An elevator guide rail perpendicularity detection system

Abstract

The present application relates to the technical field of elevator maintenance, and discloses an elevator guide rail perpendicularity detection system, which is used to solve the problem that fixed baseline detection is easy to cause light spot off-target, unstable plane fitting, and unable to consider the precision and efficiency of elevator guide rail perpendicularity under narrow working surface, installation offset or local shielding, and comprises a rotary supporting module, the rotary supporting module is connected with a control module, the control module is connected with a result output module, the rotary supporting module is connected with a rotary driving module and a measurement base, the measurement base is connected with a multi-baseline ranging module, a posture acquisition module and a calibration module, the multi-baseline ranging module comprises at least three ranging units located at different calibration installation coordinates; the present application constructs different length candidate baselines through the at least three ranging units, selects the maximum effective baseline by combining calibration, posture compensation and quality parameter screening, and encrypts the re-measurement of low-quality sections, and considers the reliable landing of light spots, stable plane fitting, perpendicularity detection precision and detection efficiency.

Description

Technical Field

[0001] This invention relates to the field of elevator maintenance technology, and in particular to an elevator guide rail verticality detection system. Background Technology

[0002] Chinese invention patent CN106370146A discloses an elevator guide rail verticality detection system. This system mounts a first ranging laser, a second ranging laser, and a dual-axis tilt sensor on a motor-driven rotating support plate. The two ranging lasers rotate with the support plate to different detection angles to collect distance data from detection points on the guide rail's working surface, simultaneously collecting attitude data. The distance and tilt data are then sent to a control module via a serial port server. The control module calculates the guide rail verticality deviation. This system automatically processes the detection data without the need for manual plumbing or leveling. However, this solution uses two ranging lasers with a fixed center distance of 6 to 8 mm. When the measurement baseline is too large, it cannot guarantee that both laser points will fall simultaneously. On the same working surface of guide rails of different specifications, if the measurement baseline is too small, it will reduce the ability of the working surface fitting to represent the true posture. Moreover, scanning the entire column of guide rails with the same density according to the preset sampling interval requires the operator to input the start and end heights in advance for local detection. It cannot automatically change the effective distance combination and sampling resolution according to the actual situation. It is easy to use abnormal distance data that falls on the edge of the guide rail, connectors or the background of the well, making it difficult to balance detection speed and abnormal identification accuracy. This results in insufficient adaptability of the same equipment to different types of guide rails and complex well environments. The detection time is long when scanning with high density, while expanding the scanning interval may miss verticality abnormalities near guide rail joints, supports or local bending positions. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide an elevator guide rail verticality detection system, which constructs candidate baselines of different lengths by using at least three ranging units, and selects the maximum effective baseline by combining calibration, attitude compensation and quality parameters, densifies and re-measures low-quality sections, and takes into account reliable spot landing, stable plane fitting, verticality detection accuracy and detection efficiency.

[0004] To achieve the above objectives, the present invention provides the following technical solution: An elevator guide rail verticality detection system includes a rotating support module, which is connected to a control module via a data acquisition module. The control module is connected to a result output module. The rotating support module is connected to a rotating drive module and a measuring base. The measuring base is connected to a multi-baseline ranging module, an attitude acquisition module, and a calibration module. The rotating drive module includes an angle detection unit that acquires the base rotation angle, and the attitude acquisition module acquires the dual-axis tilt angle. The multi-baseline ranging module includes at least three ranging units located at different calibration installation coordinates. Each ranging unit forms a candidate ranging combination, creating at least two measuring baselines of different lengths. The calibration module determines the calibration installation coordinates and calibration measurement direction of each ranging unit based on reference plane data. The calibration coordinate spacing of the ranging unit is the measurement baseline of the candidate combination; the control module includes a working surface identification unit, a baseline selection unit, a graded scanning unit, and a verticality calculation unit. The control module calculates the spatial coordinates of the measurement point based on the collected data. The working surface identification unit fits the guide rail working surface based on the spatial coordinates. The quality parameters are the echo quality value, the standard deviation of the repeated ranging values, and the residual from the point to the guide rail working surface. The baseline selection unit selects the largest candidate combination of baselines from the ranging units that meet the quality threshold. The graded scanning unit scans at the first sampling interval and re-measures the sections that do not meet the quality threshold at the second sampling interval. The second sampling interval is smaller than the first sampling interval. The verticality calculation unit calculates the verticality deviation based on the change of the lateral position of the guide rail working surface with height.

[0005] As a further aspect of the present invention, the multi-baseline ranging module includes five ranging units arranged at a center distance of 4 mm along a straight mounting rail. The five ranging units are calibrated to obtain mounting positions of -8 mm, -4 mm, 0 mm, 4 mm and 8 mm, and form candidate measurement baselines of 4 mm, 8 mm, 12 mm and 16 mm.

