Shafting error measuring equipment and method

Abstract

The invention provides a shafting error measurement device and method, the shafting error measurement device comprises a motion platform, a constant force loading device, a visual measurement system and a control unit, the motion platform is provided with a product clamp, and the product clamp is used for fixing a measured product. The motion platform is used for driving the product clamp mounted on the motion platform to move along three translational degrees of freedom and one rotational degree of freedom; the constant force loading device comprises at least one constant force loading assembly, and each constant force loading assembly is used for actively loading each shaft of the tested product; the vision measurement system is used for acquiring an image of a tested product and transmitting the acquired image of the tested product to the control unit; and the control unit is electrically connected with the motion platform, the constant force loading device and the measuring system, and is used for performing image analysis on the acquired image of the measured product to obtain axial displacement data in the loading process, and measuring the axial clearance of the measured product according to the axial displacement data.

Description

Technical Field

[0001] This invention relates to the field of precision instrument testing technology, and in particular to a shaft system error measurement device and method. Background Technology

[0002] Precision instruments and equipment in engineering applications play a crucial role in the development of science and technology. These instruments and equipment are composed of precision shaft systems, each of which undergoes a series of processes including design, parts machining, and assembly. Due to errors in parts machining and assembly, the precision shaft system exhibits erroneous motion during rotation, including radial, axial, tilting, perpendicular, and orthogonal errors. These errors contain both random and systematic components, and the systematic error significantly affects the motion accuracy of the precision shaft system. Therefore, there is an urgent need for a shaft system error measurement device. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art, and proposes a shaft system error measurement device and method.

[0004] In a first aspect, the present invention discloses a shaft system error measuring device, comprising:

[0005] A motion platform is provided, on which a product fixture is mounted. The product fixture is used to fix the product to be tested. The motion platform is used to drive the product fixture mounted on it to move along three translational degrees of freedom and one rotational degree of freedom, thereby realizing the spatial axis positioning of the product to be tested.

[0006] A constant force loading device, comprising at least one constant force loading component, each constant force loading component being used to apply active loading to each axis of the product under test;

[0007] A vision measurement system, comprising at least one vision measurement module, wherein the vision measurement module is used to acquire images of the product under test and transmit the acquired images of the product under test to a control unit;

[0008] The control unit is electrically connected to the motion platform, the constant force loading device, and the measurement system. The control unit is used to perform image analysis on the acquired images of the product under test to obtain axial displacement data during the loading process, and to measure the axial clearance of the product under test based on the axial displacement data.

[0009] In some embodiments, the control unit is used to perform image analysis on the acquired image of the product under test to obtain axial displacement data during the loading process, and to measure the axial clearance of the product under test based on the axial displacement data, including: extracting key feature points of the platform under test from the image of the product under test, calculating the three-dimensional coordinates of the feature points, calculating the coordinate difference of the feature points during the loading process to obtain axial displacement data, and associating the axial displacement data with the rotation angle of the shaft system and the loading load parameters to measure the axial clearance.

[0010] In some embodiments, the control unit is further configured to call the camera distortion parameters obtained from the pre-measurement calibration, perform distortion correction on the pixel coordinates of the feature points, and then combine the distance information collected by the laser ranging module to calculate the three-dimensional coordinates of the feature points in the camera coordinate system.

[0011] In some embodiments, the control unit is further configured to convert the three-dimensional coordinates of feature points in the camera coordinate system into three-dimensional coordinates in the measurement coordinate system using a coordinate system transformation matrix obtained through calibration, thereby providing physical space data with a unified reference for axis system error calculation.

[0012] In some embodiments, the vision measurement system further includes a laser ranging module, which is used to collect multi-point spatial position data of the product under test and transmit it to the control unit. The control unit is used to determine whether the product under test is within the optimal working range of the vision measurement module based on the multi-point spatial position data. If the product under test is not within the optimal working range of the vision measurement module, the control unit controls the motion platform to adjust the position of the product under test based on the multi-point spatial position data fed back by the laser ranging module until the product under test is within the optimal working range of the vision measurement module.

[0013] In some embodiments, the visual measurement module includes a camera and a light source, the camera and the light source are fixedly mounted on a camera mounting base, and at least one visual measurement module is provided on the visual measurement module.

[0014] In some embodiments, the shaft error measuring device of the present invention further includes a target assembly, which is fixedly installed on one side of the pitch axis of the product under test. The target assembly includes a dark field bright line crosshair reticle, a wavelength selective filter, and a backlight light source. The wavelength selective filter is located between the dark field bright line crosshair reticle and the backlight light source. The dark field bright line crosshair reticle, the wavelength selective filter, and the backlight light source are all mounted on a target adjustment mounting bracket.

[0015] In some embodiments, the constant force loading device includes a lateral constant force loading component for applying a lateral constant force to the product under test and a vertical constant force loading component for applying a vertical constant force to the product under test. The lateral constant force loading component includes a lateral loading drive device, which is connected to one end of a first tension / compression sensor. The other end of the first tension / compression sensor is connected to the roll axis end of the product under test via a first fixture. The vertical constant force loading component includes a vertical loading drive device, which is connected to one end of a second tension / compression sensor. The other end of the second tension / compression sensor is connected to the pitch axis end of the product under test via a second fixture.

[0016] In some embodiments, the motion platform includes a transverse axis assembly, a longitudinal axis assembly, a lifting axis assembly, and a rotation axis assembly. The transverse axis assembly is used to drive the product under test to move laterally, the longitudinal axis assembly is used to drive the product under test to move longitudinally, the lifting axis assembly is used to drive the product under test to move up and down, and the rotation axis assembly is used to drive the product under test to rotate around the vertical Z-axis.

[0017] In some embodiments, the control unit is used to acquire the starting voltage data of the motor of the product under test, and to obtain the frictional torque of the product under test under different axial clearances based on the starting voltage data, so as to evaluate the optimal control range of the axial clearance of the product under test.

[0018] Secondly, the present invention also discloses a method for measuring shaft system error, based on the shaft system error measuring device described in the first aspect, the method comprising the following steps:

[0019] The constant force loading component is controlled to apply active loading to each axis of the product under test.

[0020] During the loading process, the vision measurement system is controlled to acquire feature images of the product under test according to different loading loads;

[0021] Extract key feature points of the platform under test from the image of the product under test, and calculate the three-dimensional spatial coordinates of each feature point;

[0022] Calculate the coordinate difference of the same feature point under different working conditions during loading to obtain the axial displacement data of the shaft system;

[0023] By performing multidimensional correlation analysis between axial displacement data and corresponding loading parameters, shaft system error measurement can be achieved.

[0024] The present invention has at least the following beneficial effects:

[0025] This invention automates the entire measurement process, significantly improving inspection efficiency. The control unit enables coordinated and automated control of the motion platform (shaft positioning), the constant force loading device (active loading), and the vision measurement system (image acquisition). No manual step-by-step operation is required. From product clamping, positioning, loading, displacement acquisition to gap data calculation, the entire process is completed automatically. Compared with traditional manual measurement, the inspection efficiency is increased several times, which can meet the industrial inspection needs of batch precision shaft products.

[0026] This invention employs non-contact measurement, effectively protecting the precision surfaces of the product being measured. The visual measurement module acquires axial displacement data through image acquisition without any physical contact with the measured shaft system, completely avoiding scratches, indentations, and other damage to the precision mating surfaces and smooth surfaces of the shaft system caused by traditional contact measuring tools (dial indicators, dial indicators, etc.). It is especially suitable for the inspection of easily damaged products such as high-precision, high-smoothness stable platform shaft systems and precision transmission shaft systems. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the structure of a shaft system error measuring device provided in an embodiment of the present invention;

[0028] Figure 2 This is a schematic diagram of a lifting shaft assembly provided in an embodiment of the present invention;

[0029] Figure 3 This is a schematic diagram of two-axis intersection measurement provided in an embodiment of the present invention;

[0030] Figure 4 This is a schematic diagram of two-axis perpendicularity measurement provided in an embodiment of the present invention;

[0031] Figure 5 A schematic diagram of the crosshair target assembly provided in an embodiment of the present invention;

[0032] Figure 6 This is a schematic diagram of the system coordinate system definition provided in an embodiment of the present invention;

[0033] Figure 7 The diagram shows an ideal subpixel edge detection model provided in an embodiment of the present invention; wherein, a is a schematic diagram of the ideal edge model; b is a schematic diagram of the original edge image; and c is a schematic diagram of the rotated edge image.

[0034] Figure 8 This is a schematic block diagram of the shaft system error measuring device provided in an embodiment of the present invention;

[0035] Figure 9 A schematic diagram of the object-space coordinate system is established;

[0036] Figure 10 This is a schematic diagram of the structure of a vision measurement module provided in an embodiment of the present invention.

[0037] In the attached diagram, 1 is the motion platform, 11 is the transverse axis assembly, 12 is the longitudinal axis assembly, 13 is the lifting axis assembly, 131 is the second mounting plate, 132 is the lead screw, 133 is the nut, 134 is the motor, 135 is the reducer, 136 is the slide rail, 137 is the lifting frame, 138 is the grating ruler, 14 is the rotation axis assembly, 21 is the transverse constant force loading assembly, 22 is the vertical constant force loading assembly, 3 is the product under test, 4 is the vision measurement module, 41 is the camera, 42 is the light source, 43 is the laser ranging module, 5 is the measurement bracket, 6 is the base platform, 7 is the target assembly, 71 is the first adjustment knob, 72 is the second adjustment knob, 73 is the target body mounting base, 74 is the third adjustment knob, 75 is the fourth adjustment knob, 76 is the dark field bright line crosshair reticle, 77 is the target support frame, 78 is the base mounting flange, and 79 is the pressure ring. Detailed Implementation

[0038] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. In the description of this invention, unless otherwise stated, "a plurality" or "several" means two or more. Similarly, "an," "a," or "the," and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the element or object listed following the word and its equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0040] In the various figures, the same elements are represented by similar reference numerals. For clarity, not all parts in the figures are drawn to scale. Furthermore, some well-known parts may not be shown in the figures.

[0041] Many specific details of the invention, such as the structure, materials, dimensions, processing methods, and techniques of the components, are described below to provide a clearer understanding of the invention. However, as those skilled in the art will understand, the invention may be implemented without following these specific details.

[0042] Precision instruments and equipment in engineering applications play a crucial role in the development of science and technology. These instruments and equipment are composed of precision shaft systems, each of which undergoes a series of processes including design, parts machining, and assembly. During this process, errors in parts machining and assembly lead to erroneous motions in the precision shaft system during rotation, including radial, axial, tilting, perpendicular, and orthogonal errors. These errors contain both random and systematic components, and the systematic error significantly affects the motion accuracy of the precision shaft system. Therefore, to further improve the accuracy of precision shaft systems, precision measurement systems can be used to measure and correct these errors, providing a basis for precise adjustment of the shaft system.

[0043] See Figure 1 and Figure 8 The present invention provides a shaft system error measuring device, comprising:

[0044] Motion platform 1, on which a product fixture is installed, the product fixture is used to fix the product 3 to be tested, and the motion platform 1 is used to drive the product fixture installed thereon to move along three translational degrees of freedom and one rotational degree of freedom, so as to realize the spatial axis positioning of the product 3 to be tested;

[0045] A constant force loading device, comprising at least one constant force loading component, each constant force loading component being used to apply active loading to each axis of the product under test 3;

[0046] A visual measurement system, comprising at least one visual measurement module 4, wherein the visual measurement module 4 is used to acquire images of the product under test 3 and transmit the acquired images of the product under test 3 to a control unit;

[0047] The control unit is electrically connected to the motion platform 1, the constant force loading device, and the measurement system. The control unit is used to perform image analysis on the acquired image of the product under test 3 to obtain axial displacement data during the loading process, and to measure the axial clearance of the product under test 3 based on the axial displacement data.

[0048] This invention solves the problems of low spatial positioning accuracy and poor multi-axis adaptability in traditional measurement methods. Traditional axis positioning often relies on manual adjustment or a single-degree-of-freedom motion platform 1, which makes it difficult to achieve precise spatial alignment of the three axes of the tested product. Furthermore, it cannot adapt to axis products of different specifications and installation postures. Manual adjustment is not only inefficient but also prone to positioning deviations due to human error. This device, through a motion platform 1 with three translational and one rotational degrees of freedom, solves the problems of insufficient degrees of freedom, low positioning accuracy, and excessive manual intervention in axis spatial positioning, achieving automated and precise alignment of the three axes of the tested product.

