A vertical shaft rigid cage anti-shaking deformation data acquisition device and detection method

Abstract

This invention relates to the field of coal mine equipment safety inspection technology, and provides a vertical shaft rigid cage guide anti-vibration deformation data acquisition device, including a base on which a data acquisition component and a data terminal are mounted. The data acquisition component includes a bracket, on which a line laser and an industrial camera are mounted, and the industrial camera is equipped with an IMU unit. This invention forms the cage guide profile through active projection using a line laser, eliminating the need to rely on natural underground textures or stable corner points, making it suitable for the low-light, weak-texture environment of vertical shafts. This invention also provides a detection method that compensates for rotational and translational distortions caused by short-term cage vibrations, providing a more reliable data foundation for the maintenance and safety assessment of vertical shaft hoisting systems.

Description

Technical Field

[0001] This invention belongs to the field of coal mine equipment safety testing technology, and in particular relates to a data acquisition device and testing method for anti-vibration deformation of a rigid shaft guideway. Background Technology

[0002] Rigid guideways are crucial components in coal mine vertical shaft hoisting systems, guiding the hoisting container for stable operation. Affected by factors such as geological movement, foundation settlement, long-term load impact, installation errors, and operational wear, rigid guideways are prone to deformation problems during operation, including localized bulges, joint misalignment, and abnormal gaps between sections. If these deformations are not detected in time, the hoisting container may experience swaying, jamming, collisions, or even derailment during operation, thereby affecting the safe operation of the mine hoisting system.

[0003] Three-dimensional measurement methods based on line structured light can actively project laser stripes, acquire laser contours through a camera, and reconstruct point clouds of the hoistway, exhibiting good non-contact measurement and geometric characterization capabilities. However, in actual vertical shaft hoisting processes, the detection device is typically installed on top of the hoisting container (cage) or moves synchronously with it. During operation, the hoisting container is affected by fluctuations in wire rope tension, localized impacts in the hoistway, braking disturbances, and structural vibrations, causing short-term vibrations in the detection device. These vibrations are not simple, continuous, and stable oscillations, but often manifest as short-duration, abrupt amplitude changes, and locally non-stationary attitude deflections and rigid displacements.

[0004] Since the line structured light point cloud is acquired and stitched together frame by frame during the lifting process, the jitter of the detection device will directly introduce rotational distortion and translational distortion of the point cloud, causing the originally continuous tank channel point cloud to become misaligned, bent or locally jumpy, thus affecting the accurate measurement of deformation dimensions such as protrusions, misalignments and gaps.

[0005] In existing solutions, some methods suppress continuous vibration interference through filtering. However, these methods typically require vibrations to have relatively stable statistical characteristics, making them insufficient for adapting to short-term sudden shaking of containers. Other solutions estimate camera pose using visual inertial odometry, but images in vertical shaft tank passage inspection scenarios mainly consist of single laser stripes, lacking stable corner points and rich textures, making reliable multi-frame visual matching difficult. Relying solely on the quadratic integral of acceleration from the inertial measurement unit to estimate translational displacement is susceptible to bias and noise accumulation, failing to meet the millimeter-level deformation detection requirements of tank passages.

[0006] In conclusion, the existing technology obviously has inconveniences and defects in practical use, so it is necessary to improve it. Summary of the Invention

[0007] To address the aforementioned shortcomings, this invention primarily provides a vertical shaft rigid tank guide anti-vibration deformation data acquisition device, solving the technical problem of low deformation detection accuracy caused by disturbances during the detection process.

[0008] To address the aforementioned problems, this invention provides a vertical shaft rigid tank passage anti-vibration deformation data acquisition device, comprising a base, on which a data acquisition component and a data terminal are mounted; the data acquisition component includes a bracket, on which a line laser and an industrial camera are mounted, and on which an IMU unit is mounted.

[0009] According to the present invention, the vertical shaft rigid tank passage anti-vibration deformation data acquisition device has a support foot at the bottom of the base; the support foot is made of magnetic material.

[0010] According to the present invention, the data terminal of the vertical shaft rigid tank passage anti-vibration deformation data acquisition device is a laptop computer.

[0011] According to the present invention, the vertical shaft rigid tank passage anti-vibration deformation data acquisition device has a base made of magnetic material; and a magnetic block is fixed at the bottom of the support.