[0006] As a further embodiment of the present invention, the multi-baseline ranging module includes two ranging unit groups that are perpendicular to each other and share a central ranging unit. The working surface identification unit fits the guide rail working surface according to the measurement points obtained from the pre-scan, the normal of the working surface is the normal of the fitting plane, and selects the ranging unit group that detects the top working surface or the side working surface according to the normal of the working surface.

[0007] As a further embodiment of the present invention, the calibration module includes a calibration drive and a position detection component. The calibration drive drives a preset calibration reference surface to at least two calibration positions. The position detection component collects the coordinates and normals of the calibration positions. The control module calculates the calibration installation coordinates and calibration measurement direction based on the distance, coordinates and normals of each calibration position.

[0008] As a further aspect of the present invention, the quality parameters also include the invalid point ratio, which is the ratio of the number of invalid points in the detection window to the total number of points. The quality threshold includes the lower limit of echo quality, the upper limit of standard deviation, the upper limit of residual, and the upper limit of invalid point ratio read by the control module from the parameter library. When any quality parameter does not meet the preset threshold, the working surface identification unit marks the measurement point or detection window as invalid.

[0009] As a further aspect of the present invention, the working surface identification unit uses the spatial coordinates of the measurement points to compensate for the dual-axis tilt angle, and performs clustering and robust plane fitting on the compensated measurement points to form the guide rail working surface with the largest continuous set of points whose residuals are not greater than the upper limit of the residuals. Then, the two working surface boundaries are determined based on the projection boundary of the point set in the working surface width direction.

[0010] As a further aspect of the present invention, the baseline selection unit excludes candidate combinations whose distance from the measurement point to any working surface boundary is less than the safe distance threshold, and then selects the candidate combination with the largest measurement baseline. The safe distance threshold is read from the parameter library by the control module based on the spot diameter and positioning error.

[0011] As a further aspect of the present invention, the graded scanning unit is set with three levels: coarse measurement, re-measurement and fine measurement. The control module reads the first, second and third sampling intervals and the first and second threshold groups of each level from the scanning parameter library according to the guide rail type input by the user. When the quality parameters do not meet the first threshold group, it enters the re-measurement level. When the re-measurement does not meet the second threshold group, it enters the fine measurement level.

[0012] As a further aspect of the present invention, the control module determines the abnormal section by the minimum and maximum values ​​of the measurement point height that triggers the retest or fine test, and adds a buffer length read from the scanning parameter library at both ends of the abnormal section to form a section to be retested. The measurement point height is the height component of the spatial coordinate. The control module controls the rotation drive module to return and scan the section to be retested according to the mapping relationship between the height and the rotation angle stored in the initial scan.

[0013] As a further aspect of the present invention, the control module indexes the measurement points by height and performs a weighted fitting of the spatial coordinates obtained by different combinations and levels within the same height window using measurement confidence to generate the fusion working surface position. The measurement confidence is a weighted sum of the normalized echo quality positive value, the standard deviation negative value, and the residual negative value obtained according to the parameter library weights. The verticality calculation unit calculates the verticality deviation based on the fusion working surface position at each height. Abnormal sections are determined by the extreme values ​​of the measurement point height and the buffer length. The measurement baseline and the scanning level serve as the combination source identifier and the scanning source identifier, respectively. The control module encodes the verticality deviation, abnormal sections, source identifiers, and measurement confidence into a result data frame. The result output module receives and parses the result data frame via the communication interface, displays the deviation curve and the abnormal section table, and stores the data and time.

[0014] It should be noted that this invention uses the rotating support and attitude acquisition as the general detection basis, and the measurement baseline that affects the accuracy of laser target landing and working surface fitting as a selectable variable. At least three displaced ranging units are used to form at least two different candidate measurement baselines. During detection, the control module calculates the spatial coordinates of the measurement point based on the ranging, attitude, and calibration data. The working surface identification unit determines the effective ranging unit on the guide rail working surface. The baseline selection unit prioritizes the largest effective baseline and switches the candidate combination when the measurement point deviates from the working surface or the quality parameter exceeds the limit. The graded scanning unit first scans the guide rail at the first sampling interval, and only reduces the sampling interval for low-quality sections for retesting. This allows the same system to adapt to different working surface widths, installation offsets, and local obstructions, improving data fitting stability while ensuring effective target landing of the measurement point, and reducing the detection time and data volume generated by high-density scanning in elevator guide rail verticality detection.