[0049] This invention solves the problems of contact-based displacement acquisition methods, easy product damage, and limited accuracy in traditional measurement. Traditional axial displacement measurement often uses contact measuring tools such as dial indicators and micrometers, which can easily scratch the precision surfaces of the measured shaft system. In addition, contact measurement has low resolution and cannot capture minute axial movement displacements. Furthermore, contact measurement has only one measurement point, making it difficult to comprehensively reflect the overall axial clearance characteristics of the shaft system. This device solves the problems of product damage, low accuracy, and single measurement point in contact measurement by using a non-contact image acquisition system of a vision measurement system, achieving high-precision and non-destructive acquisition of minute axial displacements.

[0050] The device of this invention solves the problems of manual disconnection, non-standardized operation, and poor repeatability in each link by centrally connecting and controlling the motion platform 1, the constant force loading device, and the vision measurement system through the control unit, thereby realizing the automated linkage of the entire measurement process.

[0051] The measurement accuracy is significantly improved, and the results are more precise and reliable. The constant force loading device achieves active, constant, and precise force loading, eliminating the force fluctuation error of passive loading and ensuring that the axial displacement reflects the true clearance of the shaft system. The vision measurement system adopts non-contact image acquisition, which has high resolution and high sampling rate and can capture micron-level axial displacements, far exceeding the accuracy of traditional contact measurement. The three translational and one rotational degrees of freedom of the motion platform 1 achieve micron-level precise spatial positioning of the shaft system, eliminating the measurement error introduced by positioning deviation and greatly improving the absolute accuracy and repeatability of axial clearance measurement.

[0052] In some embodiments, the visual measurement module 4 is fixed on the measurement bracket 5, and the measurement bracket 5 is fixed on the base platform 6.

[0053] The motion platform 1 is fixedly mounted on the base platform 6. The motion platform 1 is located inside the measuring bracket 5.

[0054] In some embodiments, the control unit is used to perform image analysis on the acquired image of the product under test 3 to obtain axial displacement data during the loading process, and to measure the axial clearance of the product under test 3 based on the axial displacement data, including: extracting key feature points of the test platform (the key feature points are arranged along the axial direction of the shaft system) from the image of the product under test 3, calculating the three-dimensional coordinates of the feature points, calculating the coordinate difference of the feature points during the loading process, obtaining axial displacement data, and associating the axial displacement data with the rotation angle of the shaft system and the loading load parameters to measure the axial clearance.

[0055] In some embodiments, the control unit is further configured to call the camera distortion parameters obtained from the pre-measurement calibration, perform distortion correction on the pixel coordinates of the feature points, and then combine the distance information collected by the laser ranging module to calculate the three-dimensional coordinates of the feature points in the camera coordinate system.

[0056] In some embodiments, the control unit is further configured to convert the three-dimensional coordinates of feature points in the camera coordinate system into three-dimensional coordinates in the measurement coordinate system using a coordinate system transformation matrix obtained through calibration, thereby providing physical space data with a unified reference for axis system error calculation.

[0057] In some embodiments, the visual measurement system further includes a laser ranging module, which is used to collect multi-point spatial position data of the product under test 3 and transmit it to the control unit. The control unit is used to determine whether the product under test 3 is within the optimal working range of the visual measurement module 4 based on the multi-point spatial position data. If the product under test 3 is not within the optimal working range of the visual measurement module 4, the control unit controls the motion platform 1 to adjust the position of the product under test 3 based on the multi-point spatial position data fed back by the laser ranging module until the product under test 3 is within the optimal working range of the visual measurement module 4.

[0058] In some embodiments, the visual measurement module 4 includes a camera 41 and a light source 42, the camera 41 and the light source 42 are fixedly mounted on a camera mounting base, and at least one visual measurement module 4 is provided with a laser ranging module 43.

[0059] In some embodiments, the shaft error measuring device of the present invention further includes a target assembly 7, which is fixedly installed on one side of the pitch axis of the product under test 3.

[0060] Because of the high visual measurement accuracy of this invention, traditional visual targets are no longer sufficient; therefore, a self-developed active target is used. The optical form of the new target is as follows:

[0061] The target assembly 7 includes a dark field bright line crosshair 76, a wavelength selective filter, and a backlight source. The wavelength selective filter is located between the dark field bright line crosshair 76 and the backlight source. The dark field bright line crosshair 76, the wavelength selective filter, and the backlight source are all mounted on the target adjustment mounting bracket.

[0062] See Figure 5 The target adjustment mounting frame is a multi-degree-of-freedom precision adjustment frame. The target adjustment mounting frame includes a target body mounting base 73 and a first adjustment knob 71, a second adjustment knob 72, a third adjustment knob 74, and a fourth adjustment knob 75. The dark-field bright-line crosshair reticle 76, the wavelength selective filter, and the backlight source are all mounted in the recess of the target body mounting base 73. The target body mounting base 73 is mounted on a target support frame 77. One end of the target support frame 77 is connected to a base mounting flange 78, and the other end of the base mounting flange 78 is used to connect to the measured stable platform. The recess of the target body mounting base 73 is provided with a pressure ring 79 for fixing optical components such as the reticle and filter to ensure optical path stability.

[0063] Four adjustment knobs allow for translation and angle adjustment of the target in three-dimensional space, ensuring that the crosshair image is clear, centered, and precisely aligned with the camera's field of view.

[0064] The base mounting flange 78 is used to secure the entire target assembly 7 to the test platform. The backlight is a side-mounted LED backlight, used to provide uniform backlighting for the reticle. The wavelength selective filter is used to filter stray light and improve imaging contrast.

[0065] The target component 7 features ultra-high imaging contrast, high scribing precision, and good line uniformity, ensuring the accuracy of feature extraction from the target in the camera 41. Simultaneously, the target possesses multi-degree-of-freedom adjustment capabilities, enabling rapid alignment and focusing of the system.

[0066] The target is fixed on one side of the pitch axis of the stabilized platform being measured. The rigid connection characteristics of the multi-degree-of-freedom adjustment frame can ensure that the target rotates synchronously with the pitch axis. At the beginning of the installation, the fine adjustment function of the adjustment frame can accurately align the center of the crosshairs of the target with the center of the field of view of the camera 41, so as to avoid the impact of installation errors on the subsequent measurement accuracy.

[0067] In some embodiments, the constant force loading device includes a lateral constant force loading component 21 for applying a lateral constant force to the product under test 3 and a vertical constant force loading component 22 for applying a vertical constant force to the product under test 3. The lateral constant force loading component 21 includes a lateral loading drive device, which is connected to one end of a first tension / compression sensor. The other end of the first tension / compression sensor is connected to the roll axis end of the product under test 3 through a first fixture. The vertical constant force loading component 22 includes a vertical loading drive device, which is connected to one end of a second tension / compression sensor. The other end of the second tension / compression sensor is connected to the pitch axis end of the product under test 3 through a second fixture.

[0068] In some embodiments, the motion platform 1 includes a transverse axis assembly 11, a longitudinal axis assembly 12, a lifting axis assembly 13, and a rotation axis assembly 14. The transverse axis assembly 11 is used to drive the tested product 3 to move laterally, the longitudinal axis assembly 12 is used to drive the tested product 3 to move longitudinally, the lifting axis assembly 13 is used to drive the tested product 3 to move up and down, and the rotation axis assembly 14 is used to drive the tested product 3 to rotate around the vertical Z-axis.

[0069] In some embodiments, the motion platform 1 includes a transverse axis assembly 11, a longitudinal axis assembly 12, a lifting axis assembly 13, and a rotation axis assembly 14. The transverse axis assembly 11 is used to drive the product under test to move laterally, the longitudinal axis assembly 12 is used to drive the product under test to move longitudinally, the lifting axis assembly 13 is used to drive the product under test to move up and down, and the rotation axis assembly 14 is used to drive the product under test to rotate around the vertical Z-axis.

[0070] The rotating axis drives the product under test to rotate around the vertical Z-axis (yaw motion), simulating horizontal turning conditions, and can be linked with other axes to achieve complex motion trajectories.

[0071] The lifting shaft drives the rotating shaft to move up and down vertically with the product under test, which is used to adjust the test height or simulate vertical displacement and vibration conditions.

[0072] The longitudinal axis drives the upper mechanism to move linearly along the front-to-back direction (Y-axis), simulating the product's longitudinal displacement.

[0073] The transverse axis drives the entire upper mechanism to move linearly in the left-right direction (X-axis), simulating the product's lateral displacement.

[0074] The transverse axis assembly 11 and the longitudinal axis assembly 12 form a two-axis (XY axis) translation stage. The lifting axis assembly 13 is fixedly mounted on the first mounting plate of the two-axis (XY axis) translation stage. The rotation axis assembly 14 is fixedly mounted on the second mounting plate 131 of the lifting axis assembly 13. The rotation axis assembly 14 includes a rotation axis and a rotation drive device for driving the rotation axis to rotate. The upper end of the rotation axis is fixedly connected to the product fixture.

[0075] See Figure 2 The lifting shaft assembly 13 includes a second mounting plate 131, a lead screw assembly, and a lifting drive device. The second mounting plate 131 is connected to the nut 133 of the lead screw assembly. The lead screw 132 of the lead screw assembly is rotatably mounted on the lifting frame 137. One end of the lead screw of the lead screw assembly is connected to the lifting drive device. The lifting frame is provided with a slide rail 136, and the nut of the lead screw assembly slides in engagement with the slide rail 136. The lifting shaft assembly 13 also includes a grating ruler 138, which is connected to the moving end of the lifting shaft, such as the nut of the lead screw assembly, for real-time detection and feedback of displacement data of the lifting stroke, so as to realize high-precision closed-loop control of the lifting shaft. The lifting drive device includes a motor 134 and a reducer 135.

[0076] The dual-axis constant force loading device can apply a specific force in both directions along the axis of rotation of the workpiece under test. Its function is achieved by connecting the rotating shaft end of the workpiece under test to the tension and compression sensors through a tooling, and then connecting the motor through a structural component. The motor applies tension and compression to the ends of the pitch and roll axes of the platform under test. The tension and compression sensors provide feedback on the force, and the control unit collects the tension and compression data and controls the loading motor to change the force on the axis under test, thus realizing a constant force loading closed loop. At the same time, the visual image measurement system accurately measures the displacement or structural contour changes and measures the axial clearance of the platform.

[0077] The product under test is located at the core test position of the equipment and is the verification object of the entire system. It is usually a precision component such as a stabilizing platform or inertial device. It is mounted above the rotating axis and subjected to the combined test of multi-axis motion and load.

[0078] In some embodiments, the control unit is used to acquire the starting voltage data of the motor of the product under test 3, and to obtain the frictional torque of the product under test 3 under different axial clearances based on the starting voltage data, so as to evaluate the optimal control range of the axial clearance of the product under test 3.

[0079] Before measurement: The intrinsic parameters of the camera are obtained by using a planar target to obtain the distortion parameters of the camera lens for subsequent distortion correction; the extrinsic parameters of the vision measurement system are calibrated by using a planar target and a reference ruler, as well as the coordinate system transformation relationship between the multi-view cameras. In addition, the positional relationship parameters between the camera measurement system and the laser rangefinder system are calibrated by combining the installation position of the laser rangefinder.

[0080] During measurement: First, the product under test is mounted and moved into the measurement station area. The multi-channel laser ranging function is activated until the platform is stable enough to measure multiple points in the space that meet the requirements of the optimal working range of the vision measurement system. Then, the vision measurement system is activated, and the illumination source is started and adjusted according to the image quality. The camera is controlled by the measurement software to acquire images. After image processing and data calculation, geometric feature parameters are measured according to user requirements to obtain the three-dimensional coordinates of a series of key feature points. This allows for the measurement and calculation of basic geometric parameters such as distance, area, straightness, perpendicularity, roundness, and parallelism. The rotation axis, as well as important parameters such as axis intersection, perpendicularity, and axial clearance, are also calculated. The corresponding measurement process data and final results are displayed and stored according to user operation, and data reports can be generated.

[0081] The motion platform 1 mainly includes four motion axes: a rotary axis, a lifting axis, a longitudinal axis, and a transverse axis. Based on technical requirements, a compact yet functional multi-dimensional precision mechanical motion platform 1 is designed. The rotary axis adopts a single-axis turntable structure directly driven by a torque motor, using a circular grating as the angle feedback device. The linear motion axis uses a servo motor and reducer to drive the lead screw, and incorporates a linear grating ruler for fully closed-loop control of linear displacement to meet high-precision positioning requirements.

