocular vision coordinate system;

[0147] Contour feature extraction involves identifying the boundary shape of target detection points in a visual image (e.g., if the detection point is preset as a circular marker, the contour feature is a circular edge), and extracting the contour using an edge detection algorithm (such as the Canny algorithm) to calculate the center coordinates of the detection point.

[0148] In this step, the binocular vision system acquires real-time images of the target detection point region. First, image preprocessing is performed: grayscale conversion (converting the color image to grayscale), Gaussian filtering (eliminating noise with a 5×5 kernel), and thresholding (converting the grayscale image to a black-and-white binary image to highlight the detection point outline) enhance contour contrast. Then, the Canny edge detection algorithm is used to extract the detection point outline. If the detection point is a preset circular marker, the Hough circle detection algorithm is used to identify the circle within the outline and calculate the pixel coordinates of the circle's center. Combining the intrinsic parameters (such as focal length and pixel size) and extrinsic parameters (camera position) of the binocular vision system, the pixel coordinates are converted to three-dimensional coordinates in the binocular vision coordinate system (e.g., x = 1200mm, y = 800mm, z = 500mm) using the parallax principle.

[0149] In the above scheme, the coordinate error of contour feature extraction and disparity calculation is extremely small, providing a precise target position reference for subsequent robot pose adjustment and avoiding focus shift caused by detection point positioning deviation. Even if there is slight reflection on the surface of the base component (such as machined surfaces), the contour can still be accurately extracted through preprocessing and edge detection, avoiding positioning failure caused by ambient light interference.

[0150] Step S502: Based on the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system, the three-dimensional coordinates of the target detection point are converted into coordinates under the industrial robot coordinate system, and the initial distance between the detection head of the stress detector and the target detection point is determined.

[0151] In this step, the visual 3D coordinates of the detection point are converted into coordinates in the robot coordinate system (e.g., (x = 1180mm, y = 790mm, z = 490mm)) based on the transformation matrix between the binocular vision coordinate system and the robot coordinate system. At the same time, the current robot coordinates of the robot end effector (detection head) are read (e.g., (x = 1000mm, y = 790mm, z = 490mm)), and the straight-line distance between the two is calculated (initial distance 180mm).

[0152] If the initial distance is greater than 300mm (exceeding the upper limit of the focusing range), control the robot's waist joint (driving the upper arm and forearm) to move towards the detection point in steps of 2mm / step. For each step, the distance sensor will provide feedback on the real-time distance. When the real-time distance drops to the 100-300mm range (e.g., 250mm), switch to a small step size of 0.5mm / step for fine-tuning until the distance stabilizes at 200mm (center of the focusing range, where the signal is clearest), and then pause the joint movement.

[0153] Step S503: Control the joints of the industrial robot to adjust the position of the detection head by a preset step length, and at the same time, use the distance sensor built into the detection head to feed back the real-time distance to the target detection point until the real-time distance reaches the focusing range of the stress detector.

[0154] The preset step size is the minimum displacement increment for robot joint adjustment. If the step size is too small, it will increase the adjustment time, and if the step size is too large, it may cause focus overshoot (missing the focus range). It needs to be selected reasonably according to the initial distance.

[0155] The distance sensor is a laser distance sensor integrated into the detection head of the stress detector. It is used to provide real-time feedback on the distance between the detection head and the detection point, avoiding errors from manual visual estimation.

[0156] Step S504: During the adjustment of the detection head, an obstacle avoidance model is constructed using the artificial potential field method, and the motion constraints of each joint of the industrial robot are calculated so that the minimum distance between the detection head and the surface of the large machine tool base is greater than a preset distance value.

[0157] Artificial potential field method is an obstacle avoidance algorithm that treats the robot as a point mass, obstacles as repulsive potential field sources (the closer the distance, the greater the repulsive force), and detection points as attractive potential field sources (the closer the distance, the greater the attractive force). The robot's motion direction is determined by calculating the direction of the resultant force to avoid collisions.

[0158] Motion constraints are the maximum allowable motion angles / displacements of each joint of the robot calculated based on the obstacle avoidance model. For example, the maximum rotation angle of the forearm joint is ≤90°, ensuring that the joint movement does not exceed the mechanical limit while avoiding obstacles.

