Anti-interference detection device and detection method for roller cylindricity

Abstract

The application discloses a kind of anti-interference detection device and detection method of roller cylindricity, it is related to precision measurement technical field;Device includes roller conveyor, chain drive mechanism, line confocal sensor and auxiliary point confocal sensor, three-axis vibration acceleration sensor and incremental encoder, which are arranged in diameter;Detection method includes: sensor and accelerometer synchronous sampling;Vibration displacement obtained by using acceleration twice integration is used to carry out vibration compensation to profile point cloud;The difference operation of the measurement value of diametral sensor is separated and deducts reference runout error;The NICP algorithm based on normal-curvature constraint is used to splice each section partial circular arc profile point cloud to obtain complete profile, and finally cylindricity is evaluated;The application can realize the rapid high-precision measurement of roller cylindricity under the non-ideal working condition of existing environmental vibration and reference runout.

Description

Technical Field

[0001] This invention relates to the field of precision measurement technology for machine tool equipment components, specifically to an anti-interference detection device and method for the cylindricity of rollers. Background Technology

[0002] Roller-slider composite guideways are core transmission components in precision machine tools and automated equipment, and their motion accuracy directly determines the machining accuracy of the equipment. As the core load-bearing element of the guideway, the cylindricity of the rollers is a key indicator affecting the smoothness of the guideway transmission. If the cylindricity of the rollers exceeds the allowable tolerance range, it will not only cause vibration and noise during the operation of the guideway, but also significantly increase the heat accumulation of the equipment due to increased friction, thereby inducing thermal errors in the machine tool and seriously affecting the final machining quality. Therefore, rapid and accurate detection of the cylindricity of the rollers is of paramount importance for ensuring the overall performance of machine tool equipment.

[0003] Currently, industrial inspection of roller cylindricity mainly relies on coordinate measuring machines (CMMs) or laser scanning. While CMMs offer high accuracy, their measurement process is extremely time-consuming, failing to meet the rapid online inspection needs of large batches of rollers on production lines. Laser scanning, while improving inspection speed to some extent, is typically limited by the resolution of the optical system and is susceptible to environmental lighting and dust, leading to unstable results. Furthermore, in real-world workshop environments, interference from various vibration sources, such as the high-speed operation of other machine tools and the starting and stopping of conveying mechanisms, can easily cause minute relative displacements between the measuring device and the roller under test. Simultaneously, manufacturing and installation errors in the conveyor rollers themselves can cause reference runout during rotation. Both of these factors (environmental vibration and reference runout) significantly contribute to the final contour data, resulting in severely inaccurate cylindricity assessments.

[0004] To address the shortcomings of existing technologies, namely the difficulty in balancing accuracy and efficiency in traditional testing methods, and their poor robustness in the face of non-ideal working conditions such as environmental vibration and reference runout, it is necessary to develop a testing solution that can effectively isolate environmental vibration and reference runout errors and achieve high-precision, rapid online measurement of roller cylindricity. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an anti-interference detection device and method for roller cylindricity.

[0006] To achieve the above objectives, the present invention provides the following technical solution: This application provides an anti-interference detection method for the cylindricity of rollers, comprising the following steps: S1: Select multiple cross-sections of the roller to be tested, and simultaneously acquire point clouds of multiple partial circumferential contours of each cross-section using a first sensor and a second sensor; wherein, the first sensor and the second sensor are respectively set on two radially opposite sides of the roller to be tested; S2: While the first sensor and the second sensor are sampling, the real-time vibration acceleration signal at the installation position of the first sensor and the second sensor is collected simultaneously, and the vibration displacement compensation is performed on the multi-segment partial circumferential contour point cloud based on the real-time vibration acceleration signal to eliminate the contour error caused by environmental vibration. S3: Obtain the relative radial measurement values ​​obtained by the first sensor and the second sensor at any measurement moment, calculate the instantaneous eccentricity error of the roller under test based on the relative radial measurement values, and separate and remove the instantaneous eccentricity error from the multi-segment partial circumferential contour point cloud to eliminate the measurement error caused by the roller reference runout; S4: The point clouds of the multi-segment circumferential contours of each cross section after processing by S2 and S3 are registered and stitched together using the iterative nearest point algorithm based on normal-curvature constraints to obtain the complete true contour point cloud of all cross sections of the roller under test, and the cylindricity is evaluated based on the complete true contour point cloud.