[0012] A detection method includes the following steps: S1, Install the data acquisition device on the top of the cage and adjust its position so that the data acquisition component is close to the target cage passage to be detected. S2, the cage is in operation, and the data acquisition device continuously measures the surface profile of the cage guideway; specifically... A line laser is used to project linear laser stripes onto the surface of a target tank passage to be inspected. Industrial cameras are used to capture images of laser stripes. The IMU unit is used to acquire attitude change information of the cage during the operation of the data acquisition device; S3, reconstruct the original tank passage point cloud; Transmit the data collected by the line laser, industrial camera and IMU unit to the data terminal; The data terminal processes the laser stripe image and extracts the center line of the laser stripe; After the center line of the laser stripe, the calibration parameters and laser plane equation obtained from the industrial camera are used to convert the two-dimensional image coordinates into three-dimensional point coordinates in the camera coordinate system; one frame of laser stripe image is restored as one frame of tank passage point cloud; For each frame of laser stripe image captured by the industrial camera, the operation of extracting the center line of the laser stripe and reconstructing the three-dimensional coordinates is repeatedly performed to obtain a multi-frame tank passage point cloud distributed along the tank passage direction. These multi-frame tank runway point clouds are arranged in the order of acquisition to form a continuous three-dimensional original tank runway point cloud sequence; S4 performs rotational compensation on the original can-way point cloud, specifically: Calculate the inertial attitude change acquired by the IMU unit of each frame of the tank passage point cloud and transform it into the camera coordinate system; then apply a rotation transformation opposite to the attitude change to each frame of the tank passage point cloud to restore it to the reference attitude. S5 performs translation compensation on the original tank passage point cloud; specifically: The rotation-compensated tank passage point cloud is segmented, and principal component analysis is performed on the segmented tank passage point cloud to establish a local coordinate system consistent with the geometric direction of the tank passage itself. Peak points are extracted from the point cloud of the cross-section of the tank passage in each frame to form peak trajectory; a sliding window is used to detect jitter intervals, and a trend curve and translation amount are constructed in conjunction with a weighted penalized least squares smoothing method; Apply reverse translation compensation to the tank passage point cloud of each frame jitter interval based on the translation amount; S6 performs deformation detection on the point cloud of the tank passage after rotation compensation and translation compensation, including protrusion detection, misalignment and gap detection.

[0013] According to the detection method of the present invention, before step S2, industrial camera calibration, laser plane calibration of line laser, and joint calibration of industrial camera and IMU unit are performed respectively; a unified conversion relationship between "image coordinates - camera coordinates - laser plane - IMU coordinates" is established.

[0014] According to the detection method of the present invention, the rotation compensation step in step S4 is as follows: Z1 reads the unit quaternion at the corresponding moment of each frame of point cloud as the inertial attitude to avoid angle singularity and interpolation instability during the continuous attitude change process; the unit quaternion q0 at the initial stationary moment or the steady running moment of the data acquisition device is used as the reference attitude. Z2, the point cloud of the tank passage reconstructed from any frame is Pf, where f represents the frame number; the point cloud of the f-th frame, calculate the inertial attitude change q relative to the reference attitude in that frame. rel (t)=q0×q(t) -1 Where q(t) is the inertial attitude quaternion corresponding to time t. -1 is the conjugate of q(t), and × is quaternion multiplication; Z3 transforms the attitude changes from the IMU coordinate system to the camera coordinate system, obtaining the rotation matrix R used for compensation in the camera coordinate system. C corr (t)=R C←I R I rel (t)R T C←I Among them, R C←IR is the extrinsic rotation matrix from the IMU coordinate system to the camera coordinate system; I rel (t) is related to q rel (t) is the rotation matrix in the IMU coordinate system corresponding to the rotation matrix; Z4, relative to coordinates P in the camera coordinate system c (t) Perform rotation compensation, and the compensated coordinates P` c (t)=R C corr (t)P c (t).

[0015] According to the detection method of the present invention, step S5, which involves performing translation compensation on the original tank passage point cloud, is as follows: X1, segment the point cloud of the tank passage after rotation compensation; perform Euclidean clustering based on the spatial proximity between points, and divide points that are spatially continuous and belong to the same surface of the tank passage into a tank passage point cloud cluster. X2 performs principal component analysis on the segmented single tank passage point cloud, establishes a local coordinate system consistent with the geometric direction of the tank passage itself, and transforms the segmented tank passage point cloud from the camera coordinate system to the local coordinate system. X3, extract the peak points of the point cloud of the cross-section of the tank passage in each frame to form the peak trajectory; For the point cloud of frame f, the point with the largest cross-sectional depth coordinate is searched in the local coordinate system of the tank passage, and it is taken as the peak point of the cross-section of that frame; the peak points of multiple consecutive frames are arranged in the acquisition order to form the peak trajectory s(f); X4 uses a sliding window to analyze the local changes in the peak trajectory; A sliding window of length W and step size Δ is set on the peak trajectory and slides gradually along the frame sequence direction; Calculate the local mean μ within each window i The intensity of sudden changes is characterized by the absolute difference between the local means of adjacent windows. Δμ i =|μ i+1 -μ i |;where Δμ i μ represents the intensity of the sudden change in the local mean between the i-th window and the (i+1)-th window. i+1 μ is the local mean of the (i+1)th window; i Let be the local mean of the i-th window; When the intensity of the sudden change between adjacent windows exceeds the preset threshold τ, it indicates that there is an abnormal jump in the peak trajectory of the region. The corresponding window is determined to be a candidate jitter window, and the corresponding frame segment is affected by the jitter of the cage. Adjacent or overlapping candidate jitter windows are merged to obtain the complete jitter interval. X5 uses a weighted penalized least squares smoothing method to estimate the peak trajectory; For normal frames that are not identified as jitter, a weight of 1 is assigned; for frames identified as jitter, a weight of α is assigned. Then the weight w(f) of the f-th frame is: ; Among them, S j This is the jitter range; Introducing an unknown smooth trajectory z(f), the objective function J(z) is constructed as follows: ; Where, Δ 2 z(f) is a second-order difference operator; λ is the smoothing intensity, and λ > 0; T is the sequence length; By introducing a smoothing constraint to solve the above objective function, we obtain the smooth trajectory z*(f). Then, the trend curve š(f) of the cross-section of the tank passage is: ; X6, calculate the difference between the original peak trajectory and the trend curve, as the translation amount of the can channel point cloud; When the f-th frame is in the jitter interval, the difference between its original peak trajectory and trend curve is calculated as the translation amount of the point cloud in that frame. The calculation formula is: r(f)=s(f)-š(f); where r(f) is the translation amount of the cross section of the tank passage in the local coordinate system in the f-th frame. X7, after obtaining the translation amount of the jitter frame, applies reverse translation compensation to the corresponding frame's tunnel point cloud and outputs the compensated tunnel point cloud.