[0015] Compared with the prior art, the technical effects of the present invention are as follows: This invention automatically selects the maximum effective baseline from different measurement baselines based on the effective measurement range of the guide rail working surface, and switches combinations when the laser spot deviates from the working surface or the quality parameters exceed the limits. This not only avoids the long baseline spot missing the target, but also overcomes the unstable working surface fitting caused by the short baseline. At the same time, it reduces the sampling interval for retesting low-quality sections, so that measurement resources are allocated according to the state. It maintains the ability to identify anomalies while avoiding high-density scanning of the entire column, so that the system takes into account the target landing reliability, fitting accuracy and detection efficiency under various guide rails and obstruction environments. Attached Figure Description

[0016] Figure 1 This is a system block diagram of the present invention; Figure 2 The results of a three-dimensional simulation of the effect of the working surface width and installation offset on the measurement error of this invention are shown. Figure 3 The three-dimensional simulation results of the occlusion ratio and echo noise on the effective measurement point rate of this invention. Figure 4 The results are shown in the three-dimensional tracking simulation of local abnormal sections using the three methods of this invention. Detailed Implementation

[0017] 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. Example 1

[0018] like Figure 1As shown, the present invention proposes an elevator guide rail verticality detection system, comprising a rotating support module, a data acquisition module connected to a control module, a control module connected to a result output module, a rotating drive module connected to a measuring base, and a multi-baseline ranging module, an attitude acquisition module, and a calibration module. The rotating drive module includes an angle detection unit that acquires the base rotation angle, and the attitude acquisition module acquires the dual-axis tilt angle. The multi-baseline ranging module includes at least three ranging units located at different calibration installation coordinates. Each ranging unit forms a candidate ranging combination, creating at least two measuring baselines of different lengths. The calibration module determines the calibration installation coordinates and standard coordinates of each ranging unit based on the reference surface data. The measurement direction is determined, and the calibrated coordinate spacing between the two ranging units is used as the measurement baseline for the candidate combination. The control module includes a working surface identification unit, a baseline selection unit, a graded scanning unit, and a verticality calculation unit. The control module calculates the spatial coordinates of the measurement point based on the collected data. The working surface identification unit fits the guide rail working surface based on the spatial coordinates. The quality parameters are the echo quality value, the standard deviation of the repeated ranging values, and the residual from the point to the guide rail working surface. The baseline selection unit selects the largest candidate combination of baselines from the ranging units that meet the quality threshold. The graded scanning unit scans at the first sampling interval and re-measures the sections that do not meet the quality threshold at the second sampling interval. The second sampling interval is smaller than the first sampling interval. The verticality calculation unit calculates the verticality deviation based on the change in the lateral position of the guide rail working surface with height.

[0019] In practical implementation, the rotating support module is installed at the bottom of the elevator shaft, with the measuring base facing the guide rail. The rotating drive module drives the measuring base to rotate within the detection range. The angle detection unit collects the rotation angle, and the attitude acquisition module collects the dual-axis tilt angle. The multi-baseline ranging module uses three ranging units set at -4mm, 0mm, and 4mm positions along the width of the working surface, forming two 4mm short baseline combinations and one 8mm long baseline combination, realizing baseline switching. Before detection, the calibration module moves the reference plane to a known position. The control module determines the calibration installation coordinates and calibration measurement direction of the three ranging units based on the reference plane coordinates, normal, and ranging values, so that different... The combined data is converted to a unified coordinate system, which reduces installation errors and avoids position jumps caused by combination switching. During detection, the three ranging units simultaneously output ranging values ​​and echo quality values. The control module calculates the spatial coordinates of the measurement point by combining the calibrated installation coordinates, calibrated measurement direction, rotation angle, and dual-axis tilt angle. The horizontal coordinate system data is obtained by attitude compensation, reducing the influence of installation attitude on the verticality results. The working surface identification unit fits the guide rail working surface using the spatial coordinates obtained from adjacent angles and calculates the echo quality value, the standard deviation of three consecutive ranging values, and the residual from the point to the guide rail working surface. The ranging units that meet the quality threshold are identified as valid ranging units, while those falling on the edge of the guide rail are excluded. Abnormal points on the edge, pressure guide plate, or wellbore background; when all three ranging units are effective, the baseline selection unit selects two outer ranging units to form an 8mm long baseline to expand the measurement scale of the working surface and reduce the influence of ranging noise. If any outer ranging unit is ineffective due to a narrow working surface, installation offset, or partial obstruction, it switches to the middle ranging unit to form a 4mm short baseline with another outer ranging unit, so that the two light spots continue to fall on the same working surface. This does not only increase sensor redundancy, but also dynamically selects between target landing reliability and fitting stability according to the actual working surface condition, and synergistically solves the contradiction between the easy miss of long baselines and the unstable fitting of short baselines; the graded scanning unit first uses 10 The first sampling interval of 0mm scans the entire column of guide rails. When any quality parameter does not meet the threshold, its height is marked as the section to be retested and retested at a second sampling interval of 20mm. This ensures that normal sections maintain low data volume while abnormal sections obtain higher data density, reducing detection time and data volume while maintaining the ability to identify local anomalies. The verticality calculation unit calculates the verticality deviation based on the lateral position change of the working surface of each height guide rail and outputs the result. This allows the system to adapt to different working surface widths, installation offsets, and occlusion conditions. Through the synergistic effect of unified calibration, attitude compensation, baseline selection, and on-demand retesting, it solves the problem that fixed baseline and fixed sampling schemes are difficult to achieve simultaneously. Example 2