[0007] Furthermore, the first sensor is a line confocal sensor mounted on a line confocal sensor fixture, and the second sensor is an auxiliary point confocal sensor mounted on an auxiliary point confocal sensor fixture. The line confocal sensor and the auxiliary point confocal sensor are located directly above and directly below the roller to be tested, respectively. The real-time vibration acceleration signal is acquired by a triaxial vibration acceleration sensor mounted on the line confocal sensor fixture and the auxiliary point confocal sensor fixture, and the line confocal sensor, the auxiliary point confocal sensor and the triaxial vibration acceleration sensor keep synchronous sampling to obtain data with a unified timestamp.

[0008] Furthermore, in S1, the rollers of the roller conveyor drive the roller to be tested to rotate. Every time it rotates through a fixed angle, a high-precision incremental rotary encoder installed at the end of the driven roller sends a trigger pulse to trigger the first sensor and the second sensor to perform the next sampling, so that the arc contour point cloud obtained by two adjacent samplings has a set overlapping area in the circumferential direction. The area of ​​the defined overlapping region accounts for 15% to 20% of the total area of ​​the arc contour point cloud.

[0009] Furthermore, the vibration displacement compensation based on the real-time vibration acceleration signal described in S2 specifically includes: The acceleration signals of the X-axis, Y-axis, and Z-axis collected by the triaxial vibration accelerometer are respectively denoted as a. x (t), a y (t), a z (t); The acceleration signal is integrated twice over time to obtain the real-time vibration displacement d(t). The integration process is as follows: First, integrate the acceleration to obtain real-time velocity data: Then, the obtained velocity is integrated over time again to obtain the real-time vibration displacement: in, For acceleration signals, For real-time speed, For real-time vibration displacement, The initial velocity, This is the initial displacement.

[0010] Furthermore, the vibration displacement compensation for the multi-segment partial circular contour point cloud described in S2 specifically includes: Based on the timestamp of the measurement results, obtain any time T. i Real-time vibration displacement (dX(T) in the X and Y axes) i ),dY(T i )); The original contour coordinates (X) obtained by the first sensor or the second sensor i ,Y i Subtract the vibration displacement (dX(T)) in the X and Y axes at the corresponding time. i ),dY(T i The coordinates of the contour points after vibration compensation are obtained. The calculation formula is: .

[0011] Furthermore, the calculation of instantaneous eccentricity error described in S3 specifically includes: At any measurement moment, obtain the vibration-compensated value S(0) at the center direction of the contour point cloud measured by the first sensor, and the reading value S(180) at the opposite side direction obtained by the second sensor. Calculate the true value of the center of the point cloud of the roller profile under test based on S(0) and S(180): S(0) 真实 And the instantaneous eccentricity error X caused by roller runout 误差 The calculation formula is as follows: S(0) 真实 =(S(0)+S(180)) / 2 X 误差 =(S(0)-S(180)) / 2.

[0012] Furthermore, the separation and elimination of the instantaneous eccentricity error described in S3 specifically includes: The instantaneous eccentricity error X 误差 Subtracting the measured values ​​of all sampling points on the corresponding contour line yields the pure contour data S (pure contour data), which contains only the contour information of the roller itself. The calculation formula is: S (pure contour data) = S (including reference runout error) - X 误差 ; Wherein, S (including reference runout error) is the original profile measurement value including the roller reference runout error.

[0013] Furthermore, the registration and stitching using the Iterative Closest Point (NICP) algorithm based on normal-curvature constraints, as described in S4, specifically includes: The arc contour point cloud acquired in the i-th acquisition is used as the target point cloud, and the arc contour point cloud acquired in the (i+1)-th acquisition is used as the source point cloud. For each point in the target point cloud and the source point cloud, a covariance matrix is ​​constructed by taking local neighborhood points. The normal vector and curvature value of each point are calculated by singular value decomposition (SVD). The features of each point are then expanded into a multi-dimensional feature vector that includes position, normal vector and curvature.

[0014] Furthermore, the registration and stitching described in S4 also includes: A multi-constraint objective function is constructed, using Euclidean distance, normal angle, and curvature difference as constraints for registration optimization; wherein the normal angle is limited to less than 15°, and the curvature difference is limited to less than a preset curvature threshold. The objective function is solved using the LM algorithm to obtain the rotation matrix R and the translation vector t. The rotation matrix R and the translation vector t are then applied to the source point cloud. The solution is iterated multiple times until the error change is less than the convergence threshold, thereby achieving seamless stitching of adjacent arc point clouds and obtaining the complete true contour point cloud of all cross-sections of the roller under test.