[0016] In summary, the vertical shaft rigid cage guide anti-vibration deformation data acquisition device of the present invention fixes a line laser, an industrial camera, and an IMU unit on a rigid support. The cage guide contour is formed by active projection of the line laser, eliminating the need to rely on natural textures or stable corner points in the shaft, making it suitable for low-light, weakly textured vertical shaft environments. During operation, the cage is affected by fluctuations in wire rope tension, braking impacts, local collisions, or mechanical vibrations, causing short-term attitude changes and rigid displacements of the measuring device relative to the cage guide. Because the detection components are fixed on a rigid support, the attitude data collected by the IMU unit can be used to describe the attitude disturbance characteristics during this type of vibration process, providing a basis for subsequent data processing. The present invention also provides a detection method that compensates for rotational and translational distortions caused by short-term cage vibration, effectively reducing the impact of hoisting container vibration on point cloud measurements, allowing the cage guide point cloud to maintain good geometric continuity even under vibration conditions. The point cloud of the hoistway after rotation compensation and translation compensation can be used to calculate the hoistway protrusion height, joint misalignment, and inter-section gap value, providing a more reliable data foundation for the maintenance and safety assessment of the vertical shaft hoisting system. This is beneficial to improving the accuracy and stability of hoistway deformation measurement and meets the needs of millimeter-level accuracy assessment of rigid guide rail deformation. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the data acquisition device of the present invention; Figure 2 This is a schematic diagram illustrating the working principle of the data acquisition device of the present invention; Figure 3 This is a schematic diagram of the original tank passage point cloud reconstructed by the present invention; Figure 4 This is a schematic diagram of the point cloud of the canal passageway after the modification of this invention; Figure 5 This is a schematic diagram of the point cloud segmentation result of the canway in this invention; Figure 6 This is a schematic diagram of the jitter interval detection and smoothing trend curve replacement results of the present invention; Figure 7 This is a schematic diagram of the protrusion detection process according to Embodiment 1 of the present invention; Among them, (a) is the simulation image of the protrusion; (b) is the reconstructed original point cloud of the tank passage containing the protrusion; and (c) is the point cloud of the tank passage containing the protrusion and the detection image of the protrusion area after rotation compensation and translation compensation. Figure 8 for Figure 7 Measurement error curve of the embodiment; Figure 9 This is a schematic diagram of the misalignment and gap detection process in Embodiment 2 of the present invention; Among them, (a) is the simulation diagram of misalignment; (b) is the reconstructed original point cloud of the tank passage with misalignment; (c) is the point cloud of the tank passage with misalignment after rotation compensation and translation compensation; (d) is the point cloud model of the tank passage in the X direction of misalignment; (e) is the point cloud model of the tank passage in the Y direction of misalignment; (f) is the simulation diagram of gap; (g) is the reconstructed original point cloud of the tank passage with gap; (h) is the point cloud of the tank passage with gap after rotation compensation and translation compensation; (i) is the point cloud model of the tank passage in the Z direction of misalignment; and (j) is the point cloud model of the tank passage in the Y direction of gap. In the diagram: 1-line laser, 2-industrial camera, 21-IMU unit; 3-base, 31-support foot, 32-bracket, 321-magnetic block; 4-data terminal, 5-cage, 51-cage passage. Detailed Implementation

[0018] See Figure 1 The present invention provides a vertical shaft rigid tank passage anti-vibration deformation data acquisition device, including a base 3, on which a data acquisition component and a data terminal 4 are installed; the data acquisition component includes a bracket 32, on which a line laser 1 and an industrial camera 2 are provided, and an IMU unit 21 is installed on the industrial camera 2; The IMU unit 21 of the present invention is an inertial measurement unit.

[0019] Combination Figure 2 The base 3 is provided with a support foot 31 at its bottom; the data acquisition device is placed on top of the cage 5, so that the data acquisition component is close to the target passage 51 to be detected. As the cage 5 rises and falls, the data acquisition of the passage 51 on this side is completed and stored in the data terminal 4.

[0020] Furthermore, the data terminal 4 can be a laptop computer.

[0021] Even better, the support foot 31 is made of magnet material, which can be magnetically fixed to the top of the cage 5 and locked in the adjusted position.

[0022] Furthermore, the base 3 is made of magnetic material; a magnetic block 321 is fixed at the bottom of the bracket 32; the position of the data acquisition component can be finely adjusted and it can be attached to the base 3 by magnetic block 321.

[0023] As one embodiment, multiple data acquisition devices can be installed simultaneously on the top of the cage 5, with each device collecting data from one side of the cage passage 51. Data acquisition from all passages 51 is completed with a single lifting and lowering of the cage 5, improving detection efficiency. Under the same operating conditions, simultaneous data acquisition from all passages 51 ensures high data reliability.