[0082] The image measurement system mainly includes five measuring cameras and control analysis measurement software; it adopts optical multi-axis machine vision measurement technology to realize non-contact measurement of typical structures, and can realize the measurement of static geometric dimensions and form and position deviations of mechanisms, as well as the measurement of the spatial position (relative to the reference) of mechanical shafts and the position and angle deviations of intersecting shaft systems. With the cooperation of the measurement and control system, it can generate a visualized error model based on the structural model and the measured structure.

[0083] The constant force loading device mainly includes a lateral translation axis and a lifting axis, which can apply bidirectional force loads to the ends of the roll axis and pitch axis of the stabilizing platform, respectively, and perform constant force loading closed loop through tension and compression sensors;

[0084] The control unit controls the motion platform 1 to adjust the posture of the product under test 3 in conjunction with the product under test; it controls the multi-view machine vision image measurement system to perform non-contact measurement of the product under test and processes the test image data; it controls the two-axis constant force loading device to apply specific forces to each axis of the product. Combining subsystem information, it generates measurement result data, which can be visualized.

[0085] In some embodiments, the vision measurement system includes five cameras and a support. Four cameras are evenly distributed around the rotation axis of the motion platform 1 and fixed to the support, while the other camera is fixed directly above the rotation axis. This allows for the measurement of the product's contour linear dimensions in five directions: front, back, left, right, and top. The five cameras are distributed and fixed on the support, which is then fixed to the base platform 6 as a whole. This independent support structure ensures that the motion platform 1 is not disturbed during movement and the loading device is in operation, thereby guaranteeing the accuracy of the image vision measurement.

[0086] The lateral and vertical loading drive devices are electric cylinders. These devices are fixed to the support frame. To save space, electric cylinders are used for loading in this design. Two electric cylinders perform lateral and vertical loading respectively; the cylinders are fixed to the support frame and connected to the product under test (3) via tooling.

[0087] In some embodiments, the control unit is used to acquire the starting current data of the motor of the product under test, and to obtain the frictional torque of the product under test under different axial clearances based on the starting current data, so as to evaluate the optimal control range of the axial clearance of the product under test.

[0088] If the average starting current is in the "lower range" (indicating low frictional torque); if the starting current fluctuation is ≤ the threshold (e.g., if the threshold is 5%, it indicates stable frictional torque and no shaft movement); if the axial clearance range that meets the above two conditions is the optimal control range (e.g., 0.03mm-0.06mm).

[0089] The motion platform 1 uses three servo motors driven by reducers to drive lead screws, achieving linear motion in the X, Y, and Z directions. High-precision linear grating rulers are added in the X, Y, and Z directions to achieve a closed-loop linear position movement, reaching micron-level linear displacement positioning accuracy. The axis rotating around the Z-axis is directly driven by a torque motor, with a Renishaw circular grating added as an angular position feedback element to achieve high-precision positioning of the roll axis. The position servo control algorithm is integrated into the driver, and the four axis electrical limit switches are connected to their respective drivers. The drivers internally implement positive and negative limit functions to ensure that the movement of each axis does not exceed the limit range. The measurement and control system uses a CAN bus to send position commands, acquire axis positions, and read back driver status.

[0090] The dual-axis constant force loading device can apply a specific force in both directions along the axis of rotation of the workpiece under test. Its function is achieved by connecting the rotating shaft end of the workpiece under test to the tension and compression sensors through a tooling, and then connecting the motor through a structural component. The motor applies tension and compression to the pitch and roll shaft ends of the platform under test. The tension and compression sensors provide feedback on the force. The measurement and control system collects the tension and compression data and controls the loading motor to load, thereby changing the force on the shaft under test and realizing a constant force loading closed loop. At the same time, the visual image measurement system accurately measures the displacement or structural contour changes and measures the axial clearance of the platform.

[0091] The motor in this invention adopts a three-loop feedback control method that combines position loop, speed loop and current loop. The control algorithm uses a feedforward algorithm to make the control effect more stable and accurate.

[0092] This invention proposes to use composite control PID regulation to achieve the required performance indicators. ① A closed loop consisting of current feedback and encoder feedback; ② Composite control consisting of position, velocity, and acceleration.

[0093] In control systems, the commonly used control law is PID control. The principle of a PID controller control system is shown in the figure. It mainly consists of a PID controller and the controlled object. A PID controller is a linear controller that uses a linear combination of the proportional (P), integral (I), and derivative (D) values ​​of the deviation between the given value rin(t) and the actual output value yout(t) to form the control quantity, thereby controlling the controlled object.

[0094] Differential equation of PID controller:

[0095]

[0096] In the formula

[0097] PID controller transfer function:

[0098]

[0099] The control system includes a current loop, a speed loop, and a position loop. In a motor drive system, current-limiting feedback is used to limit the maximum current. This current-limiting effect only exists during startup or stall conditions. Once the current exceeds a certain specified maximum value, current negative feedback is activated, causing the static characteristics to "soften" drastically. As the load current increases, the motor speed continuously decreases. When the load current increases to a certain value (i.e., stall current), the motor stops rotating (i.e., stall). During normal operation, current-limiting feedback is ineffective, the static characteristics are relatively rigid, and the current changes freely with the load. However, relying solely on current-limiting feedback to limit the inrush current during startup and acceleration still results in unsatisfactory startup performance. This is because current-limiting feedback can only limit the maximum current. During the transition process, the current is constantly changing. After reaching the maximum value, due to the introduction of negative feedback and the increase in motor back EMF, the current is suppressed again, and the motor torque decreases accordingly. This prolongs the startup and acceleration process, thus failing to guarantee the system's rapid response. Therefore, to fully utilize the motor's permissible overload capacity, it is best to maintain the current at its maximum value during the transition process, enabling the system to withstand the largest possible acceleration (deceleration) during starting (braking) and achieve a rapid dynamic response. Upon reaching steady-state speed, the current should immediately decrease, balancing the motor's electromagnetic torque with the load torque, thus transitioning to steady-speed operation. Ideally, the starting or braking current waveform should be a square waveform with a linear change in speed. However, in actual systems, due to the inductance of the main circuit, the current cannot change abruptly; the ideal waveform can only be approximated. From a control perspective, the key is to obtain a constant current process that maintains the current at its maximum value. Furthermore, current negative feedback control is essentially acceleration control, so the role of the current loop should not be limited to safety protection but should also be designed and used from a control perspective.

[0100] To reduce the impact of friction, the open-loop gain of the system must be increased because, within the system's stable range, a higher open-loop gain reduces the system's sensitivity to disturbances (considering friction as a disturbance). On the other hand, if the structural stiffness of each axis of the system is not sufficiently high, its dynamic characteristics will exhibit multiple vibration modes, resulting in some zeros in the right half-plane, making the system a non-minimum-phase system. Therefore, if a traditional control strategy is used to control the platform, the open-loop gain cannot be very large due to the zeros in the right half-plane. The motor model has minimum phase, and under stable system conditions, the open-loop gain can be sufficiently large by selecting an appropriate controller. Intuitively, a high gain in the inner loop controlling the motor position will make the system insensitive to friction, which is beneficial for the outer loop design controlling the platform position. Since the outer loop model is a non-minimum-phase system, its open-loop gain cannot be very high. However, since the impact of friction is almost eliminated by the inner loop, a relatively low open-loop gain in the outer loop is not a major concern. The above intuitively explains why a robust control strategy using inner and outer loops is applied to control each axis of the system. The above analysis shows that introducing speed feedback improves the control and dynamic characteristics of the actuator, which is beneficial for suppressing interference, overcoming friction dead zones, reducing torque fluctuations, and widening the system bandwidth. Furthermore, the introduction of the speed loop reduces the pressure on the position loop, allowing it to have higher servo stiffness, static accuracy, and better dynamic characteristics.

[0101] In servo system control, once the speed loop is determined, it is treated as a transfer function of the position loop. The output of the position controller serves as the input to the speed loop, and the position loop gain is one of the most important parameters in a DC servo system. The selection of the position loop gain must consider the influence of many factors. Generally, a higher position loop gain results in a smaller position tracking error, but when the input speed changes abruptly, the output will change drastically. Furthermore, higher gains lead to lower stability, while lower gains improve stability but increase the tracking error. Therefore, selecting an appropriate position loop gain is crucial for the entire control system.

[0102] The multi-target machine vision image measurement system contains multiple cameras that can cover five directions of the product under test: front, back, left, right, and top. It can perform non-contact test measurements on the user's stable platform from all angles. Its external data interaction is via Ethernet, uploading the measured data and 3D images to the industrial control computer for system analysis.

[0103] The control unit is used to perform the shaft error measurement method as described in any of the following embodiments.

[0104] Based on the same inventive concept, this disclosure also provides a shaft system error measurement method, which is the shaft system error measurement device provided in any of the preceding embodiments; the description of the shaft system error measurement device can be found in the preceding embodiments, and will not be repeated here.

[0105] The shaft system error measurement method includes:

[0106] When a measurement command for axial clearance is received, perform the following steps:

[0107] The constant force loading component is controlled to actively load each axis of the product under test 3.

[0108] During the loading process, the vision measurement system is controlled to acquire feature images of the product under test 3 according to different loading loads;

[0109] Extract key feature points of the platform under test from the image of the product under test, and calculate the three-dimensional spatial coordinates of each feature point;

[0110] Calculate the coordinate difference of the same feature point under different working conditions during loading to obtain the axial displacement data of the shaft system;

[0111] Axial clearance measurement is achieved by performing multidimensional correlation analysis between axial displacement data and corresponding loading parameters.

[0112] In some embodiments, the shaft system error measurement method of the present invention further includes:

[0113] Upon receiving the first orthogonal error test command, execute the following steps:

[0114] The first command is issued to the product under test to control the azimuth axis of the product under test to remain stationary, and to control the pitch axis of the product under test to rotate by multiple (e.g., n angles, n≥3, with each rotation angle evenly distributed) angles. A crosshair target assembly 7 is installed on one side of the pitch axis of the product under test.

[0115] The vision measurement system is triggered to acquire target images at various angle positions on the pitch axis. The acquired target images are processed to obtain the target crosshair center coordinates corresponding to each angle position. The center coordinates of the target crosshair center at multiple angle positions are fitted to obtain the first center coordinate.

[0116] A second command is issued to the product under test to control the pitch axis of the product under test to remain stationary and to control the azimuth axis of the product under test to rotate 180°.

[0117] A third command is issued to the product under test to control the azimuth axis of the product under test to remain stationary and to control the pitch axis of the product under test to rotate multiple (e.g., n angles, n≥3, with each rotation angle evenly distributed) angles. At each angle position of the pitch axis, the vision measurement system is triggered to acquire target images. The acquired target images are processed to obtain the target crosshair center coordinates corresponding to each angle position. The target crosshair center coordinates corresponding to multiple angle positions are fitted with the center of the circle to obtain the second circle center coordinates.

[0118] Calculate the coordinate deviation between the coordinates of the first center and the coordinates of the second center, and obtain the orthogonal error of the axis system of the tested product based on the coordinate deviation.

[0119] This invention obtains the angular error based on the coordinate deviation between the coordinates of the first and second center circles, and evaluates the orthogonality error of the typical platform axis system based on the angular error. If the angular error is within a set range, i.e., less than a set threshold, it indicates that the azimuth and pitch axes are orthogonal; if the angular error is greater than the set threshold, it indicates that there is an orthogonality error. The larger the error value, the worse the axis system orthogonality, and calibration is required.

[0120] This invention uses the least squares method to fit a circle to a set of planar points. The core of this method is to obtain the circle's parameters (center coordinates, radius) by solving a system of linear equations, and then calculate the roundness (fitting accuracy). The detailed process is as follows:

[0121] Step 1: Input parameter validity check

[0122] The function accepts two input parameters: double pts[][2] (the two-dimensional point set to be fitted, each row is the X and Y coordinates of a point) and size_t ptsSize (the number of points in the point set).

[0123] First, check if the number of points is less than 3: circle fitting requires at least 3 non-collinear points. If the number of points is less than 3, directly set the state of the fitting result circleFitResult:

[0124] The success flag for fitting is set to false.

[0125] Roundness = 0.0;

[0126] Set all circle parameters (center X, center Y, radius) to 0;

[0127] Return the invalid result directly.