[0015] A testing apparatus for performing the above-described anti-interference testing method for roller cylindricity includes: A roller conveyor having multiple conveyor rollers for supporting and driving the roller to be tested to rotate; A chain drive mechanism, comprising sprockets disposed at the ends of each conveyor roller and chains sequentially meshing with each sprocket, the chain drive mechanism being used to drive all conveyor rollers to rotate synchronously at the same speed and in the same direction; The fixed bracket includes a first fixed bracket and a second fixed bracket; The lead screw drive mechanism includes a first lead screw drive mechanism for mounting and driving the first fixed bracket and the second fixed bracket to move along the X-axis direction, a second lead screw drive mechanism for mounting and driving the line confocal sensor to move along the Y-axis direction, and a third lead screw drive mechanism for mounting and driving the auxiliary point confocal sensor to move along the Y-axis direction. A linear confocal sensor, mounted on the second lead screw transmission mechanism, is used to acquire the point cloud of the arc contour of the cross-section of the roller to be measured; An auxiliary point confocal sensor is mounted on the third lead screw transmission mechanism and is positioned opposite the line confocal sensor below the roller to be measured, for acquiring the opposite point data of the arc contour point cloud; A triaxial vibration accelerometer is mounted on the fixture of the line confocal sensor and the auxiliary point confocal sensor to acquire the absolute vibration information of the measuring device in real time. A high-precision incremental rotary encoder is installed at the end of the driven roller of the roller conveyor. It is used to send a trigger signal every time the driven roller rotates through a fixed angle, so as to trigger the linear confocal sensor and the auxiliary point confocal sensor to sample all cross sections of the roller under test for the next time. Specifically, the linear confocal sensor and the auxiliary point confocal sensor detect a total of 6 cross sections in one sampling. Then, the conveyor rotates, driving the roller to rotate, and the encoder sends a trigger signal to perform the next sampling of 6 cross sections.

[0016] And a control unit, used to receive signals from the line confocal sensor, the auxiliary point confocal sensor, the triaxial vibration acceleration sensor and the high-precision incremental rotary encoder, and to execute the above-described detection method.

[0017] Compared with the prior art, this application has the following beneficial effects: This invention proposes an anti-interference detection device and method for roller cylindricity. By employing a high-resolution line confocal sensor and combining innovative mechanical structures and algorithms, it achieves rapid and accurate measurement of roller cylindricity. Its beneficial effects are as follows: Continuous transmission of multi-station rollers is achieved through a roller conveyor and chain drive mechanism, coupled with pulse triggering by a high-precision encoder, enabling the device to simultaneously handle online rapid detection and high-precision sampling; By integrating a triaxial vibration acceleration sensor into the sensor fixture of the measuring device and combining it with two time integration calculations, real-time vibration noise introduced by the operation of surrounding equipment is effectively filtered out; Simultaneously, using an auxiliary point confocal sensor located in the radial direction of the roller, the instantaneous eccentricity error caused by the runout of the mechanical reference (roller) is calculated and compensated in real time, ensuring that the final extracted contour point cloud completely eliminates external interference and truly reflects the spatial attitude of the roller itself; Combined with the NICP algorithm based on normal-curvature constraints, while ensuring an overlap rate of 15%~20% between adjacent arc segments, seamless high-precision stitching of the point cloud is achieved using multi-dimensional feature constraints, thereby significantly improving the accuracy and reliability of the final cylindricity assessment. Attached Figure Description

[0018] Figure 1 This is a diagram showing the installation position of the roller cylindricity detection device for the roller-sliding composite guide rail in this invention, measured on a roller conveyor.

[0019] Figure 2 This is an isometric view of the roller cylindricity detection device for the roller-sliding composite guide rail in an embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram illustrating the principle of the roller cylindricity detection device for the roller-sliding composite guide rail in this invention, which measures a portion of the circumference of a cross-section of a roller on a roller conveyor.

[0021] Figure 4 This is a schematic diagram of the overall measurement process of the roller cylindricity of the roller cylindricity detection device for the roller-sliding composite guide rail in this embodiment of the invention.