[0024] The vertical shaft rigid guideway anti-vibration deformation data acquisition device of the present invention, by fixing the line laser 1, industrial camera 2, and IMU unit 21 on a rigid bracket 32, can effectively acquire data under conditions of wire rope tension fluctuations, braking impacts, guide wheel disturbances, and local impacts on the guideway 51 during the operation of the cage 5. This provides data support for reconstructing the guideway point cloud.

[0025] The present invention also provides a detection method based on the aforementioned data acquisition device, comprising the following steps: S1, Install the data acquisition device on the top of the cage 5 and adjust its position so that the data acquisition component is close to the target cage channel 51 to be detected. S2, the cage 5 is in operation, and the data acquisition device continuously measures the surface profile of the cage passage 51; specifically... Line laser 1 is used to project linear laser stripes onto the surface of the target tank channel 51 to be inspected; Industrial camera 2 is used to acquire laser stripe images; IMU unit 21 is used to acquire attitude change information of cage 5 during the operation of data acquisition device; Laser stripes are the contour lines formed by the intersection of the laser plane and the surface of the tank channel. Therefore, the changes in their shape in the image can reflect the changes in the geometry of the cross-section of the rigid tank channel.

[0026] IMU unit 21 continuously outputs inertial data of the data acquisition device during operation, including angular velocity, acceleration, and attitude information obtained from inertial calculations. Since the cage 5 is affected by wire rope tension fluctuations, braking impacts, local collisions, or mechanical vibrations during actual operation, the measuring device will undergo short-term attitude changes and rigid displacements relative to the rigid cage guide 51. Therefore, the attitude data collected by IMU unit 21 can be used to describe the attitude disturbance characteristics during this type of jitter process.

[0027] As a preferred approach, to ensure that the acquired image data, laser geometric information, and inertial attitude information participate in the point cloud reconstruction and distortion compensation of the tank passage 51 under a unified spatial reference, the following calibrations are performed before step S2: industrial camera 2 calibration, line laser 1 laser plane calibration, and joint calibration of industrial camera 2 and IMU unit 21; establishing a unified transformation relationship between "image coordinates—camera coordinates—laser plane—IMU coordinates". Industrial camera calibration ensures the accuracy of image geometric relationships, laser plane calibration ensures the correctness of line structured light 3D reconstruction, and joint calibration of industrial camera and IMU ensures that inertial attitude information can be accurately applied to the camera point cloud. These three together form the basis for anti-shake measurement.

[0028] S3, reconstruct the original tank passage point cloud; The data collected by the line laser 1, industrial camera 2 and IMU unit 21 are transmitted to the data terminal 4. Data terminal 4 processes the laser stripe image and extracts the center line of the laser stripe; After the center line of the laser stripe, the calibration parameters and laser plane equation obtained from the industrial camera are used to convert the two-dimensional image coordinates into three-dimensional point coordinates in the camera coordinate system; one frame of laser stripe image is restored as one frame of tank passage point cloud; The laser stripe centerline extraction and three-dimensional coordinate reconstruction operations are repeatedly performed on each frame of laser stripe image acquired by industrial camera 2 to obtain multiple frames of tank passage point cloud distributed along the direction of tank passage 51. These multi-frame tank runway point clouds are arranged in the order of acquisition to form a continuous three-dimensional original tank runway point cloud sequence; Optionally, each center pixel of the laser stripe in the image corresponds to an imaging ray originating from the optical center of the camera. After intersecting the known laser plane with this ray, the three-dimensional coordinates of the corresponding point on the surface of the tank can be obtained.

[0029] See Figure 3 During operation, the data acquisition device experiences short-term vibrations along with the cage 5, causing the industrial camera 2 and line laser 1 to undergo attitude changes and rigid displacements relative to the cage channel 51. In localized areas, rotational misalignment, translational jumps, and even breakage may occur between different frames.

[0030] The original cage duct point cloud reconstruction result includes both the true geometric shape of the cage duct 51 itself and the motion distortion caused by the shaking of the cage 5. This motion distortion mainly includes two types: one is rotational distortion caused by changes in the attitude of the data acquisition device, manifested as tilting, twisting, and angular misalignment between different frames of point clouds; the other is translational distortion caused by short-term rigid displacement of the data acquisition device, manifested as overall jumps in some frames of point clouds relative to the preceding and following frames. Therefore, rotational compensation and translational compensation steps are further introduced after the cage duct point cloud reconstruction to restore the true continuity of the cage duct point cloud.

[0031] S4 performs rotational compensation on the original can-way point cloud, specifically: The inertial attitude change acquired by the IMU unit 21 of each frame of the tank passage point cloud is calculated and transformed into the camera coordinate system; then, a rotation transformation opposite to the attitude change is applied to each frame of the tank passage point cloud to restore it to the reference attitude. The industrial camera 2 and the IMU unit 21 are fixedly mounted on the same bracket 32, and their relative pose remains unchanged during the detection process. Therefore, the pose change acquired by the IMU unit 21 can characterize the pose change of the industrial camera 2 and the line laser 1 caused by the shaking of the cage 5.

[0032] S5 performs translation compensation on the original tank passage point cloud; specifically: See Figure 5 The rotation-compensated point cloud of the tank passage is segmented, and principal component analysis is performed on the segmented point cloud of the tank passage to establish a local coordinate system consistent with the geometric direction of the tank passage 5 itself. Peak points are extracted from the point cloud of the cross-section of the tank passage in each frame to form peak trajectory; a sliding window is used to detect jitter intervals, and a trend curve and translation amount are constructed in conjunction with a weighted penalized least squares smoothing method; Apply reverse translation compensation to the tank passage point cloud of each frame jitter interval based on the translation amount; See Figure 4 S6 performs deformation detection on the point cloud of the tank passage after rotation compensation and translation compensation, including protrusion detection, misalignment and gap detection.