[0020] In Example 1, the three ranging units are located at -4mm, 0mm, and 4mm along the width of the guide rail working surface, forming two 4mm candidate combinations and one 8mm candidate combination. However, the 8mm long baseline has only one possible combination. When one outer ranging unit fails, it can only switch directly from 8mm to 4mm, lacking an intermediate scale. When both outer ranging units fail simultaneously, or when the central ranging unit and any outer ranging unit fail simultaneously, only one effective ranging unit remains, and the working surface cannot be calculated. The probability of the above failure states increases when the working surface is narrow, the measuring base is laterally offset, or the guide plate is pressed, or oil stains and reflective areas form continuous obstructions. Example 1, due to the lack of spare combinations of equal length for the long baseline, struggles to simultaneously achieve noise suppression and spot target landing.

[0021] This embodiment establishes a coordinate system for the measurement base: the x-axis is along the width of the guide rail working surface, the y-axis is along the nominal measurement direction towards the guide rail, and the z-axis is along the vertical direction. Five ranging units L1-L5 are set up, with nominal installation coordinates of (-6mm, 0mm, 0mm), (-3mm, 0mm, 0mm), (0mm, 0mm, 0mm), (3mm, 0mm, 0mm), and (6mm, 0mm, 0mm). The five ranging units form ten candidate combinations, creating four levels of baselines: 3mm, 6mm, 9mm, and 12mm. The same baseline has multiple backup combinations at different positions. As long as any two effective ranging units remain, the system selects the largest baseline from the four levels formed by the current effective subset. Specific failure modes can lead to same-level replacement or cross-level switching. This ranging unit structure increases the probability of forming effective combinations, expands the state space of the visible spot subset-selectable baseline scale, and enables closed-loop coordination between quality discrimination, maximum effective baseline selection, and hierarchical scanning.

[0022] In practice, the five ranging units use the same type of laser triangulation rangefinder, with a single ranging resolution of 0.01mm and a sampling frequency of 100Hz. The emission axes of each ranging unit are arranged in parallel with a lateral spacing of 3mm. L1 and L5 form the longest baseline of 12mm, L1 and L4, and L2 and L5 form a 9mm baseline, and combinations with two installation positions apart form a 6mm baseline. Adjacent units form the shortest baseline of 3mm. The 12mm baseline is used in sections with wide working surfaces, small installation offsets, and stable echoes, reducing the sensitivity of surface normal estimation to single-point noise with a larger spatial scale. The 3mm or 6mm baseline is used in narrow working surfaces or edge-obstructed sections to reduce the lateral space required for two effective light spots to fall simultaneously within the guide rail working surface. The 9mm baseline provides an alternative path close to the longest baseline when one of the outer ranging units fails, thereby eliminating the selection gap caused by the three ranging units only having two settings: 4mm and 8mm.

[0023] Each ranging unit continuously acquires three distance values ​​at the same rotation angle and simultaneously outputs the echo quality value calculated from the received light intensity. The control module uses the arithmetic mean of the three distance values ​​as the current ranging value of the ranging unit and the sample standard deviation of the three distance values ​​as the stability index of repeated ranging. The larger the echo quality value, the more reliable the target echo; the smaller the standard deviation, the better the repeatability. Both, together with the residual of the subsequent working surface, constitute the quality parameter, ensuring that edge echoes, specular reflections, oil scattering, and background echoes are no longer included in the fitting with equal weight, preventing significantly distorted distance values ​​from propagating to the verticality calculation. The quality parameter also serves as a common trigger for combination selection and scan level switching, coordinating hardware redundancy and sampling resource allocation under the same observable state.

[0024] Before testing, planar calibration plates with known coordinates and normals were placed at distances of 0.8m, 1.2m, and 1.6m from the measuring base, respectively. These calibration plates were positioned in the reference orientation, at +5° and -5° around the vertical axis, and at +5° and -5° around the horizontal axis, respectively. The calibration module read the distance measurements from five distance measuring units in each orientation. An overdetermined equation was established using the intersection of the measurement ray and the reference plane as constraints. The least squares method was used to obtain the actual calibration installation coordinates and unit calibration measurement directions for L1-L5. Adding the orientation around the horizontal axis makes the vertical components of the installation coordinates and optical axis observable. Multi-position, multi-orientation calibration thus separates installation translation errors from optical axis direction errors. When switching candidate combinations, the calibration coordinate spacing is used to avoid misinterpreting manufacturing and assembly errors as lateral changes in the guide rail, providing a continuous data basis for the dynamic switching of multiple baselines in elevator guide rail verticality measurement.