[0159] During joint adjustment, an obstacle avoidance model is constructed using the artificial potential field method. The process includes:

[0160] Set the repulsive potential field parameters of the obstacle (e.g., repulsive force coefficient 50 N / m) and the attractive potential field parameters of the detection point (attractive force coefficient 30 N / m); calculate in real time the repulsive force (from the base surface and surrounding equipment) and attractive force (from the detection point) experienced by the robot end effector (detection head), and the direction of the resultant force is the safe direction of movement; calculate the motion increment of each joint based on the direction of the resultant force: if the direction of the resultant force is away from the base surface, the waist joint is allowed to move towards the detection point; if the direction of the resultant force is towards the base surface (risk of collision), the movement of the waist joint is restricted, and only the wrist joint is allowed to make minor adjustments to its posture. Simultaneously set joint motion constraints: waist joint rotation angle ≤ ±30°, upper arm joint rotation angle ≤ 90°, and forearm joint rotation angle ≤ 120° to prevent joints from exceeding mechanical limits and causing equipment damage.

[0161] Through the above technical solutions, the obstacle avoidance model can identify collision risks in real time (e.g., when the detection head is only 30mm away from the surface of the base component, the repulsive force is greater than the attractive force, and the head stops approaching), reducing the collision risk to zero and protecting the base component and detection equipment (e.g., preventing damage to the X-ray tube caused by the detection head hitting the base component). Motion constraints avoid ineffective joint movements (e.g., excessive rotation of the lumbar joint), reduce joint adjustment time, and improve focusing efficiency.

[0162] In one embodiment, optionally, the method further includes:

[0163] When adjusting the focus of the stress detector, the relative posture of the detection head and the target detection point is monitored in real time through a binocular vision system. If the angle between the axis of the detection head and the surface normal of the target detection point deviates from the preset angle range, the joint fine adjustment of the industrial robot is triggered until the posture meets the requirements.

[0164] The relative attitude is the angle between the axis of the detection head and the normal to the surface of the target detection point. It reflects the degree of tilt of the detection head and directly affects the incident angle of X-rays. An angle deviation will lead to a decrease in the intensity of the diffraction signal and an increase in the error of stress calculation.

[0165] Based on the detection principle of the stress detector (X-ray diffraction must meet the Bragg angle condition), the preset angle range that can obtain effective signals is set. The preset angle range varies slightly for different materials (such as steel and cast iron), and is usually 30°-60°.

[0166] Joint fine-tuning is a precise adjustment for posture deviations, mainly achieved through the wrist joints of industrial robots (such as wrist rotation joints and swing joints). The wrist joints have higher adjustment precision and can achieve small-amplitude posture corrections.

[0167] In this step, while adjusting the focus, the binocular vision system continuously acquires relative pose images of the detection head and the detection point (acquisition frequency 5Hz): It extracts the contour features of the detection head (e.g., the rectangular edge of the detection head shell) and calculates the pixel direction of the detection head axis (along the center line of the long side of the rectangle); it extracts the texture features of the detection point surface (e.g., surface processing textures) and calculates the normal direction of the detection point surface (perpendicular to the fitting plane) using a plane fitting algorithm; it converts the detection head axis direction and normal direction into vectors in the world coordinate system, calculates the angle between the two vectors (i.e., the relative pose angle), and compares it with a preset angle range (e.g., 45°±15°). In this way, by quantifying the angle deviation, the direction and magnitude of joint fine-tuning are clearly defined, avoiding blind adjustments.

[0168] If the relative posture angle deviates from the preset range (e.g., the included angle 25° < 30°), the joint fine-tuning process is triggered: The posture correction is calculated: the target angle is 45°, the current angle is 25°, requiring an increase of 20°. Based on the robot's kinematic model, the wrist swing joint needs to rotate upwards by 15° (the mapping relationship between joint rotation angle and posture angle is obtained through calibration); the wrist joint is controlled to rotate in small steps of 0.1° / step. Every 0.5° rotation, the binocular vision system recalculates the relative posture angle; when the relative posture angle enters the preset range (e.g., 40°), joint fine-tuning stops, and posture calibration is completed. During fine-tuning, the distance between the detection head and the detection point is monitored simultaneously via a distance sensor to ensure that the distance remains within the focus range (e.g., 200mm ± 5mm) during posture fine-tuning, avoiding focus shift due to posture adjustment. In this way, fine-tuning is achieved only through the wrist joint, without changing the distance between the detection head and the detection point, eliminating the need for refocusing, saving adjustment time, and improving the continuity of the detection process.