[0022] Figure 5 This is a schematic diagram of the data processing flow of the roller cylindricity detection device for the roller-sliding composite guide rail in this embodiment of the invention.

[0023] Figure 6 This is a schematic diagram of the movement trajectory of the line confocal sensor of the roller cylindricity detection device of the roller-sliding composite guide rail on the roller conveyor in an embodiment of the present invention.

[0024] Figure label: 1. First fixed bracket; 2. High-precision incremental rotary encoder; 3. First lead screw drive mechanism; 4. Second lead screw drive mechanism; 5. Sprocket; 6. Linear confocal sensor; 7. Roller to be tested; 8. Linear confocal sensor fixture; 9. Auxiliary point confocal sensor fixture; 10. Conveyor roller; 11. Auxiliary point confocal sensor; 12. Third lead screw drive mechanism; 13. Triaxial vibration acceleration sensor; 14. Second fixed bracket; 15. Chain. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Furthermore, in this invention, an element referred to as fixed to or disposed on another element may be directly disposed on the other element, or there may be an intermediate element. When an element is considered to be connected to another element, it may be directly connected to the other element, or there may be an intermediate element present simultaneously. The terms vertical, horizontal, left, right, and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0027] Example 1 This embodiment provides a detailed description of the online rapid anti-interference detection device for roller cylindricity under non-ideal working conditions.

[0028] like Figure 1 and Figure 2 As shown in this embodiment, an online rapid anti-interference detection device for the cylindricity of rollers under non-ideal working conditions is provided, mainly applied to roller conveyors. The core of this device lies in accurately acquiring the original data of the roller under test in a non-ideal measurement environment through reasonable mechanical layout and electromechanical coordination.

[0029] The roller conveyor is equipped with multiple conveyor rollers 10 to support and drive the roller 7 under test to rotate. To ensure that all rollers are strictly synchronized, in this embodiment, a sprocket 5 is fixedly installed at the end of every two adjacent conveyor rollers 10, and adjacent sprockets 5 are connected by a chain 15. When the motor drives the first conveyor roller 10 (or the sprocket 5 set coaxially with it) to rotate, the sprocket teeth drive the chain 15, and the power is transmitted down the chain 15 step by step, driving all conveyor rollers 10 to rotate synchronously at the same speed and in the same direction. This mechanical design of the same direction and speed fundamentally ensures that the rotation of the roller 7 under test is smooth and predictable during the measurement process, avoiding sliding friction caused by uneven speed of each roller axis, thereby ensuring accurate control of the sampling angle.

[0030] To achieve point cloud acquisition at different cross-sections and spatial locations, this device is equipped with a spatial coordinate movement system. Specifically, the system includes a first screw drive mechanism 3 fixedly mounted on the frames on both sides of the conveyor, used to drive the entire device to move along the conveyor's transmission direction (X-axis direction). Two moving supports are respectively provided on the first screw drive mechanism 3, namely a first fixed support 1 and a second fixed support 14. A second screw drive mechanism 4 is provided on the first fixed support 1, used to drive the line confocal sensor 6 to move with high precision along a direction perpendicular to the conveyor (Y-axis direction). A third screw drive mechanism 12 is provided on the second fixed support 14, used to drive the auxiliary point confocal sensor 11 to move with high precision along the Y-axis direction. Figure 6 As shown, the linear confocal sensor 6, driven by the second lead screw transmission mechanism 4, can traverse each cross-sectional position along the axial direction of the roller to be measured 7, and ensures full coverage of the roller through multi-point position scanning.

[0031] To achieve non-contact and high-precision contour scanning, the sensor layout in this embodiment has a specific cooperative relationship. See [link to documentation]. Figure 1The linear confocal sensor 6 is mounted on the second lead screw drive mechanism 4 via the linear confocal sensor fixture 8, and is located at the exact center of the two rollers of the conveyor, just above the roller 7 to be measured (corresponding to the 0° direction). The auxiliary point confocal sensor 11 is mounted on the third lead screw drive mechanism 12 via the auxiliary point confocal sensor fixture 9, and is directly opposite the linear confocal sensor 6, located directly below the roller 7 to be measured (corresponding to the 180° direction). The aforementioned sensors arranged in a diametrically opposed manner can simultaneously acquire the contour values ​​of both sides of the same cross-section of the roller, providing basic data support for subsequent separation reference runout error. A high-precision incremental rotary encoder 2 is installed at the end of the driven roller of the conveyor. The encoder is used to monitor the actual rotation angle of the roller in real time. Whenever the driven roller rotates precisely through a fixed small angle, the encoder 2 sends a trigger pulse signal. This trigger signal controls the line confocal sensor 6 and the auxiliary point confocal sensor 11 to perform the next sampling of all cross sections of the roller, and obtain the partial arc contour data of all cross sections of the roller at the current angle. The advantage of this setting is that even if the roller under test 7 and the conveyor roller 10 slip slightly due to surface dirt or gravity, the encoder can still send commands based on the real rotation angle, so that the contour data obtained by two adjacent samplings can always maintain an overlap area of ​​15%~20% in the circumferential direction, providing a solid mathematical premise for subsequent registration and splicing.