[0025] The data acquisition module sends the task number, timestamp, rotation angle, dual-axis tilt angle, five-channel ranging values, and five-channel echo quality values ​​to the control module. For the i-th ranging unit, the control module calculates the measurement point pi+diui in the base coordinate system based on its calibration installation coordinates pi, calibration measurement direction ui, and current ranging value di. Then, it multiplies this by a rotation matrix composed of rotation angle, roll angle, and pitch angle, and superimposes the reference coordinates of the measuring base in the horizontal coordinate system to obtain the spatial coordinates of the measurement point. This compensates for system errors caused by inaccurate leveling of the measuring base and slight tilt of the rotation axis, and ensures that data from different sensors, different rotation angles, and different scanning levels have a unified meaning of height and lateral position.

[0026] The working face identification unit uses all spatial coordinates obtained from three adjacent rotation angles as its neighborhood. It first eliminates background points in the shaft according to the preset normal range of the guide rail working face, and then uses a random sampling consistency method to obtain an initial working face used only for residual gating. In this embodiment, the echo quality value is normalized to 0-1 according to the calibration upper and lower limits in the parameter library, and 0.60, 0.18mm, and 0.20mm are used as example thresholds for echo quality, repeatability distance standard deviation, and point-to-initial working face residual, respectively. The thresholds are determined by the 95th percentile of the calibration samples. When all three conditions are met simultaneously, the ranging unit is recorded as valid, excluding abnormal points on the guide rail edge, pressure plate, and background wall. After baseline selection, only the measurement points obtained from adjacent angles in the selected combination are used for the final local working face and lateral position estimation, ensuring that the selected baseline actually participates in the verticality calculation, rather than just being used as an incidental output parameter.

[0027] Let the first and the The calibration installation coordinates of each ranging unit are as follows: and candidate combination baseline The baseline selection unit is first selected from pairwise combinations of all currently valid ranging units. The largest possible combination is selected to ensure compliance with the maximum effective baseline selection rule, and then the best combination is chosen only among multiple equal-length maximum baseline combinations based on combination confidence. The positive normalized value for echo quality is... The back normalized standard deviation is 1. The residual inverse normalized value is The upper and lower limits are derived from calibration samples and stored in the parameter database; the confidence level of the ranging unit. The combined confidence level is calculated by weighting the three factors by 0.40, 0.30, and 0.30 respectively. This represents the average confidence level of the two ranging units. Hysteresis hold is used only between the maximum baseline combinations of equal lengths to prevent jitter in the spare equal length combination from creating false steps.

[0028] The hierarchical scanning unit first scans the 0-30m guide rail at a first sampling interval of 100mm. When there are fewer than two effective ranging units, no candidate combination can be formed, the quality parameter of any unit in the selected combination exceeds the limit, the combination confidence is lower than the example threshold of 0.58, or the difference in lateral position between adjacent fusion points exceeds the example threshold of 0.25mm, the current height is recorded in the abnormal candidate set. The minimum and maximum values ​​of the heights of consecutive abnormal points are used as the initial boundaries of the abnormal section, and a buffer length of 100mm is added at each end. The current section is then retested at a second sampling interval of 20mm. If the retest still does not meet the threshold, the result is marked as invalid without background echo compensation. This avoids unnecessary retesting due to the individual exceeding of the limit by an unselected unit among the five units. The five ranging units increase the probability of obtaining a valid combination during retesting and reduce repeated scanning caused by consecutive invalid results.

[0029] The verticality calculation unit first uses a coarse scan measurement point height ±50mm or a fine scan measurement point height ±10mm as the same height window. At each scan moment, the baseline selection unit selects a combination and its adjacent rotation angle measurement points to fit the final local working surface, obtaining the lateral position of the elevator guide rail working surface at the scan moment. Then, the lateral positions of different scan moments, different selected combinations, and different scan levels within the same height window are normalized and weighted according to the corresponding combination confidence level to obtain the fused lateral position. For any 5m height interval, the maximum value of the fused lateral position minus the minimum value is used as the verticality deviation, and the 30m full height variation curve is retained, taking into account both positive and negative extreme values ​​within the interval, avoiding comparison only with the starting point of the interval and missing the deviation in the opposite direction.

[0030] The control module uses the task serial number written by the host computer at the start of detection as the task number, the z-coordinate of the measurement point after transformation to the horizontal coordinate system as the measurement point height, and the combined confidence level as the fusion weight of the verticality estimate at each scanning time. The control module associates the verticality deviation, abnormal sections, measurement baseline, scan level, and measurement confidence level with the task number and measurement point height, and encodes them into a result data frame. The verticality deviation originates from the verticality calculation unit, the abnormal sections originate from the extreme height of abnormal points and the 100 mm buffer length, the measurement baseline originates from the Bi,j of the combination selected by the baseline selection unit, the scan level originates from the status of the graded scan unit, and the fused measurement confidence level is calculated from the ci,j of the participating data according to the fusion weight. The result output module uses the CRC polynomial check data frame stored in the parameter library, parses it in height order, and outputs the above parameters as numerical values, curves, and abnormal section markers, ensuring that each output quantity has a clear source.