[0169] Figure 6 A block diagram of a residual stress detection system for a large machine tool base component according to an embodiment of this application is shown.

[0170] like Figure 6 As shown, in a second aspect, embodiments of this application provide a residual stress detection system for large machine tool foundation components, comprising:

[0171] A binocular vision system 61 is set on one side of the large machine tool base component and is used to acquire physical space images of the large machine tool base component and extract the contour features and three-dimensional coordinates of the detection points.

[0172] The mobile chassis 62 is used to receive the movement command sent by the industrial PC, move along the preset optimal movement trajectory to the preset stopping position of each target detection point, and achieve stable stopping through the braking function;

[0173] The industrial robot 63 is detachably connected to the mounting flange of the mobile chassis via a bottom connecting seat. The end effector of the robot is equipped with a clamping mechanism for fixing the stress detector. The working radius of the robot covers all target detection points of the casting. After the mobile chassis stops, the position of the end effector is adjusted by joint control to drive the stress detector to focus on the target detection points.

[0174] The stress detector 64 is fixed to the end effector of the industrial robot by a clamping mechanism. The detector head of the analyzer faces the direction of the casting inspection and is equipped with a distance sensor to receive stress detection commands from the industrial PC, emit X-rays and collect diffraction signals, and calculate the residual stress data of the target detection point.

[0175] An industrial PC 65 is placed on a safe operating platform outside the casting inspection area. It communicates with a binocular vision system, a mobile chassis, an industrial robot, and a stress detector via an industrial bus. It is used to generate movement commands, posture adjustment commands, and stress detection commands, while storing residual stress data and displaying the operating status of each device.

[0176] Figure 7 A block diagram of a residual stress detection device for a large machine tool base component according to an embodiment of this application is shown.

[0177] like Figure 7 As shown, in a second aspect, embodiments of this application provide a residual stress detection device 70 for large machine tool foundation components, comprising:

[0178] The acquisition module 71 is used to acquire the three-dimensional model of the large machine tool base component, and generate the detection point distribution based on the three-dimensional model to obtain the detection point sequence in the model coordinate system;

[0179] The coordinate transformation module 72 is used to acquire physical space images of the large machine tool base component through the binocular vision system, establish the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system by combining hand-eye calibration technology, and convert the detection point sequence under the model coordinate system into the target detection point coordinates under the world coordinate system.

[0180] The mobile module 73 is used to generate the optimal movement trajectory of the mobile chassis based on the coordinates of the target detection points, and control the mobile chassis to carry the industrial robot to move along the optimal movement trajectory to the preset stopping position of each target detection point.

[0181] The adjustment module 74 is used to adjust the stress detector carried by the industrial robot by controlling the joints of the mobile chassis after the mobile chassis stops, so that the stress detector is focused on the target detection point.

[0182] The acquisition module 75 is used to acquire residual stress data of the target detection point through the stress detector according to the stress detection command sent by the industrial PC.

[0183] In one embodiment, optionally, the acquisition module includes:

[0184] The calling unit is used to directly call the preset three-dimensional model and verify the consistency between the model size and the actual size if the large machine tool basic component has a preset three-dimensional model.

[0185] The acquisition unit is used to acquire point cloud data of the surface of the large machine tool base component if there is no preset three-dimensional model.

[0186] The processing unit is used to sequentially perform denoising, registration and simplification processing on the point cloud data, and then generate an initial three-dimensional model through the Poisson surface reconstruction algorithm.

[0187] A calibration unit is used to calibrate the initial three-dimensional model based on the measured key dimensions of the large machine tool base component, so as to obtain a three-dimensional model that matches the actual object.

[0188] In one embodiment, optionally, the acquisition module further includes:

[0189] The marking unit is used to call the stress concentration region identification algorithm based on the structural features of the three-dimensional model to mark the key areas that are prone to residual stress concentration.