[0032] Furthermore, a triaxial vibration acceleration sensor 13 is rigidly mounted on both the linear confocal sensor fixture 8 and the auxiliary point confocal sensor fixture 9. In a real workshop environment, the vibration generated by the operation of peripheral equipment (such as grinding machines, milling machines, etc.) will be transmitted to the measuring bracket, causing a relative micro-displacement between the sensor and the roller 7 to be measured. The triaxial vibration acceleration sensor 13 can acquire the absolute vibration information of the bracket in the X, Y, and Z axes in real time. The acquisition of this vibration signal and the sampling of the linear confocal and auxiliary point confocal sensors adopt a synchronous triggering mechanism, so that the data acquired by the three have the same timestamp, ensuring the real-time performance and accuracy of subsequent error compensation.

[0033] Example 2 Based on Example 1, this example elaborates on the online rapid anti-interference detection method for roller cylindricity under non-ideal working conditions: Based on the apparatus described in Example 1, this embodiment provides a detection method. Combined with... Figure 4 As shown, the actual operation process of this embodiment is as follows: The first step is to control the lead screw mechanism to move the sensor to the position to be measured, and select multiple cross sections of the roller 7 to be measured (taking 6 cross sections as an example) as measurement target points in an equally spaced manner; The second step involves the line confocal sensor 6 and the auxiliary point confocal sensor 11 sampling all selected cross sections (taking 6 cross sections as an example), sampling a portion of the circular arc contour of 6 cross sections at a time, while the triaxial vibration acceleration sensor 13 synchronously collects the vibration acceleration data at the current moment. Third, after the online confocal sensor 6 and the auxiliary point confocal sensor 11 have completed one sampling of all cross sections, the roller conveyor starts to run, thereby driving the roller under test 7 to rotate continuously. When the encoder detects that the roller has rotated through a preset fixed angle, it immediately sends a trigger signal to drive the sensor to perform the next sampling. Fourth, repeat the above rotation and sampling process until the system obtains the complete circumferential contour point cloud of all cross sections; Fifth step: After all selected cross-sections have been sampled, the system proceeds according to... Figure 5 The data processing flow shown performs anti-interference processing on the original point cloud, and finally obtains the true contour of the roller and outputs the cylindricity evaluation result.

[0034] Furthermore, in order to restore the true profile of the roller under non-ideal working conditions with environmental vibration and reference runout, this embodiment adopts a high-precision data post-processing algorithm in the fifth step, which is specifically divided into three parts: vibration error compensation, reference runout error separation, and point cloud stitching based on multi-dimensional feature constraints.

[0035] For the aforementioned vibration error compensation, firstly, the system uses a unified timestamp as a reference to compare the contour data collected from the linear confocal sensor 6 and the auxiliary point confocal sensor 11 with the acceleration signal a collected from the triaxial vibration acceleration sensor 13. x (t), a y (t), a z (t) Data alignment is performed; in order to convert the acceleration signal into a displacement signal to eliminate spatial jitter, this embodiment uses a precise double time integration method: first, the acceleration signal is integrated over time to obtain real-time velocity data, i.e. In this formula, This represents the acceleration signal along any axis. The real-time speed obtained through integration. The initial velocity at the measurement moment is set to 0; then, the obtained velocity signal is integrated over time again to obtain the real-time vibration displacement, i.e. ,in, For the calculation of real-time vibration displacement, To measure the initial displacement at the start.