[0031] To compare the prior art (referring to the methods used in the background art cited documents), Embodiment 1, and the improvements described in this embodiment, a Monte Carlo simulation model compatible with MATLAB syntax was established, with a random seed of 20250825 and a guide rail height of 30m. The three methods employed the same ranging noise, quality criterion, and coordinate error model, only changing the ranging unit position, candidate combination rules, and scanning strategy to avoid misinterpreting differences in device accuracy as structural improvements. The cited document method established a fixed two-point model with an 8mm center distance as disclosed in Embodiment CN106370146A. Embodiment 1 used three ranging units: -4mm, 0mm, and 4mm; Embodiment 2 used five ranging units: -6mm, -3mm, 0mm, 3mm, and 6mm. 8000 random trials were conducted for each typical operating condition. The numerical values ​​in the figures are used to compare the relative trends under the established models and do not replace prototype calibration and field testing data.

[0032] Under the above simulation conditions, we statistically analyzed whether two fixed ranging units are effective at the same time, and whether three or five ranging units can form at least one candidate combination that meets the quality threshold. We also calculated the average baseline and conditional root mean square error of the effective combination. The root mean square error of the three methods only counts the samples with effective combinations. When there is no effective combination, it is recorded as a rejection and no large error is artificially filled in. The typical working condition comparison data are shown in Table 1.

[0033] Table 1. Effective combinations and error data of the three methods under typical working conditions. Simulation conditions Detection methods Effective portfolio rate (%) Average selected baseline (mm) Verticality RMSE (mm) Normal working surface Citation document method 96.0 8.00 0.136 Normal working surface Example 1 Method 99.9 7.85 0.099 Normal working surface Example 2 Method 100.0 11.88 0.074 Narrow working face Citation document method 44.5 8.00 0.158 Narrow working face Example 1 Method 94.8 5.91 0.133 Narrow working face Example 2 Method 99.7 7.20 0.110 Horizontal offset Citation document method 45.1 8.00 0.161 Horizontal offset Example 1 Method 94.6 5.88 0.133 Horizontal offset Example 2 Method 100.0 8.70 0.100 Partial occlusion Citation document method 49.1 8.00 0.167 Partial occlusion Example 1 Method 79.0 6.46 0.135 Partial occlusion Example 2 Method 97.0 9.76 0.107 Combined unfavorable working conditions Citation document method 0.0 0.00 — Combined unfavorable working conditions Example 1 Method 47.1 4.00 0.191 Combined unfavorable working conditions Example 2 Method 77.1 4.86 0.177 As shown in Table 1, all three methods can complete the detection on the normal working surface. However, Example 2 utilizes a 12mm long baseline to reduce the conditional root mean square error. Under adverse working conditions such as lateral offset and composite conditions, the fixed two-point method in the cited document's example is prone to rejection due to a spot extending beyond the working surface. Example 1 can switch to a 4mm short baseline when one outer point fails, but there are still insufficient effective points when two outer points fail simultaneously, or when the center point and any outer point fail simultaneously. Example 2 can utilize five positions to form multiple backup combinations of the same level, and select the largest one from 12mm, 9mm, 6mm, or 3mm baselines based on the actual effective subset. This example expands the candidate combination set with a spatially dispersed five-point structure, and then balances target landing reliability and fitting stability using a quality threshold and the maximum baseline rule.

[0034] Further, the narrow working plane, lateral offset, occlusion ratio, and echo noise were set as functions that continuously varied with the 30m measurement height, concentrating unfavorable local conditions near the 16.2m abnormal convexity. The cited document's embodiments used a fixed height step of 8mm. Embodiments 1 and 2 first performed a coarse scan of 100mm, then only re-measured sections without effective combinations, with insufficient confidence in the selected combination, or abrupt changes in adjacent positions, using a 20mm step. The effective point rate, conditional error, abnormal missed detection rate, actual requested points, and estimated scan time were directly calculated using the same scanning program; the results are shown in Table 2. The scan time was uniformly set to 1s per stable measurement position, including 20% ​​for rotation and communication overhead. The confidence error was defined as the average absolute difference between the combination confidence and exp(-absolute error / 0.15mm). Since the cited document did not output the quality confidence, the corresponding cell was marked as "—".