[0190] The generation unit is used to generate detection points in a grid pattern at a first preset interval in the critical area, and to generate detection points at a second preset interval in the non-critical area, wherein the second preset interval is greater than the first preset interval;

[0191] The filtering unit is used to filter the validity of the generated detection points, and remove detection points located on non-detection surfaces of the large machine tool base and those exceeding the working radius of the industrial robot, thus forming a sequence of detection points in the model coordinate system.

[0192] In one embodiment, optionally, the moving module includes:

[0193] The map building unit is used to construct an environmental map containing various obstacles by using a path planning algorithm, with the current position of the mobile chassis as the starting point and the preset stopping positions of each target detection point as the waypoints.

[0194] The first calculation unit is used to calculate an initial optimized trajectory based on the environment map, with the goal of minimizing the total path length, the number of turns, and the distance between the mobile chassis and each obstacle being greater than a preset distance.

[0195] A smoothing unit is used to smooth the initial optimized trajectory to obtain the optimal movement trajectory.

[0196] In one embodiment, optionally, the adjustment module includes:

[0197] The extraction unit is used to acquire real-time images of the target detection point area through a binocular vision system, extract the contour features of the target detection point and calculate its three-dimensional coordinates in the binocular vision coordinate system.

[0198] The distance determination unit is used to convert the three-dimensional coordinates of the target detection point into coordinates in the industrial robot coordinate system according to the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system, and to determine the initial distance between the detection head of the stress detector and the target detection point.

[0199] The position adjustment unit is used to control the joints of the industrial robot to adjust the position of the detection head by a preset step size, and at the same time, the real-time distance between the detection head and the target detection point is fed back by the distance sensor built into the detection head until the real-time distance reaches the focusing range of the stress detector.

[0200] The second calculation unit is used to construct an obstacle avoidance model using the artificial potential field method during the adjustment of the detection head, and to calculate the motion constraints of each joint of the industrial robot so that the minimum distance between the detection head and the surface of the large machine tool base is greater than a preset distance value.

[0201] In one embodiment, optionally, the apparatus further includes:

[0202] The monitoring module is used to monitor the relative posture of the detection head and the target detection point in real time through a binocular vision system when adjusting the focus state of the stress detector. If the angle between the axis of the detection head and the normal of the surface of the target detection point deviates from the preset angle range, the industrial robot joint fine adjustment is triggered until the posture meets the requirements.

[0203] Fourthly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for detecting residual stress in large machine tool foundation components.

[0204] Fifthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described method for detecting residual stress in large machine tool foundation components.

[0205] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the residual stress detection device and each module of the large machine tool foundation described above can be referred to the corresponding process in the aforementioned embodiments of the residual stress detection method for large machine tool foundations, and will not be repeated here.

[0206] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the model training device and each module described above can be referred to the corresponding process in the aforementioned embodiment of the residual stress detection method for large machine tool foundation components, and will not be repeated here.

[0207] The aforementioned residual stress detection device for large machine tool foundation components can be implemented as a computer program, which can, for example... Figure 8 It runs on the computer device shown.

[0208] Figure 8 A block diagram of a computer device according to one embodiment of this application is shown.

[0209] See Figure 8 The computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include storage media and internal memory.

[0210] The storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any of the residual stress detection methods for large machine tool foundation components based on multi-source data provided in the embodiments of this application.

[0211] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0212] Internal memory provides an environment for the execution of computer programs stored in the storage medium. When executed by a processor, these programs can enable the processor to perform methods for analyzing the transmission paths of any infectious disease or training predictive neural networks. The storage medium can be non-volatile or volatile.

[0213] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0214] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0215] In addition, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions for performing the steps described in the first aspect embodiment.

[0216] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or electronic device described above can be referred to the relevant descriptions in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0217] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0218] It should be understood that although the terms "first," "second," etc., may be used to describe the setting units in the embodiments of this application, these setting units should not be limited to these terms. These terms are only used to distinguish the setting units from each other. For example, without departing from the scope of the embodiments of this application, the first setting unit may also be referred to as the second setting unit, and similarly, the second setting unit may also be referred to as the first setting unit.

[0219] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0220] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0221] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.