[0036] Through the above integration process, the measurement time T can be obtained. i Vibration displacement components (dX(T)) i ),dY(Ti Finally, based on the timestamp correspondence, the original contour coordinates (X) obtained by the line confocal sensor are... i ,Y i Subtracting the vibration displacement component at that moment yields the true contour coordinates after vibration compensation, and the calculation formula is as follows: This step effectively filters out the interference of various low-frequency vibration noises in the workshop on the measurement data.

[0037] Regarding the real-time separation of the reference runout error, various factors, such as manufacturing shape errors of the roller itself, rotational errors of the supporting bearings, eccentricity in roller installation, and speed fluctuations in the transmission chain, cause irregular spatial runout of the actual rotation center of the roller during rotation. This runout is transmitted to the measured roller through the contact interface between the roller and the roller shaft, causing the overall axis of the roller to translate relative to the fixed sensor support. To eliminate this influence, this embodiment uses a confocal sensor 11 with a diametrically aligned auxiliary point for compensation. At any sampling moment, the vibration-compensated value S(0) at the midpoint of the contour point collected by the line confocal sensor 6 in the 0° direction and the synchronously opposite reading value S(180) collected by the auxiliary confocal sensor 11 in the 180° direction are obtained. The average value of these two diametrically aligned measurement results is used as the true radial reference of the center point of the contour point cloud, i.e., S(0). 真实 =(S(0)+S(180)) / 2; Through mathematical difference operations, the instantaneous eccentricity error X caused by the roller bounce at that sampling moment can be calculated in real time with high precision. 误差 The calculation formula is X 误差 =(S(0) S(180)) / 2.

[0038] Then, the instantaneous eccentricity error X 误差 The raw values ​​of all sampling points along this contour line are directly subtracted to obtain pure contour point cloud data that retains only the true geometric information of the roller itself. The processing formula is S(pure contour data) = S(including reference runout error) - X 误差 In this formula, S (including the reference runout error) represents the original profile measurement value before deducting the reference runout error; through this separation mechanism, even if the conveyor rollers have large runout, accurate roller profile information can still be extracted.

[0039] For the NICP precision stitching based on normal-curvature constraints, after the above two steps of error elimination, each measured cross section will obtain multiple partial arc point clouds, and adjacent point clouds will have 15%~20% physical overlap.

[0040] The specific stitching process is as follows: The arc contour point cloud acquired in the i-th acquisition is used as the target point cloud, and the arc contour point cloud acquired in the (i+1)-th acquisition is used as the source point cloud. Using the fixed rotation angle provided by encoder 2, the estimated overlapping area of ​​the two point clouds can be quickly located, thereby reducing the computational load of subsequent iterations. Then, for each point in the target and source point clouds, local neighborhood points are selected, and a covariance matrix is ​​constructed. Singular value decomposition (SVD) is then performed on this covariance matrix. Based on the decomposition result, the normal vector and curvature value of each point are calculated, and the original 3D position information, normal vector, and curvature of each point are expanded into a high-dimensional feature vector. Next, a multi-constraint objective function is constructed, specifically including the Euclidean distance between point pairs, ... The registration optimization process considers three constraints: the angle between the normals of two matching points (which must be less than a preset threshold, preferably 15° in this embodiment) and the difference in curvature between them (which must also be less than a preset threshold). Only point pairs that simultaneously satisfy these three constraints are included in the registration optimization. Then, the Levenberg-Marquardt (LM) algorithm is used to solve the above multi-constraint objective function, thereby solving for the optimal rotation matrix R and translation vector t. The calculated R and t are applied to the source point cloud to update its spatial position, and the above solution process is repeated iteratively until the relative displacement error between the source point cloud and the target point cloud is less than the set convergence threshold. When the iteration terminates, the source point cloud has been accurately transformed into the coordinate system of the target point cloud, and the two naturally merge in the overlapping area.

[0041] By sequentially performing this splicing process on all arc segments of the same cross-section, a complete circular outline can be spliced ​​out; by reconstructing the complete outline of all cross-sections, a complete and realistic three-dimensional outline point cloud of the roller under test can be obtained.

[0042] After obtaining the aforementioned realistic 3D contour point cloud, the conventional least-squares cylindricity assessment algorithm in this field can be used to perform cylindrical fitting on the contour point cloud and calculate the maximum radial deviation between the actual contour and the fitted cylinder, thereby outputting the final cylindricity value of the roller 7 to be tested. This device and method utilize efficient data compensation and stitching algorithms to successfully solve the problems of slow detection speed and poor accuracy in complex and non-ideal workshop environments in existing technologies, and has strong industrial application value.