[0035] Table 2. Comprehensive simulation results of the three methods under a 30m composite working condition. Detection methods Effective point rate (%) RMSE (mm) 95% error (mm) Abnormal missed detection rate (%) Number of sampling points Scan time (s) Confidence error (dimensionless) Citations 89.0 0.125 0.243 100.0 3751 4501.2 — Example 1 89.9 0.113 0.223 51.1 573 687.6 0.230 Example 2 95.5 0.093 0.180 31.1 397 476.4 0.186 As shown in Table 2, the cited document embodiment, due to its 8mm height step and stable waiting at each position, has the largest number of sampling points and estimated scanning time. The graded scanning in Embodiments 1 and 2 significantly reduces the amount of data in normal sections, with the structure of the five ranging units having a higher probability of retaining effective combinations in unfavorable sections. The anomaly miss rate in the table is calculated using the same true anomaly function and the same judgment threshold. Conditional RMSE and effective point rate are listed separately to avoid amplifying the error of a method by artificially assigning values ​​to failure points. The confidence error verifies whether the output confidence level changes with the actual error.

[0036] To visually observe the error distribution when the working surface width and installation offset change together, the width was set to 8-20mm, the lateral offset to -5 to 5mm, the local occlusion ratio to 10%, and the echo noise standard deviation to 0.07mm. A uniform evaluation penalty of 0.80mm was assigned to each rejected sample from all three methods. The equivalent RMSE was synthesized from the conditional RMSE and the rejection ratio to avoid using different penalty calibers. (See the 3D surface diagram.) Figure 2 Red, blue, and green represent the cited document method, the method of Example 1, and the method of Example 2, respectively.

[0037] Figure 2 The results show that when the working surface narrows and the installation offset increases, the red surface rises the fastest, indicating that there is a clear geometric target boundary in the fixed 8mm two-point structure; the blue surface can decrease due to the switching between 8mm and 4mm, but there is still a relatively steep change near the offset boundary; the green surface is lower and flat in most areas, because the five-point structure can select a combination with a larger span to suppress noise, and can also form a 3-9mm combination with the other two units located inside the working surface when the edge fails. The model results show that the multi-level baseline and the same-level backup combination can jointly reduce the coupling between the installation offset and the working surface width.

[0038] To examine the combined retention capability under echo degradation conditions, the effective width of the guide rail working surface was fixed at 14 mm, the lateral offset was fixed at 1.5 mm, and the local obstruction ratio was varied between 0-45%, while the standard deviation of echo noise was varied between 0.03-0.18 mm. Effective measurement points were determined according to the quality threshold, and a three-dimensional efficiency surface was generated. The results are shown in [Figure number missing]. Figure 3 .

[0039] Figure 3The results show that as the occlusion ratio and echo noise increase, the effective measurement point rate of all three methods decreases, but the green surface in Example 2 decreases the slowest. This is because as long as any two of the five units meet the quality threshold, a candidate combination can be formed, while the two fixed units must be effective simultaneously, and the three units are prone to having fewer than two effective points when adjacent occlusion occurs. More importantly, Example 2 does not use low-quality data to obtain continuous output on the surface, but instead feeds back the invalid state to the hierarchical scanning unit, re-acquiring spatial samples at smaller height intervals. This improves the efficiency of elevator data measurement while keeping the error under control, demonstrating the synergy between redundant point placement, quality gating, and local retesting.

[0040] Finally, the true lateral deviation of the 30m guide rail was set as the sum of a slow tilt term, a periodic bending term, and a local convexity term near 16.2m. Obstruction and high noise were superimposed in the 14.7-17.0m range. Three methods were used to calculate the lateral position and absolute measurement error at each height, generating... Figure 4 The three-dimensional trajectory is shown; all three color curves use the same real-world guide function and random seed.

[0041] Figure 4 In the results, the red curve shows the highest error in the local convexity and occlusion sections, the blue curve reduces the error through short baseline switching and 20mm retesting, and the green curve is closest to the actual change because it can select the longest effective combination at multiple locations and fuse the retest data according to the confidence level. The results show that Example 2 not only improves the effective point rate in a static sense, but also maintains the continuity of the coordinate chain when the abnormal height changes continuously, so that the actual guide rail convexity is not covered by the artifact of combination switching. At the same time, the boundary of the abnormal section is automatically formed by the extreme value of the height exceeding the limit plus the buffer length, and the result data frame retains the baseline, scan level and confidence level, which can distinguish between the actual structural anomaly and the deterioration of the measurement conditions, and provide a basis for elevator track calibration decision.