[0128] Step 2: Accumulate and calculate the point set statistics

[0129] If the number of points is greater than or equal to 3, iterate through all points and accumulate the various statistical sums required for subsequent solutions (to simplify formula derivation, aggregate the coordinate powers and product terms of the points in advance):

[0130] Calculate the sum of the X coordinates of all points, sumX1. The formula (single-step accumulation) is: sumX1 += x (where x is the X coordinate of the current point).

[0131] Calculate the sum of the Y coordinates of all points, sumY1. The formula (single-step accumulation) is: sumY1 += y (where y is the Y coordinate of the current point).

[0132] Calculate the sum of squares of the X coordinates of all points, sumX2. The formula (single-step accumulation) is: sumX2 += x*x;

[0133] Calculate the sum of squares of the Y coordinates of all points, sumY2. The formula (single-step accumulation) is: sumY2 += y*y;

[0134] Calculate the sum of the cubes of the X coordinates of all points, sumX3, using the formula (single-step accumulation): sumX3 += x*x*x;

[0135] Calculate the sum of the cubes of the Y coordinates of all points, sumY3. The formula (single-step accumulation) is: sumY3 += y*y*y;

[0136] Calculate the sum of the X*Y products of all points, sumX1Y1, using the formula (single-step accumulation): sumX1Y1 += x*y;

[0137] Calculate the sum of the products of X*Y² for all points, sumX1Y2. The formula (single-step accumulation) is: sumX1Y2 += x*y*y;

[0138] Calculate the sum of the products of X²*Y for all points, sumX²Y1. The formula (single-step accumulation) is: sumX²Y1 += x*x*y.

[0139] The specific code is as follows:

[0140] doublesumX1 = 0.0;

[0141] doublesumY1 = 0.0;

[0142] doublesumX2 = 0.0;

[0143] doublesumY2=0.0;

[0144] doublesumX3 = 0.0;

[0145] doublesumY3 = 0.0;

[0146] doublesumX1Y1=0.0;

[0147] doublesumX1Y2=0.0;

[0148] doublesumX2Y1=0.0;

[0149] constdoubleN=(double)ptsSize;

[0150] for(size_ti=0;i<ptsSize;++i)

[0151] {

[0152] doublex=pts[i][0];

[0153] doubley=pts[i][1];

[0154] doublex2=x*x;

[0155] doubley2=y*y;

[0156] doublex3=x2*x;

[0157] doubley3=y2*y;

[0158] doublexy=x*y;

[0159] doublex1y2=x*y2;

[0160] doublex2y1=x2*y;

[0161] sumX1+=x;

[0162] sumY1+=y;

[0163] sumX2+=x2;

[0164] sumY2+=y2;

[0165] sumX3+=x3;

[0166] sumY3+=y3;

[0167] sumX1Y1+=xy;

[0168] sumX1Y2+=x1y2;

[0169] sumX2Y1+=x2y1;

[0170] }

[0171] Step 3: Construct the coefficients of the linear equation system

[0172] Based on the accumulated statistics, calculate the intermediate coefficients required for least-squares fitting (corresponding to the coefficients of the system of equations after linearization of the circular equation):

[0173] C = N*sumX2 - sumX1*sumX1;

[0174] D = N*sumX1Y1 - sumX1*sumY1;

[0175] E=N*sumX3+N*sumX1Y2-(sumX2+sumY2)*sumX1;

[0176] G = N * sumY2 - sumY1 * sumY1;

[0177] H=N*sumX2Y1+N*sumY3-(sumX2+sumY2)*sumY1;

[0178] Where N is the number of points (the floating-point conversion value of ptsSize).

[0179] Step 4: Solve for the linearization parameters of the circle equation

[0180] The general equation of a circle is: x² + y² + ax + by + c = 0. The coefficients a, b, and c are solved by least squares.

[0181] Calculate the denominator: denominator = C * GD * D (avoid division by zero; if it is 0, the fitting will fail. This is not explicitly handled in the code, so please pay attention).

[0182] Solve for the coefficient a: a = (H * DE * G) / (denominator);

[0183] Solve for the coefficient b: b = (H * CE * D) / (-denominator);

[0184] Solve for the coefficient c: c = -(a*sumX1 + b*sumY1 + sumX2 + sumY2) / N.

[0185] Step 5: Calculate the center coordinates and radius of the circle based on the coefficients a, b, and c.

[0186] Convert the coefficients of the general equation into standard parameters of the circle (center coordinates, radius), and store them in the result structure:

[0187] Calculate the X coordinate of the circle center: circleParam[0] = -0.5*a;

[0188] Calculate the Y-coordinate of the circle center: circleParam[1] = -0.5 * b;

[0189] Calculate the radius: circleParam[2]=0.5*sqrt(a*a+b*b-4*c) (sqrt is the square root, and the square root must be ≥0, otherwise there is no real solution, which is not explicitly handled in the code);

[0190] Set the success flag: success = true.

[0191] Step 6: Calculate roundness (fitting accuracy)

[0192] Roundness reflects the degree to which the point set fits the fitted circle; the smaller the value, the higher the fitting accuracy.

[0193] For each point, calculate the difference between the distance from that point to the center of the fitted circle and the fitted radius (residual): Residual = sqrt[(center X - point X)² + (center Y - point Y)²] - radius;

[0194] The specific code is as follows:

[0195] round.push_back((sqrt(pow((cFR.circleParam[0]-pts[i][0]),2)+pow((cFR.circleParam[1]-pts[i][1]),2)))-cFR.circleParam[2]), where pts[i][0] is the X coordinate of the i-th original point, pts[i][1] is the Y coordinate of the i-th original point; cFR.circleParam[0] is the X coordinate of the center of the fitted circle (obtained by the least squares method); cFR.circleParam[1] is the Y coordinate of the center of the fitted circle, and cFR.circleParam[2] is the radius of the fitted circle;

[0196] Sum the absolute values ​​of all residuals to obtain the total residual sum, roundSum;

[0197] Roundness = Total residuals ÷ Number of points (mean residuals).

[0198] Step 7: Return the fitting results

[0199] Returns a circleFitResult structure containing "fit success flag, circle parameters (center X / Y, radius), and roundness".

[0200] Algorithm basis: The general equation of a circle is linearized, and the overdetermined system of equations is solved by the least squares method, which is suitable for fitting point sets without significant exterior points;

[0201] The significance of roundness: The smaller the average residual, the closer the point set is to the ideal circle.

[0202] The product under test can be, but is not limited to, a three-axis stabilized platform, including pitch axis, roll axis and azimuth axis.

[0203] Three-axis stabilization platform: In addition to the two-axis system, an azimuth axis is added, which can simultaneously compensate for three types of disturbances of the carrier: pitch, roll and yaw. It has higher stabilization accuracy and is suitable for complex carrier scenarios such as UAVs and airborne reconnaissance equipment.

[0204] The shaft system error measurement method for stabilization platforms designed in this invention is mainly used for measuring the structural shaft system errors of a positioning scale stabilization platform. Using a non-contact measurement method, it achieves spatial position measurement and orthogonal error measurement of the mechanical shafts of the stabilization platform. Furthermore, by applying bidirectional constant force to the measured shaft system, the axial clearance can be measured, and the shaft system stiffness can be evaluated through axial displacement, providing a basis for precise structural adjustment. In addition, it can also be used to measure the geometric dimensions of typical contour parts.

[0205] In some embodiments, when a first orthogonal error test command is received, the motion platform 1 is first controlled to move the product under test to send the target into the field of view of the vision measurement system, triggering the vision measurement system to acquire the target image and perform image processing to obtain the center coordinates of the first crosshair.

[0206] Then, the first command is issued to the product under test: keep the azimuth axis stationary, rotate the pitch axis at multiple angles, and repeatedly trigger the vision measurement system to acquire target images and perform image processing to obtain the center coordinates of multiple first crosshairs. (n is not less than 3);

[0207] The coordinates of the first circle center are obtained by fitting the center coordinates of the first crosshair.

[0208] In some embodiments, a second command is issued to the product under test to control the pitch axis of the product under test to remain stationary and to control the azimuth axis of the product under test to rotate 180°, thereby triggering the vision measurement system to acquire target images and perform image processing to obtain the center coordinates of the second crosshair.

[0209] Then, a third command is issued to the product under test: keep the azimuth axis stationary, rotate the pitch axis at multiple angles, and repeatedly trigger the vision measurement system to acquire target images and perform image processing to obtain the center coordinates of multiple second crosshairs. (n is not less than 3);

[0210] The coordinates of the second center of the crosshair are obtained by fitting the center coordinates of the second crosshair.

[0211] In some embodiments, triggering a vision measurement system to acquire target images at various angle positions along the pitch axis, and performing image processing on the acquired target images to obtain the target crosshair center coordinates corresponding to each angle position includes: triggering a single-camera vision measurement system or a multi-camera vision measurement system to acquire target images at various angle positions along the pitch axis; when triggering a multi-camera vision measurement system to acquire target images, performing image processing on the target images acquired by each camera to obtain the target crosshair center coordinates corresponding to each camera; and fusing the target crosshair center coordinates corresponding to multiple cameras to obtain the target crosshair center coordinates corresponding to each angle position.

[0212] In some embodiments, the target crosshair center coordinates corresponding to multiple cameras are fused, specifically including: fusing the target crosshair center coordinates corresponding to multiple cameras using the least squares method. The optimal coordinate values ​​are solved with the objective of minimizing the sum of squared residuals between each set of coordinates and the fused coordinates, and finally, the three-dimensional physical coordinates of the target center corresponding to each angular position are output.

[0213] In some embodiments, image processing is performed on the acquired target image to obtain the center coordinates of the target crosshair at various angle positions, specifically including:

[0214] Binarize the target image;

[0215] Edge extraction and fitting are performed on the crosshairs of the binarized target image to determine the center of the crosshairs;

[0216] The camera distortion parameters obtained from the pre-measurement calibration are used to correct the distortion of the extracted crosshair center pixel coordinates.

[0217] By combining the distance information collected by the laser ranging module, the three-dimensional coordinates of the center point of the crosshairs in the camera coordinate system are calculated.

[0218] The coordinate transformation matrix obtained through calibration is used to convert the three-dimensional coordinates of the crosshair center point in the camera coordinate system to the three-dimensional coordinates of the crosshair center point in the measurement coordinate system.

[0219] join Figure 10 The visual measurement module 4 includes a camera 41 and a light source 42. The camera is used to acquire images of the product 3 under test. At least one laser ranging module is installed on the visual measurement module 4. The camera, light source, and laser ranging module 43 are fixedly mounted on the camera mounting base. This invention corrects lens distortion through a high-precision intrinsic parameter calibration algorithm. Based on a high-precision planar target, an arbitrary pose planar target parameter calibration method can be used to easily calibrate the intrinsic parameters. The optical axis of the laser ranging module is parallel to the visual axis of the camera.

[0220] The extrinsic parameters are the coordinate system parameters of the binocular vision sensor and the positional relationship parameters between the two cameras. Typically, the coordinate system parameters between the binocular cameras are derived by adjusting and inverting a series of calibration field control points. This invention proposes to use a calibration ruler containing a standard distance as the calibration target, and to incorporate the reference distance into the calculation to achieve high-precision calibration.

[0221] The target's motion speed is 1-10 mm / s for translation and 0.05° / s for rotation. If the cameras are not synchronized, or if there is an exposure time, the target's motion will cause positional errors. The camera selected in this solution has a frame rate of 60 frames per second, meeting the system's 10Hz measurement requirement. The exposure time can be less than 1 ms under normal lighting conditions. If the translation is calculated at 10 mm / s, the object moves m7 = 0.01 mm during each exposure; for rotation at 0.05° / s, the rotation error is 0.0005° / s, which is negligible. Furthermore, camera synchronization also causes differences in target points. However, due to the use of a synchronous trigger acquisition module (which synchronously triggers multiple cameras), the synchronization accuracy can reach the microsecond level, therefore this is also negligible.

[0222] N (N≥3) target points are evenly distributed on the surface of the target being measured. The distribution of the target points should meet the following requirements:

[0223] 1) It should be distributed as evenly as possible on the surface of the object;

[0224] 2) During the movement and rotation of the object, ensure that the binocular vision measurement system can measure at least 3 target points;

[0225] 3) To ensure the accuracy of attitude measurement, the distance between target points should not be too small.