[0033] The principle of rotation compensation and translation compensation in this invention is as follows: the attitude information of IMU unit 21 compensates for the attitude change of each frame of the cannula point cloud; then, based on the rotationally compensated cannula point cloud, the translation amount is estimated by using the geometric continuity of the rigid cannula cross section, and rigid translation compensation is performed on the frame cannula point cloud affected by jitter.

[0034] When cage 5 is not jittering, the point cloud of the cage track in consecutive frames should be smoothly arranged along the direction of the cage track and reflect the true geometric changes of the cage track itself. When cage 5 jitters, the point cloud of the cage track in that frame can be understood as an attitude rotation and a rigid translation superimposed on the true point cloud of the cage track.

[0035] Therefore, the point cloud compensation for the tank passage does not directly compensate for the final measurement result, but rather processes the acquired data at the point cloud level for each frame. Specifically, this includes: First, eliminate rotational deviations caused by changes in camera attitude.

[0036] Second, eliminate the rigid translation deviation between frames that still exists after rotation compensation.

[0037] Third, the internal geometric relationship of the single-frame tank channel point cloud remains unchanged during the compensation process, avoiding the mistaken removal of actual protrusions, misalignments, and gaps in the tank channel as jitter errors.

[0038] This processing method enables the compensated tank passage point cloud to maintain overall continuity while preserving the true local defect features.

[0039] As one embodiment, the rotation compensation step in step S4 is as follows: Z1 reads the unit quaternion at the corresponding moment of each frame of point cloud as the inertial attitude to avoid angle singularity and interpolation instability during the continuous attitude change process; the unit quaternion q0 at the initial stationary moment or the steady running moment of the data acquisition device is used as the reference attitude. Z2, the point cloud of the tank passage reconstructed from any frame is Pf, where f represents the frame number; the point cloud of the f-th frame, calculate the inertial attitude change q relative to the reference attitude in that frame. rel (t)=q0×q(t) -1 Where q(t) is the inertial attitude quaternion corresponding to time t. -1 is the conjugate of q(t), and × is quaternion multiplication; Z3 transforms the attitude changes from the IMU coordinate system to the camera coordinate system, obtaining the rotation matrix R used for compensation in the camera coordinate system. C corr (t)=R C←I R I rel (t)R T C←I Among them, R C←I R is the extrinsic rotation matrix from the IMU coordinate system to the camera coordinate system; I rel (t) is related to q rel (t) is the rotation matrix in the IMU coordinate system corresponding to the rotation matrix; Z4, relative to coordinates P in the camera coordinate system c (t) Perform rotation compensation, and the compensated coordinates P` c (t)=R C corr (t)P c (t); The above processing is performed on all time frames, unifying the point clouds of the cage passage at different acquisition times to the same reference attitude standard. This step eliminates the tilting, rotation, and attitude misalignment of the point cloud caused by cage 5 shaking, ensuring that subsequent translation compensation is no longer affected by attitude deflection.

[0040] As one embodiment, step S5, which involves performing translation compensation on the original tank passage point cloud, is as follows: X1, segment the point cloud of the tank passage after rotation compensation; perform Euclidean clustering based on the spatial proximity between points, and divide points that are spatially continuous and belong to the same surface of the tank passage into a tank passage point cloud cluster. X2, perform principal component analysis on the segmented single tunnel point cloud, and establish a local coordinate system consistent with the geometric direction of tunnel 51 itself, that is, transform the segmented tunnel point cloud from the camera coordinate system to the local coordinate system. X3, extract the peak points of the point cloud of the cross-section of the tank passage in each frame to form the peak trajectory; For the point cloud of frame f, the point with the largest cross-sectional depth coordinate is searched in the local coordinate system of the tank passage, and it is taken as the peak point of the cross-section of that frame; the peak points of multiple consecutive frames are arranged in the acquisition order to form the peak trajectory s(f); X4 uses a sliding window to analyze the local changes in the peak trajectory; A sliding window of length W and step size Δ is set on the peak trajectory and slides gradually along the frame sequence direction; Calculate the local mean μ within each window i The intensity of sudden changes is characterized by the absolute difference between the local means of adjacent windows. Δμ i =|μ i+1 -μ i |;where Δμ i μ represents the intensity of the sudden change in the local mean between the i-th window and the (i+1)-th window. i+1 μ is the local mean of the (i+1)th window; i Let be the local mean of the i-th window; When the intensity of the sudden change between adjacent windows exceeds the preset threshold τ, it indicates that there is an abnormal jump in the peak trajectory of the region. The corresponding window is determined to be a candidate jitter window, and the corresponding frame segment is affected by the jitter of cage 5. Adjacent or overlapping candidate jitter windows are merged to obtain the complete jitter interval. X5 uses a weighted penalized least squares smoothing method to estimate the peak trajectory; For normal frames that are not identified as jitter, a weight of 1 is assigned; for frames identified as jitter, a weight of α is assigned. Then the weight w(f) of the f-th frame is: ; Among them, S j This is the jitter range; Introducing an unknown smooth trajectory z(f), the objective function J(z) is constructed as follows: ; Where, Δ 2 z(f) is a second-order difference operator; λ is the smoothing intensity, and λ > 0; T is the sequence length; By introducing a smoothing constraint to solve the above objective function, we obtain the smooth trajectory z*(f). Then, the trend curve š(f) of the cross-section of the tank passage is: ; X6, calculate the difference between the original peak trajectory and the trend curve, as the translation amount of the can channel point cloud; When the f-th frame is in the jitter interval, the difference between its original peak trajectory and trend curve is calculated as the translation amount of the point cloud in that frame. The calculation formula is: r(f)=s(f)-š(f); where r(f) is the translation amount of the cross section of the tank passage in the local coordinate system in the f-th frame. X7, after obtaining the translation amount of the jitter frame, applies reverse translation compensation to the corresponding frame's tunnel point cloud and outputs the compensated tunnel point cloud.