[0043] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0044] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for anti-interference detection of roller cylindricity, characterized in that, Includes the following steps: S1: Select multiple cross sections of the roller (7) to be tested, and use the first sensor and the second sensor to simultaneously acquire the point cloud of the partial circumferential contour of each cross section; wherein, the first sensor and the second sensor are respectively set on the radially opposite sides of the roller (7) to be tested; S2: While the first sensor and the second sensor are sampling, the real-time vibration acceleration signal at the installation position of the first sensor and the second sensor is collected simultaneously, and the vibration displacement compensation is performed on the multi-segment partial circumferential contour point cloud based on the real-time vibration acceleration signal to eliminate the contour error caused by environmental vibration. S3: Obtain the relative radial measurement values ​​measured by the first sensor and the second sensor at any measurement moment, calculate the instantaneous eccentricity error of the roller to be measured (7) based on the relative radial measurement values, and separate and remove the instantaneous eccentricity error from the multi-segment partial circumferential contour point cloud to eliminate the measurement error caused by the roller reference runout; S4: The point clouds of the multi-segment circumferential contours of each cross section after processing by S2 and S3 are registered and stitched together using the iterative nearest point algorithm based on normal-curvature constraints to obtain the complete true contour point cloud of all cross sections of the roller (7) to be tested, and the cylindricity is evaluated based on the complete true contour point cloud.

2. The anti-interference detection method for roller cylindricity according to claim 1, characterized in that, The first sensor is a line confocal sensor (6) mounted on a line confocal sensor fixture (8), and the second sensor is an auxiliary point confocal sensor (11) mounted on an auxiliary point confocal sensor fixture (9). The line confocal sensor (6) and the auxiliary point confocal sensor (11) are located directly above and directly below the roller (7) to be tested, respectively. The real-time vibration acceleration signal is acquired by a triaxial vibration acceleration sensor (13) mounted on the line confocal sensor fixture (8) and the auxiliary point confocal sensor fixture (9), and the line confocal sensor (6), the auxiliary point confocal sensor (11) and the triaxial vibration acceleration sensor (13) keep synchronous sampling to obtain data with a unified timestamp.

3. The anti-interference detection method for roller cylindricity according to claim 1, characterized in that, In S1, the roller (10) of the roller conveyor drives the roller (7) to be tested to rotate. Every time it rotates through a fixed angle, a trigger pulse is sent by the high-precision incremental rotary encoder (2) installed at the end of the driven roller to trigger the first sensor and the second sensor to perform the next sampling, so that the arc contour point cloud obtained by two adjacent samplings has a set overlapping area in the circumferential direction. The area of ​​the defined overlapping region accounts for 15% to 20% of the total area of ​​the arc contour point cloud.

4. The anti-interference detection method for roller cylindricity according to claim 1, characterized in that, The vibration displacement compensation based on the real-time vibration acceleration signal described in S2 specifically includes: The acceleration signals of the X-axis, Y-axis, and Z-axis collected by the triaxial vibration accelerometer (13) are respectively denoted as a x (t), a y (t), a z (t); The acceleration signal is integrated twice over time to obtain the real-time vibration displacement d(t). The integration process is as follows: First, integrate the acceleration to obtain real-time velocity data: Then, the obtained velocity is integrated over time again to obtain the real-time vibration displacement: in, For acceleration signals, For real-time speed, For real-time vibration displacement, The initial velocity, This is the initial displacement.

5. The anti-interference detection method for roller cylindricity according to claim 4, characterized in that, The vibration displacement compensation for multi-segment partial circular contour point clouds described in S2 specifically includes: Based on the timestamp of the measurement results, obtain any time T. i Real-time vibration displacement (dX(T) in the X and Y axes) i ),dY(T i )); The original contour coordinates (X) obtained by the first sensor or the second sensor i ,Y i Subtract the vibration displacement (dX(T)) in the X and Y axes at the corresponding time. i ),dY(T i The coordinates of the contour points after vibration compensation are obtained. The calculation formula is: 。 6. The anti-interference detection method for roller cylindricity according to claim 1, characterized in that, The calculation of instantaneous eccentricity error described in S3 specifically includes: At any measurement moment, obtain the vibration-compensated value S(0) at the center direction of the contour point cloud measured by the first sensor, and the reading value S(180) at the opposite side direction obtained by the second sensor. Calculate the true value of the center of the point cloud of the profile of the roller (7) to be measured based on S(0) and S(180): S(0) 真实 And the instantaneous eccentricity error X caused by roller runout 误差 The calculation formula is as follows: S(0) 真实 =(S(0)+S(180)) / 2 X 误差 =(S(0)-S(180)) / 2。 7. The anti-interference detection method for roller cylindricity according to claim 6, characterized in that, The separation and elimination of the instantaneous eccentricity error described in S3 specifically includes: The instantaneous eccentricity error X 误差 Subtracting the measured values ​​of all sampling points on the corresponding contour line yields the pure contour data S (pure contour data), which contains only the contour information of the roller itself. The calculation formula is: S (pure contour data) = S (including reference runout error) - X 误差 ; Wherein, S (including reference runout error) is the original profile measurement value including the roller reference runout error.