[0226] In the initial state, the three-dimensional coordinates of each target point are obtained using the principle of binocular vision measurement. where i=1,2, N. Establish an object frame using three initial points P1, P2, and P3 (other suitable points can be selected according to the actual situation).

[0227] like Figure 9 As shown, suppose points P1, P2, and P3 are not on a straight line. Generate a new coordinate system from these three points. In the new coordinate system, the coordinates of these three points are respectively , and ,and >0, >0, the scaling factor remains unchanged.

[0228] Then the object coordinate system P1- The origin parameters are (X1, Y1, Z1), and its three rotation parameters (εz, εy, εx) are solved using the least squares principle:

[0229]

[0230] Coordinate system P1- Once the seven transformation parameters (X1, Y1, Z1, εz, εy, εx, 1) are determined, the target points on the surface of the object can be measured. All coordinates are transformed to the object coordinate system (the transformation principle is omitted). .

[0231] In actual measurements, due to the rotation and movement of the object, the point number measured by the binocular vision measurement system cannot be determined. It is assumed that any three points are measured at time Ti. , , Then, by using the transformation parameters (X1, Y1, Z1, εz, εy, εx, 1), we can transform to the initial object-space coordinate system and reconstruct the object-space coordinate system Pi at time Ti. .

[0232] At this point, the initial object-space coordinate system P1- and the object-space coordinate system Pi at time Ti The difference is the change in position and orientation of the object in space.

[0233] As can be seen from the above principle, the position change accuracy of the object coordinate system is the three-dimensional coordinate accuracy of the measurement point. At this time, the attitude change accuracy of the object coordinate system is mainly affected by the point measurement error and the point distance.

[0234] The spatial measurement range of the equipment is 300×300×300mm. Assuming the maximum size of the object being measured is 300mm, and the positional accuracy is... If the error is 0.0033 mm, then the axial attitude error of the object coordinate system measured at any time is:

[0235]

[0236] The three-dimensional coordinate system is constructed by determining two axis systems and then using the cross product to determine the third axis. Therefore, the overall axial error of the coordinate system is:

[0237]

[0238] Similarly, under the premise of multi-point fusion data processing, it is theoretically possible to satisfy... The requirements are that, taking into account other influencing factors such as vibration, there is a certain margin, which can still meet the requirements of system orthogonality.

[0239] Based on the principle of pinhole imaging, assuming For the optical center of the camera, The focal length of the camera is half the horizontal physical length of the object surface. Half of the horizontal physical length of the image plane is The distance between the target and the camera is .

[0240] The ratio between camera chip size and field of view is obtained as follows:

[0241]

[0242] According to the requirement that the spatial measurement range be no less than 300×300×300mm, it is known that when the working distance... It is 300mm, and the horizontal physical length of the object surface is... Physical length perpendicular to the object surface When both are 300mm, the camera sensor size and focal length should be equal. Common camera sensor sizes are typically 3-16mm in one direction, and the focal length should also be 3-16mm. If the working distance... As the size increases, the focal length can also increase accordingly.

[0243] According to the requirements, the resolution of linear dimension measurement of typical contours should not be greater than 3um. Assuming the camera focal length is 15mm, when the working distance is 300mm, the size of a single pixel should not be greater than 0.15um. However, commonly used pixel sizes include 1.67um, 2um, 2.2um, 2.4um, 3.45um, 4.8um, and 5.5um. Therefore, a single pixel size of not more than 0.15um is impractical.

[0244] In summary, conventional cameras and lenses cannot simultaneously meet the above requirements. To satisfy both conditions at the same time, it is necessary to reduce the resolution of typical contour linear dimension measurement or reduce the spatial measurement range. Specifically, if the resolution of typical contour linear dimension measurement is r, then:

[0245]

[0246]

[0247] According to the above formula, we can obtain:

[0248]

[0249] To simultaneously satisfy the requirements of a spatial measurement range of no less than 300×300×300mm and a typical contour linear dimension measurement resolution of no more than 3µm, the resolution in a single direction must be no less than 100,000 pixels, and the camera resolution must be no less than 1,000,000MP, which is unrealistic.

[0250] If the camera resolution res is 5328×4608px, the horizontal and vertical pixel size S is 2.74um, the lens focal length f is 50mm, and the working distance D is 200mm, then the measured resolution is:

[0251]

[0252] The camera's field of view in the x-direction is:

[0253]

[0254] In this case, a small target can be used to measure the orthogonality error of a typical platform axis system. Specifically, a crosshair target (e.g., 25.4 mm in diameter) is used, which is actively emitting light and fixed to one side of the turntable's pitch axis. In the image processing stage, a sub-pixel subdivision method based on Zernike moments can be used for edge detection, ensuring that the measurement resolution is no greater than 3 μm.

[0255] In summary, when measuring the orthogonality error of a typical platform axis system, a small target can be used.

[0256] Because cameras have distortion errors, only calibrated cameras can be used for measurement, and the accuracy of camera calibration directly affects the point accuracy of the photogrammetric system. For digital photogrammetric cameras, the main factors interfering with imaging are radial distortion, eccentric distortion, image plane distortion, and scaling and orthogonal distortion within the image plane. Furthermore, inaccurate internal orientation elements (x0, y0, f) can interfere with the collinearity equation. Therefore, the image point coordinate errors caused by these internal parameters are systematic and are thus called systematic image point errors.

[0257] The intrinsic parameters of a camera are currently mainly described using a 10-parameter model, which includes: radial distortion parameters K1, K2, and K3; eccentric distortion parameters P1 and P2; image plane distortion parameters b1 and b2; principal point deviation (x0, y0); and camera focal length f. The 10-parameter camera model is a physical model, and each parameter has a clear physical meaning.

[0258] Currently, there are many types of methods for calibrating camera intrinsic parameters, among which the commonly used ones include the self-calibrated bundle adjustment model, the Zhang Zhengyou model, and the RAC model. The Zhang Zhengyou model belongs to the planar calibration method. This model uses a planar calibration plate with regularity information and is a method between traditional calibration methods and self-calibration methods, with high calibration accuracy.

[0259] In some embodiments, the planar calibration method employs a two-step calculation process:

[0260] (1) Initialize the distortion parameters to zero, select a point with small distortion near the center of the image, and substitute its world coordinates and image coordinates as known conditions to solve the camera's intrinsic and extrinsic parameters using the collinearity equation;

[0261] (2) Solve for the camera's intrinsic and extrinsic parameters using optimization algorithms;

[0262] (3) Using the solution results as the initial values ​​of the internal and external parameters, select all points on the plane template, establish an optimization model, and solve for the optimal values ​​of the internal parameters, external parameters, and distortion parameters.

[0263] Define the template plane as lying in the world coordinate system Z=0, and let K be the intrinsic parameter matrix of the camera. Let be the homogeneous coordinates of a point on the template plane. Let be the homogeneous coordinates of the points on the template plane projected onto the corresponding points on the image plane. t and t are the rotation matrix and translation vector of the camera coordinate system relative to the world coordinate system, respectively.

[0264] Using the collinearity equation from photogrammetry, we can obtain the following equation:

[0265]

[0266] make , Since r1 and r2 are two unit orthogonal vectors, we have and Two constraints can be obtained for each image:

[0267]

[0268]

[0269] Therefore, when there are enough images, the camera's intrinsic parameter matrix K can be solved, thereby achieving camera calibration.

[0270] The vision measurement system needs to be fixed on the measurement bracket 5. Before the system starts measuring, the relationship between the vision measurement system coordinate system and the workpiece initial coordinate system needs to be calibrated, and this needs to be done in real time during measurement. The coordinate systems are defined as follows:

[0271] Coordinate system for visual measurement system:

[0272] The coordinate system of a visual measurement system refers to the coordinate system of the visual measurement system itself, which is the inherent coordinate system of the camera. The coordinate system transformation relationship between cameras can be obtained by measuring the feature target.

[0273] The workpiece coordinate system can be defined at specific locations on the workpiece according to user requirements. Multiple targets are attached to each surface of the workpiece; the workpiece coordinate system can be defined by specifying the positions of these targets. For example, by specifying targets P1, P2, and P3 on the workpiece, the position and orientation of the workpiece coordinate system can be defined using these three points. (P1 is the origin of the coordinate system, the line connecting P1 and P2 is the +X axis, the vector product of P1P2 and P1P3 is the +Z axis, and the +Y axis is determined according to the right-handed coordinate system.) Figure 6 As shown.

[0274] The coordinates of other targets in the workpiece coordinate system can be measured using high-precision measuring equipment such as laser trackers to determine their specific positions in the workpiece coordinate system.

[0275] In a vision measurement system, the coordinates of each target can be obtained by identifying and locating them using a camera. By combining the target's coordinates in the camera's measurement coordinate system with its coordinates in the workpiece's coordinate system, the coordinate transformation relationship between the workpiece and measurement system coordinate systems can be solved using the least squares transformation principle. This yields the real-time pose of the workpiece within the measurement system. The coordinate transformation matrix includes the rotation matrix R and the translation matrix T.

[0276] The coordinate transformation calibration algorithm is as follows:

[0277] The system determines the transformation relationship between two coordinate systems by measuring the corresponding coordinates of points in different coordinate systems using a common coded point. Solving the coordinate system transformation relationship is essentially solving a least-squares problem. Let the two coordinate systems be M1 and M2, and the coordinate sets of the corresponding points in the two coordinate systems be {P}. i}, {P i '}; i=1,2,…,n. P i =(x i , y i , z i T, P i '=(x i ',y i ',z i Then T has:

[0278]

[0279] In the formula N i Given the error vector, establish the objective function:

[0280]

[0281] The optimal solution for the transformation matrix between the two coordinate systems can be obtained when the objective function reaches its minimum value.

[0282] There are several common methods for solving transformation and translation matrices. The method used in this system is a distributed solution method based on the least squares principle for solving transformation and translation matrices.

[0283] The step-by-step solution method transforms the problem of solving rotation and translation matrices into a method of first solving the rotation matrix and then solving the translation matrix. Given two coordinate systems M1 and M2, and point P... i The vectors in the two coordinate systems are p i q i Then we have:

[0284]

[0285] Suppose there are n common points. Theoretically, if we don't consider errors, the coordinates of each point in coordinate system M1 after coordinate transformation should coincide with the coordinates of the points in coordinate system M2. However, due to the influence of various factors during the measurement process, errors are inevitable. Therefore, we assume that the centroids of the two rigid bodies formed by the data points in the two coordinate systems coincide after coordinate transformation.

[0286]

[0287] Then we have:

[0288]

[0289] From the above two equations, we can obtain:

[0290]

[0291] make

[0292] Then establish the least squares function:

[0293]

[0294] Depend on After sorting, we can obtain:

[0295]

[0296] Transposing both sides of the equation, we get:

[0297]

[0298] The above formula can be expressed as:

[0299]

[0300] Matrix M is a diagonalizable real symmetric matrix, and R is a rotation transformation matrix, therefore it satisfies the following condition:

[0301]

[0302] Therefore, performing singular value decomposition on the square matrix N yields:

[0303]

[0304] Calculate matrix

[0305] If |X|=1, then R=X; otherwise, compute the singular value decomposition of NT. ,calculate If the determinant of matrix R is 1, then the solution for the rotation matrix is ​​found; otherwise, the algorithm fails (this situation generally does not occur).

[0306] After obtaining the rotation matrix R, the translation matrix T can be obtained:

[0307] .

[0308] In some embodiments, edge extraction and fitting of the crosshairs in the binarized target image to determine the center of the crosshairs includes: extracting the horizontal bright line edge and the vertical bright line edge of the crosshairs in the binarized target image respectively to obtain two sets of continuous edge point coordinates;

[0309] Linear fitting is performed on the two sets of edge points of the horizontal bright line to obtain the equations of the upper and lower edge lines of the horizontal bright line. The perpendicular bisectors of the two horizontal edge lines are calculated to obtain the horizontal center line of the crosshair.

[0310] Linear fitting is performed on the two sets of edge points of the vertical bright line to obtain the equations of the left and right edge lines of the vertical bright line. The perpendicular bisectors of the two vertical edge lines are calculated to obtain the vertical center line of the crosshair.

[0311] The intersection of the horizontal center line and the vertical center line is the center of the crosshairs.

[0312] In some embodiments, the horizontal and vertical bright line edges of the crosshairs in the binarized target image are extracted to obtain two sets of continuous edge point coordinates. Specifically, this includes: first, finding candidate edge points of the crosshairs through edge detection.