[0041] If the point cloud of the tank passage in a given frame shifts outwards, the peak trajectory will deviate positively from the trend curve; if the point cloud of the tank passage in a given frame shifts inwards, a negative deviation will occur. This deviation is the displacement that needs to be compensated. The translation amount in this invention is not obtained through double integration of IMU acceleration, but rather by extracting the peak points of the point cloud of the tank passage cross-section for each frame to form a peak trajectory; a sliding window is used to detect jitter intervals, and a low-frequency trend is constructed in conjunction with the weighted penalized least squares smoothing method (WPLS) to calculate the translation amount. This avoids the influence of zero bias, noise, and drift in inertial integration on millimeter-level displacement estimation.

[0042] See Figure 6 This demonstrates the effectiveness of the trajectory reconstruction strategy based on sliding window and WPLS in processing non-stationary signals. The green curve represents the peak trajectory of the stable segment, while the gray curve shows the peak trajectory exhibiting large-amplitude rigid displacement changes within two jitter segments, i.e., jitter intervals. After processing by the method of this invention, the abrupt change intervals marked with light red shading, i.e., jitter intervals, are identified, and the abnormal fluctuations within them are replaced by smooth trend curves shown by red line segments. The results show that this invention achieves targeted filtering of high-frequency jitter noise, while preserving the original geometric characteristics of the signal.

[0043] Combination Figure 4 Compared with the original reconstructed tank passage point cloud, the rotated and translated point cloud has the following characteristics: First, the point clouds of different frames of the tank channel are unified under the same reference attitude, which reduces the rotational misalignment caused by attitude shaking. Second, the abnormal position jumps of jittery frames relative to the preceding and following frames are compensated, and the continuity of the tank channel point cloud along the lifting direction is improved. Third, the compensation process mainly uses frame-level rigid compensation, which does not change the internal geometry of a single frame and can preserve the true local defects in the tank passage; fourth, the translation compensation does not mainly rely on IMU acceleration integration, which reduces the impact of integral drift on millimeter-level detection accuracy.

[0044] Therefore, the point cloud of the tank passage after rotation compensation and translation compensation can more accurately reflect the true spatial shape of the rigid tank passage and can serve as the data basis for subsequent protrusion detection, misalignment and gap detection. Example

[0045] See Figure 7 In this invention, a sample 51 with a raised structure is selected, and data acquisition and processing are performed using the aforementioned method. Figure 7 The highlighted red area represents the protrusion analysis window from the statistical point cloud of the tank passage after compensation. For each protrusion thickness, multiple scans and reconstructions were performed under different levels of jitter to calculate the measurement mean, average error, maximum error, and standard deviation.

[0046] Combination Figure 8 The measurement error statistics show that after rotation compensation and translation compensation, the measured values ​​for various protrusion thicknesses converge to the calibrated values, with the average error and standard deviation being less than 0.3 mm.

[0047] Example 2: Due to long-term operation, the guide rail may deviate from its original position, resulting in crack-like misalignment at the joint of tank passage 5 and a vertical gap between the upper and lower tank passages 5. To quantitatively evaluate and verify the accuracy of the point cloud after distortion compensation for misalignment and gap detection, this invention selects a tank passage 5 sample with misalignment and gap structures and uses the aforementioned method for data acquisition and processing.

[0048] The detection method involves setting a fixed point cloud curvature threshold to filter out planar points and extract edge region point clouds from the 3D tank passage point cloud model after rotation and translation compensation. By identifying the two tank passage point cloud contour lines with the smallest distance between the upper and lower tank passages, local tank passage point clouds of misalignment and gaps are extracted. The distance between the tank passage point clouds in the direction of the corresponding edge point pairs of the upper and lower tank passage point cloud contour lines is calculated, and the average value is finally used as the detection result of misalignment and gaps.

[0049] This embodiment of the invention simulates the detection of misalignment and gaps in a tank channel, using calipers with an accuracy of 0.1 mm for horizontal (X) measurement. dis ), vertical (Y) dis ), depth (Z) dis Multiple measurements in the direction were taken, and the X value was finally recorded. dis =1.50mm, Y dis =1.10mm and Z dis =4.90mm, gap value Y gap =9.30mm as the actual value. See also Figure 9 Point cloud models and error results of the guideway misalignment and gap in different directions were obtained. Finally, 10 independent measurements of guideway misalignment and gap were performed under different levels of vibration, and the mean, average error, maximum error and standard deviation of the measurements were calculated. The statistical results of the measurement errors are shown in Table 1.