[0222] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0223] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for detecting residual stress in a large machine tool foundation component, characterized in that, The method includes: A three-dimensional model of the large machine tool base component is obtained, and a detection point distribution is generated based on the three-dimensional model to obtain a sequence of detection points in the model coordinate system; The physical space image of the large machine tool base component is acquired by a binocular vision system. The transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system is established by combining hand-eye calibration technology. The detection point sequence under the model coordinate system is converted into the target detection point coordinates under the world coordinate system. Based on the coordinates of the target detection points, an optimal movement trajectory for the mobile chassis is generated, and the mobile chassis carrying the industrial robot is controlled to move along the optimal movement trajectory to the preset stopping position of each target detection point. Once the mobile chassis comes to a stop, the joints of the industrial robot are controlled to adjust the stress detector it carries, so that the stress detector is focused on the target detection point. Based on the stress detection command received from the industrial PC, the residual stress data of the target detection point is collected by the stress detector. The detection point distribution is generated based on the three-dimensional model, including: Based on the structural features of the three-dimensional model, a stress concentration region identification algorithm is invoked to mark key regions that are prone to residual stress concentration. In the critical area, detection points are generated in a grid pattern at a first preset interval, and in the non-critical area, detection points are generated at a second preset interval, wherein the second preset interval is greater than the first preset interval; The generated detection points are filtered for validity, and detection points located on non-detection surfaces of the large machine tool base and those exceeding the working radius of the industrial robot are removed, forming a sequence of detection points in the model coordinate system; The step of generating the optimal movement trajectory of the mobile chassis based on the coordinates of the target detection point includes: Using a path planning algorithm, starting from the current position of the mobile chassis and taking the preset stopping positions of each target detection point as waypoints, an environmental map containing all obstacles is constructed. Based on the environmental map, the initial optimized trajectory is calculated with the goal of minimizing the total path length, the number of turns, and ensuring that the distance between the mobile chassis and each obstacle is greater than a preset distance. The initial optimized trajectory is smoothed to obtain the optimal movement trajectory; Adjusting the stress detector carried by the joints of the industrial robot to focus the stress detector on the target detection point includes: Real-time images of the target detection point area are acquired through a binocular vision system, the contour features of the target detection point are extracted, and its three-dimensional coordinates in the binocular vision coordinate system are calculated. Based on the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system, the three-dimensional coordinates of the target detection point are converted into coordinates under the industrial robot coordinate system to determine the initial distance between the detection head of the stress detector and the target detection point. The joints of the industrial robot are controlled to adjust the position of the detection head by a preset step size. At the same time, the distance sensor built into the detection head provides feedback on the real-time distance to the target detection point until the real-time distance reaches the focusing range of the stress detector. During the adjustment of the detection head, an obstacle avoidance model is constructed using the artificial potential field method, and the motion constraints of each joint of the industrial robot are calculated so that the minimum distance between the detection head and the surface of the large machine tool base is greater than a preset distance value.

2. The method according to claim 1, characterized in that, Obtaining the three-dimensional model of the large machine tool basic component includes: If the large machine tool base component has a preset three-dimensional model, directly call the preset three-dimensional model and verify the consistency between the model size and the actual size; If the large machine tool base component does not have a preset three-dimensional model, collect point cloud data of its surface; The point cloud data is sequentially subjected to denoising, registration, and simplification processes, and then an initial 3D model is generated using the Poisson surface reconstruction algorithm. The initial 3D model is calibrated based on the measured key dimensions of the large machine tool base component to obtain a 3D model that matches the actual object.

3. The method according to claim 1, characterized in that, The method further includes: When adjusting the focus of the stress detector, the relative posture of the detection head and the target detection point is monitored in real time through a binocular vision system. If the angle between the axis of the detection head and the surface normal of the target detection point deviates from the preset angle range, the joint fine adjustment of the industrial robot is triggered until the posture meets the requirements.