[0313] Calculate the ideal edge parameters within the coarse-edge window;

[0314] The edge candidate points are corrected using the ideal edge parameters to obtain the edge point coordinates.

[0315] Edges exist between objects and between objects and the background. Edges are one of the most fundamental features of an image, containing rich intrinsic information (such as shape, orientation, or step properties). They are crucial features relied upon for pattern recognition, image registration, image segmentation, and image classification. Accurate edge detection is of great significance for image measurement, analysis, and recognition. Edge detection is an important aspect of image processing. Boundaries are the most basic features of an image. A boundary refers to the set of pixels whose surrounding pixels exhibit a step-like or roof-like change in grayscale. Edge detection is the foundation and key to improving image detection accuracy; the accuracy of edge localization directly determines the quality of the detection system. To improve image detection accuracy, a sub-pixel edge point localization method based on Zernike moments is adopted.

[0316] The principle of the Zernike moment subpixel edge detection algorithm is to calculate the four edge parameters required for edge detection based on the rotation invariance of Zernike moments, thereby achieving precise edge localization. The four parameters for determining the edge include the background grayscale value. Step height The vertical distance from the center of the disk to its edge and perpendicular line Angle between axes A Zernike moment before an image is rotated. With the rotated Zernike moment The relationship is .set up For the rotated image, we have:

[0317]

[0318] Figure 7 This is an ideal edge model diagram. The straight line L enclosed by the unit circle represents the ideal edge, with gray values ​​h and h+k on either side of line L, where k is the gray-level difference and l is the perpendicular distance from the origin to the edge. Let l be the angle between the l-axis and the x-axis.

[0319] When determining the edge parameters, three Zernike moments are required, namely... , and Let's calculate the complex numbers corresponding to each of them. , , Since the moment remains invariant before and after rotation, we have:

[0320]

[0321] Due to rotation The image after the angle is symmetrical about the x-axis, so If the imaginary part is equal to 0, then we have:

[0322]

[0323] In the formula, for The imaginary part, for The imaginary part. Then we have:

[0324]

[0325] Therefore, it can be further deduced that the Zernike moments of each order after image rotation are:

[0326]

[0327]

[0328]

[0329] Finally, using the above formula, we can derive the four parameters of the ideal edge as follows:

[0330] .

[0331] In some embodiments, the three-dimensional coordinates of the crosshair center point in the camera coordinate system are calculated by combining the distance information collected by the laser ranging module, specifically including:

[0332] The corrected crosshair center pixel coordinates are converted to normalized coordinates in the camera coordinate system (x... corr ,y corr );

[0333] Based on the distance information acquired by the laser ranging module, the three-dimensional coordinates (x, y, z) in the camera coordinate system are calculated. c ,y c ,z c ), where x c =x corr *z c y c =y corr *z c , z c The depth value of the feature point in the camera coordinate system is obtained by combining the laser ranging value with the calibration parameters;

[0334] The corrected crosshair center pixel coordinates are converted to normalized coordinates in the camera coordinate system, and the calculation formula is: x corr =(u corr- c x ) / f x ,y corr=(v corr- c y ) / f y c x f y c y f y It is a core parameter of the camera intrinsic parameter matrix K, u corr v corr These are the pixel coordinates of the center of the crosshair after correction.

[0335] z c =Laser ranging value * cosθ, where θ is the angle between the laser beam and the camera optical axis. This parameter is known after prior calibration.

[0336] Image analysis is performed on the collected images of the product under test to obtain axial displacement data during the loading process. Based on the axial displacement data, the axial clearance of the product under test is measured.

[0337] The intersection of two axes is measured using the roll axis ( Figure 3 Using axis 2 as a reference, rotate the roll axis and measure the pitch axis using a multi-view machine vision image measurement system. Figure 3 The following are examples of measurement methods for the contour variation of the characteristic point at the end of axis 1) in the X direction, and the intersection error L between the two axes: Figure 3 As shown. During the two-axis intersection test, the pitch axis of the test piece rotates in opposite directions, and three cameras measure the displacement at three positions: the pitch axis system, the azimuth axis system, and the base.

[0338] In some embodiments, the shaft system error measurement method of the present invention further includes:

[0339] Upon receiving the second orthogonal error test command, execute the following steps:

[0340] The visual measurement system is triggered to acquire feature point images at both ends of the pitch axis of the product under test, and obtain the feature point contour position images of two points, A1 and A2, at both ends of the pitch axis of the product under test.

[0341] A fourth command is issued to the product under test to control the roll axis of the product under test to rotate 180°.

[0342] The vision measurement system is triggered again to collect feature point images at both ends of the pitch axis of the product under test, and the feature point contour position images of two points A1' and A2' at both ends of the pitch axis of the product under test are obtained.

[0343] Image analysis is performed on the feature point contour position images of points A1 and A2, as well as the feature point contour position images of points A1' and A2', to obtain the displacement of the feature point at one end of the pitch axis in the X direction and the displacement of the feature point at the other end of the pitch axis in the X direction. Based on the displacement of the feature points at both ends of the pitch axis in the X direction, the orthogonality / intersection degree between the roll axis and the pitch axis is calculated.

[0344] The following section will provide a detailed description of the biaxial intersection measurement technique disclosed herein, using a specific example.

[0345] In some embodiments, the shaft system error measurement method of the present invention further includes: a two-shaft intersection measurement step, including:

[0346] Add feature points: on the pitch axis of the product under test ( Figure 3 Feature points are added to both ends of axis 1 in the diagram. The feature points are targets (targets can be, but are not limited to, target balls).

[0347] Positioning: The motion platform 1 is used to position the feature points of the product under test within the focal length range of the camera for subsequent feature point measurement;

[0348] Reference measurement: The two cameras in the measurement system are used to measure the feature points at both ends of the pitch axis to obtain the feature point contour position images of points A1 and A2.

[0349] Variation measurement: Rotate the roll axis 180°, that is, rotate axis 1 180° around axis 2. The feature point of axis 1 changes from position A1A2 to A1'A2'. Use the camera again to measure the feature points at both ends of the axis system to obtain the feature point contour position images of points A1' and A2'.

[0350] Analysis and Calculation: Due to errors in the roll axis rotation coding feedback of the tested product, feature point fitting calculations at both ends can be used. Since the orthogonality L value is small, the coding error value is also small. Therefore, the solid and dashed axes in the diagram can be approximated as the same straight line. The orthogonality can be calculated based on spatial geometric relationships.

[0351] Orthogonal error calculation: The displacement of a feature point at one end of the pitch axis system in the X direction |a1-a2| and the displacement of a feature point at the other end in the X direction |a3-a4| can be calculated using image measurement software. The intersection degree between axis 1 and axis 2 is represented by L. Therefore, the theoretical formula for calculating the intersection degree of the two axes of the measured product should be:

[0352]

[0353] In some embodiments, the shaft system error measurement method of the present invention further includes: when a two-axis perpendicularity test command is received, performing the following steps:

[0354] The visual measurement system is triggered to acquire feature point images at both ends of the pitch axis of the product under test, and obtain the feature point contour position images of two points, A1 and A2, at both ends of the pitch axis of the product under test.

[0355] The fifth command is issued to the product under test to control the roll axis of the product under test to rotate 180°.

[0356] The vision measurement system is triggered again to collect feature point images at both ends of the pitch axis of the product under test, and the feature point contour position images of two points A1' and A2' at both ends of the pitch axis of the product under test are obtained.

[0357] Image analysis was performed on the feature point contour position images of points A1 and A2, and the feature point contour position images of points A1' and A2', to obtain the displacement |a1-a2| of the feature point at one end of the pitch axis in the Y direction, and the distance between A2 and A2'. Based on the displacement of the characteristic point at one end of the pitch axis in the Y direction |a1-a2| and the distances of A2 and A2' The perpendicularity between the roll axis and the pitch axis is calculated.

[0358] The perpendicularity measurement of the two axes uses the roll axis (axis 2 in the diagram) as a reference. By rotating the roll axis, the contour change value of the Y-axis feature point at the end of the pitch axis (axis 1 in the diagram) is measured using a multi-view machine vision image measurement system. The method for measuring the perpendicularity error angle θ of the two axes is as follows: Figure 4 As shown. During the two-axis perpendicularity test, the test piece is rotated by the azimuth turntable, and the azimuth axis of the test piece rotates in the opposite direction. Three cameras measure the displacement of the pitch axis system, azimuth axis system, and base respectively.

[0359] The following will provide a detailed description of the biaxial perpendicularity measurement technology solution disclosed herein, using a specific example.

[0360] In some embodiments, the shaft system error measurement method of the present invention further includes:

[0361] Add feature points: Add feature points at both ends of the pitch axis (axis 1 in the figure) of the product under test. The feature points are precision steel balls.

[0362] Positioning: The product under test is positioned within the focal length range of the camera using motion platform 1 for subsequent feature point measurement;

[0363] Reference measurement: The two cameras in the measurement system are used to measure the feature points at both ends of the pitch axis to obtain the feature point contour position images of points A1 and A2.

[0364] Variation measurement: Rotate the roll axis 180°, that is, rotate axis 1 180° around axis 2. The feature point of axis 1 changes from position A1A2 to A1'A2'. Use the camera to measure the feature point again to obtain the feature point contour position image of the two points A1' and A2'.

[0365] Analysis and Calculation: The displacement of the feature point in the Y direction, |a1-a2|, can be calculated using image measurement software. The perpendicularity between axis 1 and axis 2 is represented by angle θ, and the distance between A2 and A2' is represented by... This indicates that the theoretical formula for calculating the perpendicularity of the two axes of the tested product should be: θ = arctan , where θ is the angle representing the perpendicularity between the roll axis and the pitch axis.

[0366] In other embodiments, the shaft system error measurement method of the present invention further includes: upon receiving an axial clearance test command, performing the following steps:

[0367] The constant force loading device is controlled to apply axial tensile and compressive loads to the product under test. The magnitude of the loading force is detected by tensile and compressive force sensors. At the same time, the vision measurement system is controlled to capture images of feature points and record the loading force and the captured feature images.

[0368] Select the first feature image and the second feature image corresponding to two different loading forces, extract the first feature distance value L1 of two feature points in the first feature image and the second feature distance value L2 of two feature points in the second feature image, and obtain the axial clearance of the shaft system based on the first feature distance value L1 and the second feature distance value L2.

[0369] Axial clearance measurement uses the feature profile as a reference. A multi-view machine vision image measurement system is used to capture the profile of the feature before and after axial tensile and compressive loading. The distance values ​​L1 and L2 between the two features are simulated and calculated by vision measurement software. The calculated |L2-L1| is the axial clearance.

[0370] During the axial clearance test, the loading device applies horizontal tensile and compressive loads to the pitch axis system and longitudinal tensile and compressive loads to the azimuth axis system. Three cameras measure the displacement of the pitch axis system, azimuth axis system, and base at three positions.

[0371] The following will provide a detailed description of the axial clearance testing technology solution disclosed herein, using a specific example.

[0372] In some embodiments, the shaft system error measurement method of the present invention further includes: an axial clearance test step, comprising:

[0373] Setting up marker points: Circular measurement marker points and coded markers are set up on the surface of the feature being measured;

[0374] Positioning of the test piece: The circular measurement markers and coded markers on the test product are moved to the focal length and field of view of the multi-view vision image measurement system by the motion platform 1. Each axis of the motion platform 1 is locked to ensure that the selected markers are all within the field of view and focal length of the camera.

[0375] Camera measurement: The three-dimensional coordinates of the marker points are measured using high-precision measuring equipment (such as a single-camera measuring system). Images are taken by dual cameras under dynamic conditions of the target under test, and the coordinates of the measurement points are calculated using camera algorithms.

[0376] Loading channel positioning: The loading mechanism is moved to the shaft end position of the product being tested via the motion axis of the loading device;

[0377] Constant force loading: The motion axis of the low-speed loading device applies axial tensile and compressive loads to the product under test. The magnitude of the loading force is fed back by tensile and compressive sensors, and the feature is continuously photographed by a camera. The loading and photographing are carried out simultaneously, recording the loading force and the photographed feature images.

[0378] Analysis and Calculation: Camera images 1 and 2 with two different loading forces were selected. The feature spacing values ​​L1 and L2 of images 1 and 2 were analyzed using visual measurement software. Under a specific loading force, the axial clearance of the shaft system is:

[0379] △L=│L2-L1│

[0380] By comparing and measuring multiple images, a force-displacement curve can be obtained, thereby measuring the axial displacement and providing a reference for adjusting the axial clearance.