[0042] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0043] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An elevator guide rail verticality detection system, comprising a rotating support module, the rotating support module being connected to a control module via a data acquisition module, the control module being connected to a result output module, the rotating support module being connected to a rotating drive module and a measuring base, and the measuring base being connected to a multi-baseline ranging module, an attitude acquisition module, and a calibration module, characterized in that: The rotation drive module includes an angle detection unit that acquires the base rotation angle, and an attitude acquisition module that acquires the dual-axis tilt angle. The multi-baseline ranging module includes at least three ranging units located at different calibration installation coordinates. The ranging units form candidate ranging combinations in pairs and form at least two different lengths of measurement baselines. The calibration module determines the calibration installation coordinates and calibration measurement direction of each ranging unit based on the reference surface data. The distance between the calibration coordinates of two ranging units is the measurement baseline of the candidate combination. The control module includes a working surface identification unit, a baseline selection unit, a graded scanning unit, and a verticality calculation unit. The control module calculates the spatial coordinates of the measurement point based on the acquired data. The working surface identification unit fits the guide rail working surface based on the spatial coordinates. The quality parameters are the echo quality value, the standard deviation of the repeated ranging values, and the residual from the point to the guide rail working surface. The baseline selection unit selects the largest candidate combination of baselines from the ranging units that meet the quality threshold. The graded scanning unit scans at the first sampling interval and re-measures the sections that do not meet the quality threshold at the second sampling interval. The second sampling interval is smaller than the first sampling interval. The verticality calculation unit calculates the verticality deviation based on the change in the lateral position of the guide rail working surface with height.

2. The elevator guide rail verticality detection system according to claim 1, characterized by: The multi-baseline ranging module includes five ranging units arranged at a center distance of 4 mm along a straight mounting rail. The five ranging units are calibrated to obtain mounting positions of -8 mm, -4 mm, 0 mm, 4 mm and 8 mm, forming candidate measurement baselines of 4 mm, 8 mm, 12 mm and 16 mm.

3. The elevator guide rail verticality detection system of claim 1, wherein: The multi-baseline ranging module includes two ranging unit groups that are perpendicular to each other and share a central ranging unit. The working surface identification unit fits the guide rail working surface according to the measurement points obtained from the pre-scan. The normal of the working surface is the normal of the fitting plane, and the ranging unit group that detects the top working surface or the side working surface is selected according to the normal of the working surface.

4. The elevator guide rail verticality detection system of claim 1, wherein: The calibration module includes a calibration drive and a position detection component. The calibration drive drives the preset calibration reference surface to at least two calibration positions. The position detection component collects the coordinates and normals of the calibration positions. The control module calculates the calibration installation coordinates and calibration measurement direction based on the distance, coordinates and normals of each calibration position.

5. The elevator guide rail verticality detection system of claim 1, wherein: The quality parameters also include the invalid point ratio, which is the ratio of the number of invalid points in the detection window to the total number of points. The quality thresholds include the lower limit of echo quality, the upper limit of standard deviation, the upper limit of residual, and the upper limit of invalid point ratio read by the control module from the parameter library. When any quality parameter does not meet the preset threshold, the working surface identification unit will mark the measurement point or detection window as invalid.

6. The elevator guide rail verticality detection system of claim 5, wherein: The working surface identification unit uses the spatial coordinates of the measurement points to compensate for the dual-axis tilt angle, and performs clustering and robust plane fitting on the compensated measurement points. The maximum continuous set of points with residuals not exceeding the upper limit of residuals is used to form the guide rail working surface. Then, the two working surface boundaries are determined based on the projection boundary of this set of points in the working surface width direction.

7. The elevator guide rail verticality detection system of claim 6, wherein: The baseline selection unit excludes candidate combinations where the distance from the measurement point to any working surface boundary is less than the safe distance threshold, and then selects the candidate combination with the largest measurement baseline. The safe distance threshold is read from the parameter library by the control module based on the spot diameter and positioning error.

8. The elevator guide rail verticality detection system of claim 1, wherein: The hierarchical scanning unit sets three levels of coarse measurement, repeated measurement and precise measurement, and the control module reads first, second and third sampling intervals and first and second threshold groups of each level from a scanning parameter library according to a guide rail category input by a user, enters the repeated measurement level when a quality parameter does not meet the first threshold group, and enters the precise measurement level when a quality parameter does not meet the second threshold group after repeated measurement.

9. The elevator guide rail verticality detection system of claim 8, wherein: The control module determines an abnormal section by using minimum and maximum values of a measurement point height triggering repeated measurement or precise measurement, increases a buffer length read from the scanning parameter library at both ends of the abnormal section to form a section to be measured, and controls the rotary driving module to return and scan the section to be measured according to a mapping relationship between the height and the rotation angle stored by initial scanning, where the measurement point height is a height component of a spatial coordinate.

10. The elevator rail verticality detection system of claim 9, wherein: The control module indexes the measurement point height, weights and fits spatial coordinates obtained by different combinations and levels in a same height window according to a measurement confidence to generate a fused working face position, where the measurement confidence is a weighted sum of a normalized echo quality positive value, a standard deviation reverse value and a residual error reverse value according to a parameter library weight; the verticality calculation unit calculates a verticality deviation according to the fused working face position of each height, the abnormal section is determined by extreme values of the measurement point height and the buffer length, the measurement baseline and the scanning level are respectively taken as a combination source identifier and a scanning source identifier; the control module encodes the verticality deviation, the abnormal section, the source identifier and the measurement confidence into a result data frame, the result output module receives and analyzes the result data frame through a communication interface, displays a deviation curve and an abnormal section table, and stores data and time.

Images