[0050] The results show that after compensation, the average error calculated in each direction does not exceed 0.3 mm, which is within the allowable error range. Moreover, the standard deviation of the 10 measurement datasets is not greater than 0.1 mm, indicating that the compensated tank passage point cloud has high stability and measurement repeatability.

[0051] Table 1 Measurement Error .

[0052] Under vibration conditions, the reconstructed tank guide point cloud exhibits significant geometric deviations, making deformation detection impossible. However, by correcting the vibration-affected tank guide point cloud using the method of this invention, the true three-dimensional information of the tank guide surface can be effectively restored. The protrusion, misalignment, and gap detection performed on the tank guide point cloud corrected by this invention can reliably reflect the size of defects, with a measurement error of less than 0.3 mm, meeting the millimeter-level accuracy requirements for rigid guide rail deformation assessment.

[0053] This invention uses line laser active projection to form the tank channel profile, eliminating the need for natural downhole textures or stable corner points, making it suitable for low-light, low-texture vertical shaft environments. Compared to existing solutions, this invention extracts peak points from the point cloud of each frame's tank channel cross-section to form a peak trajectory; it employs a sliding window to detect jitter intervals and combines this with a weighted penalized least squares smoothing method to construct a trend curve and calculate the translation amount. This avoids the impact of zero bias and noise accumulation from the second integral of acceleration on millimeter-level measurements, and avoids the drift caused by simply using an IMU accelerometer for second integral calculation of translation.

[0054] In summary, this invention provides a vertical shaft rigid cage guide anti-vibration deformation data acquisition device, which fixes a line laser, an industrial camera, and an IMU unit on a rigid support. The cage guide contour is formed by active projection of the line laser, eliminating the need to rely on natural textures or stable corner points in the shaft, making it suitable for low-light, low-texture vertical shaft environments. During operation, the cage is affected by fluctuations in wire rope tension, braking impacts, local collisions, or mechanical vibrations, causing short-term attitude changes and rigid displacements of the measuring device relative to the cage guide. Because the detection components are fixed on a rigid support, the attitude data acquired by the IMU unit can be used to describe the attitude disturbance characteristics during this type of vibration process, providing a basis for subsequent data processing.

[0055] This invention also provides a detection method that compensates for rotational and translational distortions caused by short-term cage vibration. This effectively reduces the impact of hoisting container vibration on point cloud measurements, ensuring good geometric continuity of the cage guideway point cloud even under vibration conditions. The cage guideway point cloud after rotational and translational compensation can be used to calculate the cage guideway protrusion height, joint misalignment, and inter-segment gap values, providing a more reliable data foundation for the maintenance and safety assessment of vertical shaft hoisting systems. This improves the accuracy and stability of cage guideway deformation measurements and meets the millimeter-level accuracy assessment requirements for rigid guideway deformation.

[0056] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A data acquisition device for anti-vibration deformation of a rigid shaft guideway, characterized in that, The device includes a base on which a data acquisition component and a data terminal are mounted; the data acquisition component includes a bracket on which a line laser and an industrial camera are mounted, and an IMU unit is mounted on the industrial camera.

2. The vertical shaft rigid guideway anti-vibration deformation data acquisition device as described in claim 1, characterized in that, The base is provided with support feet at the bottom; the support feet are made of magnetic material.

3. The vertical shaft rigid guideway anti-vibration deformation data acquisition device as described in claim 1, characterized in that, The data terminal is a laptop computer.

4. The vertical shaft rigid guideway anti-vibration deformation data acquisition device as described in claim 1, characterized in that, The base is made of magnetic material; a magnetic block is fixed at the bottom of the bracket.

5. A detection method, characterized in that, Includes the following steps: S1, the data acquisition device as described in any one of claims 1 to 4 is installed on the top of the cage, and its position is adjusted so that the data acquisition component is close to the target cage passage to be detected; S2, the cage is in operation, and the data acquisition device continuously measures the surface profile of the cage guideway; specifically... A line laser is used to project linear laser stripes onto the surface of a target tank passage to be inspected. Industrial cameras are used to capture images of laser stripes. The IMU unit is used to acquire attitude change information of the cage during the operation of the data acquisition device; S3, reconstruct the original tank passage point cloud; Transmit the data collected by the line laser, industrial camera and IMU unit to the data terminal; The data terminal processes the laser stripe image and extracts the center line of the laser stripe; After the center line of the laser stripe, the calibration parameters and laser plane equation obtained from the industrial camera are used to convert the two-dimensional image coordinates into three-dimensional point coordinates in the camera coordinate system; one frame of laser stripe image is restored as one frame of tank passage point cloud; For each frame of laser stripe image captured by the industrial camera, the operation of extracting the center line of the laser stripe and reconstructing the three-dimensional coordinates is repeatedly performed to obtain a multi-frame tank passage point cloud distributed along the tank passage direction. These multi-frame tank runway point clouds are arranged in the order of acquisition to form a continuous three-dimensional original tank runway point cloud sequence; S4 performs rotational compensation on the original can-way point cloud, specifically: Calculate the inertial attitude change acquired by the IMU unit of each frame of the tank passage point cloud and transform it into the camera coordinate system; then apply a rotation transformation opposite to the attitude change to each frame of the tank passage point cloud to restore it to the reference attitude. S5 performs translation compensation on the original tank passage point cloud; specifically: The rotation-compensated tank passage point cloud is segmented, and principal component analysis is performed on the segmented tank passage point cloud to establish a local coordinate system consistent with the geometric direction of the tank passage itself. Peak points are extracted from the point cloud of the cross-section of the tank passage in each frame to form peak trajectory; a sliding window is used to detect jitter intervals, and a trend curve and translation amount are constructed in conjunction with a weighted penalized least squares smoothing method; Apply reverse translation compensation to the tank passage point cloud of each frame jitter interval based on the translation amount; S6 performs deformation detection on the point cloud of the tank passage after rotation compensation and translation compensation, including protrusion detection, misalignment and gap detection.