[0381] In some embodiments, the shaft system error measurement method of the present invention further includes:

[0382] When a measurement command for shaft end runout and circular runout is received, the measurement steps for shaft end runout and circular runout are executed.

[0383] The core of shaft end runout and circular runout measurement is to measure the outer contour of the part by a multi-view machine vision image measurement system, rotate each axis of the product under test, continuously measure multiple feature points on the axis system, and measure the runout amount through software fitting and spatial analysis.

[0384] Baseline contour acquisition: Control the measurement axis to rotate slowly at a constant angular velocity, triggering the multi-view vision system to synchronously and continuously acquire cylindrical surface feature images; acquire one set of data for each set rotation angle (10°-15°), covering 360° in the entire process;

[0385] Image processing (distortion correction, edge extraction, subpixel localization) is performed on each group of images to extract the three-dimensional coordinates of the cylindrical surface feature points;

[0386] Reference axis fitting: Import all collected feature points into the software, use the least squares cylinder fitting algorithm to fit the actual cylindrical surface profile of the rotating shaft, and extract the central axis of the cylindrical surface (as the reference axis for circular runout measurement).

[0387] Circular runout calculation: Calculate the radial distance from each feature point to the reference axis to obtain the maximum (Dmax) and minimum (Dmin) radial distance within a 360° range; Circular runout = Dmax - Dmin, which is the radial runout error of the shaft.

[0388] In some embodiments, the shaft system error measurement method of the present invention further includes:

[0389] Reference end face acquisition: Keep the reference axis of the measurement axis parallel to the Z-axis of the global coordinate system, control the measurement axis to slowly rotate 360° at a constant angular velocity, and trigger the multi-view vision system to acquire one set of end face feature images every time the axis is rotated by a set angle (10°-15°);

[0390] Process each set of images to extract the three-dimensional coordinates of the end face feature points (or edge points);

[0391] Reference plane fitting: Import all end face feature points into the software and use the least squares plane fitting algorithm to fit the ideal end face of the shaft (as the reference plane for end runout measurement).

[0392] End runout calculation: Calculate the axial distance from each feature point to the reference plane, and obtain the maximum (Hmax) and minimum (Hmin) axial distance within a 360° range; End runout = Hmax - Hmin, which is the end face circular runout error of the shaft.

[0393] In some embodiments, the shaft system error measurement method of the present invention further includes:

[0394] When a planar geometric tolerance measurement command is received, the planar geometric tolerance measurement steps are executed.

[0395] Commonly used planar geometric tolerance measurements include straightness, perpendicularity, roundness, and parallelism. Multi-view machine vision image measurement technology is used to measure the contour features of parts.

[0396] The orthogonality measurement of the axis of the stabilized turntable is mainly achieved by measuring the three-dimensional coordinates of a feature point on the turntable using a vision measurement system. Multiple measurements are taken at multiple points during a full rotation, and a circle is fitted to determine the center. Additionally, measurements and fitting are performed at another point at a different height, and an axis can be fitted from these two points. Alternatively, multiple measurements can be taken to fit a cylinder, determining the central axis of the cylinder. Based on this, the horizontal and vertical axes are measured, and the included angle and translation of the two axes are calculated.

[0397] Other geometric parameters are calculated and analyzed after image feature extraction and point coordinate measurement, including various system parameters, distances, radii, areas, etc. It has common planar geometric tolerance measurement functions, such as straightness, perpendicularity, roundness, and parallelism. Therefore, the system functions include displacement calculation, attitude calculation, axis measurement, plane measurement and construction, and distance and axis angle analysis. The specific measurement methods for geometric parameter calculation are described below.

[0398] The shaft system error measurement method of the present invention further includes: displacement calculation, and the displacement calculation scheme is as follows:

[0399] The displacement of the workpiece is the change of the workpiece coordinate system origin relative to its initial position at different times. The dual-camera vision measurement system obtains the three-dimensional coordinates of each target point in the global measurement coordinate system through forward intersection measurement. The position of each target relative to the workpiece coordinate system is fixed. At the same time, by calculating the positions of different targets and based on the transformation relationship between each target and the workpiece coordinate system origin, the position of the workpiece coordinate system origin can be obtained. The workpiece origin position calculated based on the different target positions is denoted as (X...). i Y i Z i If we calculate the current position of the workpiece origin, then the final position obtained is:

[0400]

[0401] The displacement of the workpiece is calculated based on the position of the origin of the workpiece coordinate system at different times. Let the coordinates of the workpiece at the origin be (x1, y1, z1) at time t1 and (x2, y2, z2) at time t2. Then the displacement of the workpiece from time t1 to time t2 can be obtained as follows:

[0402] .

[0403] The shaft system error measurement method of the present invention further includes: attitude calculation, and the attitude calculation scheme is as follows:

[0404] During attitude calculation, the camera measures the spatial three-dimensional coordinates of specific target points on the workpiece, and the attitude angles of the moving target are calculated based on these coordinates. The relationship between the workpiece's rotation and transformation in the body coordinate system before and after movement is determined. By transforming the relationship between the body coordinate system and the measurement coordinate system, the attitude angles of the workpiece relative to the measurement coordinate system can be calculated.

[0405] Let the measurement coordinate system be O. r -x r y r z r The initial coordinate system of the workpiece is O. t -x t y t zt The coordinate system of the workpiece after rotation around a fixed point is considered as O. k -x k y k z k The workpiece's position and orientation can be obtained in a user-defined coordinate system, and its orientation parameters are returned as three rotation angles (R). x R y R z ) or four elements (q0, q1, q2, q3). Let the unit vector corresponding to the y-axis in the initial coordinate system of the workpiece be n1(0 1 0)T, n2(1 0 0)T. Paste the target at the center of the coordinate origin, and its center coordinate is (x b y b z b )T; Paste the target in the positive y-axis direction, with its center coordinates marked as (x c y c z c )T; Paste the target in the negative x-axis direction, with its center coordinates marked as (x a y a z a )T.

[0406] After rotation, n1 and n2 will appear in coordinate system O. k -x k y k z k The vector below is:

[0407]

[0408] in,

[0409]

[0410] Let R represent the rotation matrix of a space target performing a fixed-point rotation.

[0411]

[0412] The rotation matrix R can be expressed using the quaternion method as follows:

[0413]

[0414] Based on the relationship between Euler attitude angles and coordinate transformation matrices, the relationship between quaternions and Euler angles can be derived as follows:

[0415]

[0416] These three angles correspond to the yaw angle, roll angle, and pitch angle of the object being measured. The system can calculate the yaw angle, roll angle, and pitch angle of the object being measured at different times, thereby obtaining the real-time attitude of the system.

[0417] There are three ways to construct a point: two straight lines, one straight line and one plane, or three planes.

[0418] Finding the intersection of two lines involves constructing a point using those two lines. First, determine if the two lines intersect based on their direction vectors and error tolerances. If they are parallel, return to the previous step; if they intersect, solve directly using their equations. Let the equations of the two lines be... and Then the formula is as follows:

[0419]

[0420] The equation can be written as a matrix as follows.

[0421]

[0422] It is easy to obtain from the above equation that... By finding the value of , the intersection point of the two lines can be determined.

[0423] When constructing a point through a straight line and a plane, first determine whether the line and plane intersect based on the direction vector of the line, the normal vector of the plane, and the error tolerance. If the line and plane are parallel or the line lies on the plane, return to the previous step. If they intersect, solve directly using the equations of the line and plane. Let the equations of the line and plane be... Then we have the following formula:

[0424]

[0425] The matrix equation is as follows:

[0426]

[0427] The coordinates of the intersection point of the line and the equation can be easily obtained from the above equation.

[0428] When constructing a point through three planes, first determine whether the three planes intersect at a point based on the normal vectors and error tolerances of the three planes. If any two planes are parallel, return to the previous state. If they intersect, first find the intersecting line based on the two planes, and then solve the equation of the third plane using the method of directly constructing a point with the planes.

[0429] The shaft system error measurement method of the present invention further includes: shaft axis measurement and analysis, and the scheme for shaft axis measurement and analysis is as follows:

[0430] The measurement of the axis is based on the measurement of feature points on the axis of rotation. A cylinder is constructed to obtain the axis of the cylinder. The commonly used methods for constructing a straight line are point construction and the intersection of two planes.

[0431] The system constructs straight lines from points using two methods: solving linear equations and nonlinear equations. First, we examine the construction method for solving nonlinear equations. Let the parametric equation of the desired spatial straight line be:

[0432]

[0433] in , , The direction cosine of the line. Let be a point on the straight line. Suppose that the distance measured on the straight line is... The three-dimensional coordinates of each point, in the point set Indicates, point The distance to the line is:

[0434]

[0435] Then according to The unconstrained optimal objective function is established by summing the squared distances from each point to the line as follows:

[0436]

[0437] Solving this nonlinear equation yields the six unknown parameters, thus constructing a straight line in space. The direct construction method is a general approach for constructing straight lines in space, and its principle is relatively simple. However, this method is computationally complex, requires solving nonlinear equations, and is difficult to implement in programming.

[0438] The following describes a method for constructing a straight line by solving linear equations. This method involves projecting a straight line in space onto a coordinate system. , , On a plane, this is transformed into a planar straight line to construct a spatial straight line. Based on this, a method for determining the two most suitable planes to project onto is investigated. This is still based on measuring the planes on which the proposed straight line is to be constructed. Construct a straight line using points, and let its three-dimensional coordinates be . In order to be in , Taking the construction of a straight line by projection as an example, first, let's consider the coordinates of the projection point... and In respectively , Construct a straight line in a plane above and Then these two lines are the required spatial lines in the desired space. , The projected straight line on the coordinate system. And it can be considered that, in the three-dimensional coordinate system... middle, and Indicates two perpendicular to each other , The intersection of these two planes is the spatial straight line to be constructed. Therefore, to complete the construction of the straight line, it is only necessary to find the plane. , and , .

[0439] In a plane coordinate system In China, the following is adopted Projection points Construct a straight line, and substitute the known points into the equation:

[0440]

[0441] Using the least squares principle, its normal equation is derived:

[0442]

[0443] From the above formula, we can obtain:

[0444]

[0445] Similarly, we can obtain , Therefore, the equation of the desired spatial line is:

[0446]

[0447] To determine which two projection planes are ideal and avoid significant deviations in the direction of the constructed line, a method is used to assign a basic direction to the line based on the construction points. Generally, the measurement error of the construction points deviating from the line is much smaller than the distance between points along the line. Based on this characteristic, the direction determined by the two points with the largest set distance in the construction point set is selected as the approximate vector direction of the line, and this is used as the basis for determining which two projection planes to use. For example, the coordinates of the two points with the largest set distance are... and Calculate the direction of the line connecting two points, when the components When it is large, use in , A straight line is constructed by projecting it onto a plane.

[0448] When constructing a straight line through the intersection of two planes, first determine whether the two planes intersect based on their normal vectors. If the two planes are parallel, return to the previous step; if they intersect, directly solve for the intersecting line using their equations. Let the equations of the two planes be... The direction vector of the intersecting line can be obtained by the cross product of the normal vectors of the two planes. Then, the coordinates of a point on the intersection line can be determined by the two equations, and the line is determined.

[0449] When constructing a straight line from points, RMS, standard deviation, maximum deviation, minimum deviation, and morphological error are used respectively. To evaluate the effectiveness of the construction. Let the distance from the point to the line be... The error formula is as follows:

[0450]

[0451] The scheme for constructing the plane is as follows:

[0452] Planes are typically constructed from a series of points; therefore, the optimal point construction method based on the least squares principle is adopted. Let the equation of the plane be... Then its least squares normal equation is as follows:

[0453]

[0454] Because the most reliable value of the measurement result obtained by the least squares method is determined under the condition that the sum of squared residual errors is minimized, the error equation of the plane equation is as follows:

[0455]

[0456] The parameters of the plane equation are in the least squares condition. The solution is obtained under the given conditions. Solving in matrix form within the program, the normal equation can be expressed as:

[0457]

[0458] The least squares formula can be used to find:

[0459]

[0460] At this point, the normal vector of the plane has been determined, and the equation of the entire plane can be determined based on a point on the plane.