8. The anti-interference detection method for roller cylindricity according to claim 1, characterized in that, The registration and stitching process described in S4 uses the Iterative Closest Point (NICP) algorithm based on normal-curvature constraints, specifically including: The arc contour point cloud acquired in the i-th acquisition is used as the target point cloud, and the arc contour point cloud acquired in the (i+1)-th acquisition is used as the source point cloud. For each point in the target point cloud and the source point cloud, a covariance matrix is ​​constructed by taking local neighborhood points. The normal vector and curvature value of each point are calculated by singular value decomposition (SVD). The features of each point are then expanded into a multi-dimensional feature vector that includes position, normal vector and curvature.

9. The anti-interference detection method for roller cylindricity according to claim 8, characterized in that, The registration and stitching described in S4 also includes: A multi-constraint objective function is constructed, using Euclidean distance, normal angle, and curvature difference as constraints for registration optimization; wherein the normal angle is limited to less than 15°, and the curvature difference is limited to less than a preset curvature threshold. The objective function is solved using the LM algorithm to obtain the rotation matrix R and the translation vector t. The rotation matrix R and the translation vector t are then applied to the source point cloud. The solution is iterated multiple times until the error change is less than the convergence threshold, thereby achieving seamless splicing of adjacent arc point clouds and obtaining the complete true contour point cloud of all cross-sections of the roller (7) to be tested.

10. A testing apparatus for performing an anti-interference testing method for the cylindricity of rollers as described in any one of claims 1 to 9, characterized in that, include: A roller conveyor having multiple conveyor rollers (10) for supporting and driving the roller (7) to be tested to rotate; The chain drive mechanism includes sprockets (5) disposed at the ends of each conveyor roller (10) and chains (15) that mesh with each sprocket (5) in sequence. The chain drive mechanism is used to drive all conveyor rollers (10) to rotate synchronously at the same speed and in the same direction. The fixed bracket includes a first fixed bracket (1) and a second fixed bracket (14); The lead screw drive mechanism includes a first lead screw drive mechanism (3) for mounting and driving the first fixed bracket (1) and the second fixed bracket (14) to move along the X-axis direction, a second lead screw drive mechanism (4) for mounting and driving the line confocal sensor (6) to move along the Y-axis direction, and a third lead screw drive mechanism (12) for mounting and driving the auxiliary point confocal sensor (11) to move along the Y-axis direction. A line confocal sensor (6) is mounted on the second lead screw transmission mechanism (4) to acquire the arc contour point cloud of the cross section of the roller (7) to be measured; An auxiliary point confocal sensor (11) is installed on the third lead screw transmission mechanism (12) and is positioned opposite to the line confocal sensor (6) below the roller to be measured (7) to acquire the opposite point data of the arc contour point cloud. A triaxial vibration acceleration sensor (13) is mounted on the clamp of the line confocal sensor (6) and the auxiliary point confocal sensor (11) to acquire the absolute vibration information of the measuring device in real time. A high-precision incremental rotary encoder (2) is installed at the end of the driven roller of the roller conveyor and is used to send a trigger signal when the driven roller rotates through a fixed angle to trigger the line confocal sensor (6) and the auxiliary point confocal sensor (11) to sample all cross sections of the roller to be measured (7) for the next time. The control unit is configured to receive signals from the line confocal sensor (6), the auxiliary point confocal sensor (11), the triaxial vibration acceleration sensor (13), and the high-precision incremental rotary encoder (2), and to perform the detection method as described in any one of claims 1 to 9.

Images