6. The detection method as described in claim 5, characterized in that, Before step S2, the industrial camera, the laser plane of the line laser, and the joint calibration of the industrial camera and IMU unit are performed respectively; a unified conversion relationship between "image coordinates - camera coordinates - laser plane - IMU coordinates" is established.

7. The detection method as described in claim 5, characterized in that, The rotation compensation step in step S4 is as follows: Z1 reads the unit quaternion at the corresponding moment of each frame of point cloud as the inertial attitude to avoid angle singularity and interpolation instability during the continuous attitude change process; the unit quaternion q0 at the initial stationary moment or the steady running moment of the data acquisition device is used as the reference attitude. Z2, the point cloud of the tank passage reconstructed from any frame is Pf, where f represents the frame number; the point cloud of the f-th frame, calculate the inertial attitude change q relative to the reference attitude in that frame. rel (t)=q0×q(t) -1 Where q(t) is the inertial attitude quaternion corresponding to time t. -1 is the conjugate of q(t), and × is quaternion multiplication; Z3 transforms the attitude changes from the IMU coordinate system to the camera coordinate system, obtaining the rotation matrix R used for compensation in the camera coordinate system. C corr (t)=R C←I R I rel (t)R T C←I Among them, R C←I R is the extrinsic rotation matrix from the IMU coordinate system to the camera coordinate system; I rel (t) is related to q rel (t) is the rotation matrix in the IMU coordinate system corresponding to the rotation matrix; Z4, relative to coordinates P in the camera coordinate system c (t) Perform rotation compensation, and the compensated coordinates P` c (t)=R C corr (t)P c (t).

8. The detection method as described in claim 7, characterized in that, The step S5, which involves performing translation compensation on the original tank passage point cloud, is as follows: X1, segment the point cloud of the tank passage after rotation compensation; perform Euclidean clustering based on the spatial proximity between points, and divide points that are spatially continuous and belong to the same surface of the tank passage into a tank passage point cloud cluster. X2 performs principal component analysis on the segmented single tank passage point cloud, establishes a local coordinate system consistent with the geometric direction of the tank passage itself, and transforms the segmented tank passage point cloud from the camera coordinate system to the local coordinate system. X3, extract the peak points of the point cloud of the cross-section of the tank passage in each frame to form the peak trajectory; For the point cloud of frame f, the point with the largest cross-sectional depth coordinate is searched in the local coordinate system of the tank passage, and it is taken as the peak point of the cross-section of that frame; the peak points of multiple consecutive frames are arranged in the acquisition order to form the peak trajectory s(f); X4 uses a sliding window to analyze the local changes in the peak trajectory; A sliding window of length W and step size Δ is set on the peak trajectory and slides gradually along the frame sequence direction; Calculate the local mean μ within each window i The intensity of sudden changes is characterized by the absolute difference between the local means of adjacent windows. Δμ i =|μ i+1 -μ i |;where Δμ i μ represents the intensity of the sudden change in the local mean between the i-th window and the (i+1)-th window. i+1 μ is the local mean of the (i+1)th window; i Let be the local mean of the i-th window; When the intensity of the sudden change between adjacent windows exceeds the preset threshold τ, it indicates that there is an abnormal jump in the peak trajectory of the region. The corresponding window is determined to be a candidate jitter window, and the corresponding frame segment is affected by the jitter of the cage. Adjacent or overlapping candidate jitter windows are merged to obtain the complete jitter interval. X5 uses a weighted penalized least squares smoothing method to estimate the peak trajectory; For normal frames that are not identified as jitter, a weight of 1 is assigned; for frames identified as jitter, a weight of α is assigned. Then the weight w(f) of the f-th frame is: ； Among them, S j This is the jitter range; Introducing an unknown smooth trajectory z(f), the objective function J(z) is constructed as follows: ； Where, Δ 2 z(f) is a second-order difference operator; λ is the smoothing intensity, and λ > 0; T is the sequence length; By introducing a smoothing constraint to solve the above objective function, we obtain the smooth trajectory z*(f). Then, the trend curve š(f) of the cross-section of the tank passage is: ； X6, calculate the difference between the original peak trajectory and the trend curve, as the translation amount of the can channel point cloud; When the f-th frame is in the jitter interval, the difference between its original peak trajectory and trend curve is calculated as the translation amount of the point cloud in that frame. The calculation formula is: r(f)=s(f)-š(f); where r(f) is the translation amount of the cross section of the tank passage in the local coordinate system in the f-th frame. X7, after obtaining the translation amount of the jitter frame, applies reverse translation compensation to the corresponding frame's tunnel point cloud and outputs the compensated tunnel point cloud.

Images