[0461] When constructing a plane from a point, the effectiveness of the construction is evaluated using RMS, standard deviation, maximum deviation, minimum deviation, and morphological error. Let the distance from the point to the plane be... The error formula is as follows:

[0462]

[0463] The scheme for constructing the circle is as follows:

[0464] The construction of spatial circles was studied using two methods: projection solution and direct solution.

[0465] 1) Projection solution

[0466] Projection calculations are relatively common. After selecting a series of observation points, first choose a projection plane, project all points onto this plane, and then construct a circle on this plane using these projected points. Let's assume N observation points are measured, in the coordinate system... The Middle The coordinates of the points are Let the equation of the projection plane be:

[0467]

[0468] in Let be the unit normal vector of the plane. From the distance formula, the... The distances from each point to the plane are:

[0469]

[0470] Therefore, the projected coordinates of this point on the plane for:

[0471]

[0472] To simplify the calculation of spatial circle fitting, the spatial points are transformed to the projection plane coordinate system. The process of establishing this coordinate system is as follows:

[0473] a)Origin exist The coordinates in the diagram are the centers of the projection points of each observation point onto the plane.

[0474]

[0475] b) New coordinate system of The axial direction is the direction of the normal vector of the plane, that is... The vector in is .

[0476] c) New coordinate system of The axis direction is chosen as the origin. With any projection point The direction of the line connection, in The unit vector in is :

[0477]

[0478] d) New coordinate system of axial direction, is Direction vector and Cross product of direction vectors:

[0479]

[0480] Thus, the new coordinate system was completed. The transformation relationship between the two coordinate systems is as follows:

[0481]

[0482] After projection and the above transformations, the observation points are all in the new coordinate system. On a plane, with Indicates the projection point on the plane The coordinates in the system can be determined by first constructing the plane, and then transforming back to the original coordinate system.

[0483] There are two methods for constructing a planar circle; the system uses the second method.

[0484] Establish the objective function based on the equation of the plane circle:

[0485]

[0486] Solve this nonlinear equation to find the center of the circle. coordinate values ​​in and circle radius .

[0487] Construct a circle based on the principle that the common perpendicular of the line connecting any two points on the circle passes through the center of the circle.

[0488] Let the center of the desired plane circle be... There are two known points on the plane circle. and Then the midpoint of the two known points. Coordinates are From the properties of a circle, we know that... and If perpendicular, then:

[0489]

[0490] The above equation can be formulated for every two known points. If multiple points exist, a system of linear equations can be established, and the center of the circle can be determined using the least squares principle. The radius of the circle is chosen as the average distance between each point and the center. Using the above method, a planar circle can be constructed on the projection plane, and then the coordinates of the circle's center can be transformed to... Next. At this point, the construction of the spatial circle is complete, represented by its center, radius, and normal vector.

[0491] Direct solution eliminates the need to select planar information; only the points where the circle to be constructed need to be chosen. Here, we investigate methods for intersecting a sphere with a plane, and based on this method, we explore a simplified algorithm.

[0492] The basic principle of the method for finding the intersection of a sphere and a plane is that a spatial circular curve can be considered as the intersection of a sphere and a plane, where the plane is the plane containing the circle, the center of the sphere is the center of the circle, and the radius of the sphere is the radius of the circle. Therefore, the plane and the sphere can be constructed from the observation point, and thus the circle can be constructed.

[0493] First, construct the plane, whose equation is:

[0494]

[0495] According to the projection method, the observation points Projected onto this plane .

[0496] Second, let the equation of the sphere be:

[0497]

[0498] The center of the ball The center of the desired circle is the sphere, and the radius R of the sphere is the radius of the circle. Since the projection point lies both on the plane and on the sphere, the objective function is constructed as follows:

[0499] Solving this nonlinear equation will suffice. This method can construct a spatial circle relatively well, but it requires solving the nonlinear equation.

[0500] Based on the above method, a simplified method for intersecting a sphere and a plane is proposed. The method is simple, the principle is straightforward and feasible, and it does not require projecting the observation point onto the plane containing the circle. After selecting the points to participate in the construction of the circle, a plane is first constructed using these points. This plane is the plane containing the target circle to be constructed. Therefore, the normal vector of this plane is the normal vector of the circle to be constructed. Then, a sphere is constructed using these points. The projection point of the center of the sphere onto the plane is the center of the circle, and the average distance from the center of the circle to each point is the radius.

[0501] First, construct a plane from the measured observation points of the circle, which is the plane containing the circle in question. Its equation is:

[0502]

[0503] Following the method for constructing a sphere mentioned in the next section, construct a sphere from the observation points of the measuring circle. The equation of the sphere is:

[0504]

[0505] The sphere constructed at this point is different from the sphere constructed using the method described above. The circular curve lies on the surface of the sphere, but the center of the sphere... It's not necessarily the center of the circle, and the radius R of the sphere isn't necessarily the radius of the circle. (The center of the sphere is...) Projected onto a plane, we obtain That is the center of the circle we are looking for. The projection method is based on... and the normal vector of the plane Forming a straight line:

[0506]

[0507] The projection point is formed by the intersection of a plane and a straight line. The radius of a circle is obtained by the average distance between a point and the center of the circle.

[0508]

[0509] Thus, the normal vector of the circle center Both the radius R and the radius R have been obtained.

[0510] When constructing a circle from points, the effectiveness of the construction is evaluated using RMS, standard deviation, maximum deviation, minimum deviation, and morphological error.

[0511] The following is a scheme for constructing a cylinder:

[0512] The characteristics of a cylinder are described by its axis and diameter. Let the equation of the cylinder's axis be... , diameter is , The coordinates of the observation points are ( Then, based on the distance from the observation point to the axis... The objective function is established as follows:

[0513]

[0514] Since the diameter of the cylinder is unknown, solving this nonlinear equation can only determine the direction vector of the axis. However, the exact location of the diameter and axis, i.e., a point on the axis, cannot be determined. Construct any plane whose normal vector is the direction vector of the axis, and by constructing a plane circle, find the radius of the cylinder and a point on the axis, thus determining the seven parameter values ​​of the cylinder.

[0515] When constructing a cylinder from a point, the RMS, standard deviation, maximum deviation, minimum deviation, and shape error are used to evaluate the construction effect. Let the distance from the point to the cylinder surface be... The error formula is as follows:

[0516]

[0517] The distance analysis scheme is as follows:

[0518] To make things easier for users, the application implements the calculation of distances between corresponding elements.

[0519] Point / Point: The required element consists of two points. This length can be obtained using the formula for the distance between two points in space.

[0520] Point / Line: First, use the formula for the distance between two points in space to find the distance between the point and any point on the line. Then, calculate the projection of this distance onto the line; this is the desired length. This length can also be solved using the formula method. Let the equation of the line be... , point is The formula is as follows:

[0521]

[0522] Point / Plane: This length can be obtained from the formula for the distance from any point in space to the plane.

[0523] Line / Line: The required element is two lines. This length is equivalent to the length of the common perpendicular segment of the two lines. Let the equations of the two lines be... The formula is as follows:

[0524]

[0525] Plane / Plane: Requires two planes. If the two planes intersect, there is no distance between them; if the two planes are parallel, the distance between them is equivalent to the distance from any point on one plane to the other.

[0526] The scheme for analyzing the included angle of the axes is as follows:

[0527] To make things easier for users, the application implements the calculation of the angle between corresponding elements.

[0528] Line / Line: The required elements are two straight lines. This angle can be solved using a formula. Let the equations of the two lines be... The angle formula is as follows:

[0529]

[0530] Line / Plane: The required elements are a straight line and a plane. After calculating the angle between the plane's normal vector and the straight line using the formula for the angle between two lines, the complementary angle is the desired angle.

[0531] Plane / Plane: The angle between two planes can be converted into the normal vectors of the two planes, i.e., the angle between two lines.

[0532] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A shaft system error measuring device, characterized in that, include: A motion platform is provided, on which a product fixture is mounted. The product fixture is used to fix the product to be tested. The motion platform is used to drive the product fixture mounted on it to move along three translational degrees of freedom and one rotational degree of freedom, thereby realizing the spatial axis positioning of the product to be tested. A constant force loading device, comprising at least one constant force loading component, each constant force loading component being used to apply active loading to each axis of the product under test; A vision measurement system, comprising at least one vision measurement module, wherein the vision measurement module is used to acquire images of the product under test and transmit the acquired images of the product under test to a control unit; The control unit is electrically connected to the motion platform, the constant force loading device, and the measurement system. The control unit is used to perform image analysis on the acquired images of the product under test to obtain axial displacement data during the loading process, and to measure the axial clearance of the product under test based on the axial displacement data.

2. The device according to claim 1, characterized in that: The control unit is used to perform image analysis on the acquired images of the product under test to obtain axial displacement data during the loading process. Based on the axial displacement data, the axial clearance of the product under test is measured, including: extracting key feature points of the platform under test from the image of the product under test, calculating the three-dimensional coordinates of the feature points, calculating the coordinate difference of the feature points during the loading process to obtain axial displacement data, and associating the axial displacement data with the rotation angle of the shaft system and the loading load parameters to achieve the measurement of axial clearance.

3. The device according to claim 2, characterized in that: The control unit is also used to call the camera distortion parameters obtained from the pre-measurement calibration, correct the distortion of the pixel coordinates of the feature points, and then combine the distance information collected by the laser ranging module to calculate the three-dimensional coordinates of the feature points in the camera coordinate system.

4. The device according to claim 3, characterized in that: The control unit is also used to convert the three-dimensional coordinates of feature points in the camera coordinate system into three-dimensional coordinates in the measurement coordinate system through the coordinate system transformation matrix obtained by calibration, so as to provide physical space data with a unified reference for axis system error calculation.

5. The device according to claim 1 or 2, characterized in that: The vision measurement system also includes a laser ranging module, which is used to collect multi-point spatial position data of the product under test and transmit it to the control unit. The control unit is used to determine whether the product under test is within the optimal working range of the vision measurement module based on the multi-point spatial position data. If the product under test is not within the optimal working range of the vision measurement module, the control unit controls the motion platform to adjust the position of the product under test based on the multi-point spatial position data fed back by the laser ranging module until the product under test is within the optimal working range of the vision measurement module.

6. The device according to claim 1, characterized in that: It also includes a target assembly, which is fixedly installed on one side of the pitch axis of the product under test. The target assembly includes a dark field bright line crosshair reticle, a wavelength selective filter, and a backlight light source. The wavelength selective filter is located between the dark field bright line crosshair reticle and the backlight light source. The dark field bright line crosshair reticle, the wavelength selective filter, and the backlight light source are all mounted on the target adjustment mounting bracket.

7. The device according to claim 1, characterized in that: The constant force loading device includes a lateral constant force loading component for applying a lateral constant force to the product under test and a vertical constant force loading component for applying a vertical constant force to the product under test. The lateral constant force loading component includes a lateral loading drive device, which is connected to one end of a first tension / compression sensor. The other end of the first tension / compression sensor is connected to the roll axis end of the product under test through a first fixture. The vertical constant force loading component includes a vertical loading drive device, which is connected to one end of a second tension / compression sensor. The other end of the second tension / compression sensor is connected to the pitch axis end of the product under test through a second fixture.

8. The device according to claim 1, characterized in that: The motion platform includes a transverse axis assembly, a longitudinal axis assembly, a lifting axis assembly, and a rotation axis assembly. The transverse axis assembly is used to drive the product under test to move laterally, the longitudinal axis assembly is used to drive the product under test to move longitudinally, the lifting axis assembly is used to drive the product under test to move up and down, and the rotation axis assembly is used to drive the product under test to rotate around the vertical Z-axis.

9. The device according to claim 1, characterized in that: The control unit is used to acquire the starting voltage data of the motor of the product under test, and to obtain the friction torque of the product under test under different axial clearances based on the starting voltage data, so as to evaluate the optimal control range of the axial clearance of the product under test.

10. A method for measuring shaft system error, characterized in that, Based on the shaft system error measuring device according to any one of claims 1 to 9, the shaft system error measuring method includes the following steps: The constant force loading component is controlled to apply active loading to each axis of the product under test. During the loading process, the vision measurement system is controlled to acquire feature images of the product under test according to different loading loads; Extract key feature points of the platform under test from the image of the product under test, and calculate the three-dimensional spatial coordinates of each feature point; Calculate the coordinate difference of the same feature point under different working conditions during loading to obtain the axial displacement data of the shaft system; By performing multidimensional correlation analysis between axial displacement data and corresponding loading parameters, shaft system error measurement can be achieved.

Images