4. A residual stress detection system for large machine tool foundation components, employing the residual stress detection method for large machine tool foundation components as described in any one of claims 1 to 3, characterized in that, The system also includes: A binocular vision system is installed on one side of the large machine tool base component to acquire physical space images of the large machine tool base component and extract the contour features and three-dimensional coordinates of the detection points. The mobile chassis is used to receive movement commands sent by an industrial PC, move along a preset optimal movement trajectory to the preset stopping position of each target detection point, and achieve stable stopping through the braking function; The industrial robot is detachably connected to the mounting flange of the mobile chassis via a bottom connecting seat. The robot's end effector is equipped with a clamping mechanism for fixing the stress detector. The robot's working radius covers all target detection points of the casting. After the mobile chassis stops, the position of the end effector is adjusted by joint control to drive the stress detector to focus on the target detection points. The stress detector is fixed to the end effector of the industrial robot by a clamping mechanism. The detector head of the analyzer faces the direction of the casting inspection and is equipped with a distance sensor to receive stress detection commands from the industrial PC, emit X-rays and collect diffraction signals to calculate the residual stress data of the target detection point. An industrial PC is placed on a safe operating platform outside the casting inspection area. It communicates with a binocular vision system, a mobile chassis, an industrial robot, and a stress detector via an industrial bus. It is used to generate movement commands, posture adjustment commands, and stress detection commands, while storing residual stress data and displaying the operating status of each device.

5. A residual stress detection device for a large machine tool foundation component, characterized in that, include: The acquisition module is used to acquire the three-dimensional model of the large machine tool base component, and generate the detection point distribution based on the three-dimensional model to obtain the detection point sequence in the model coordinate system; The coordinate transformation module is used to acquire physical space images of the large machine tool base components through a binocular vision system, establish the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system by combining hand-eye calibration technology, and convert the detection point sequence under the model coordinate system into the target detection point coordinates under the world coordinate system. The mobile module is used to generate the optimal movement trajectory of the mobile chassis based on the coordinates of the target detection points, and control the mobile chassis to carry the industrial robot to move along the optimal movement trajectory to the preset stopping position of each target detection point. An adjustment module is used to adjust the stress detector carried by the industrial robot by controlling the joints of the mobile chassis after the mobile chassis stops, so that the stress detector is focused on the target detection point. The acquisition module is used to acquire residual stress data of the target detection point through the stress detector according to the stress detection command sent by the industrial PC. The acquisition module further includes: The marking unit is used to call the stress concentration region identification algorithm based on the structural features of the three-dimensional model to mark the key areas that are prone to residual stress concentration. The generation unit is used to generate detection points in a grid pattern at a first preset interval in the critical area, and to generate detection points at a second preset interval in the non-critical area, wherein the second preset interval is greater than the first preset interval; The filtering unit is used to filter the validity of the generated detection points, and remove detection points located on the non-detection surface of the large machine tool base and those exceeding the working radius of the industrial robot, thus forming a sequence of detection points in the model coordinate system. The mobile module includes: The map building unit is used to construct an environmental map containing various obstacles by using a path planning algorithm, with the current position of the mobile chassis as the starting point and the preset stopping positions of each target detection point as the waypoints. The first calculation unit is used to calculate an initial optimized trajectory based on the environment map, with the goal of minimizing the total path length, the number of turns, and the distance between the mobile chassis and each obstacle being greater than a preset distance. A smoothing unit is used to smooth the initial optimized trajectory to obtain the optimal movement trajectory; The adjustment module includes: The extraction unit is used to acquire real-time images of the target detection point area through a binocular vision system, extract the contour features of the target detection point and calculate its three-dimensional coordinates in the binocular vision coordinate system. The distance determination unit is used to convert the three-dimensional coordinates of the target detection point into coordinates in the industrial robot coordinate system according to the transformation relationship between the binocular vision coordinate system and the industrial robot coordinate system, and to determine the initial distance between the detection head of the stress detector and the target detection point. The position adjustment unit is used to control the joints of the industrial robot to adjust the position of the detection head by a preset step size, and at the same time, the real-time distance between the detection head and the target detection point is fed back by the distance sensor built into the detection head until the real-time distance reaches the focusing range of the stress detector. The second calculation unit is used to construct an obstacle avoidance model using the artificial potential field method during the adjustment of the detection head, and to calculate the motion constraints of each joint of the industrial robot so that the minimum distance between the detection head and the surface of the large machine tool base is greater than a preset distance value.

6. A computer device, characterized in that, include: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, the instructions being configured to perform the method according to any one of claims 1 to 3.

7. A computer-readable storage medium, characterized in that, The device stores computer-executable instructions for performing the method as described in any one of claims 1 to 3.

Images