A method and system for high-precision continuous measurement of the flatness of a point laser array plane

Abstract

The application relates to the technical field of industrial automation detection, and particularly relates to a point laser array plane flatness high-precision continuous measurement method and system. The method obtains height measurement data and motion axis position data of a point laser array, performs collinear calibration on the point laser array in combination with a preset calibration reference, obtains corresponding calibration parameters, and constructs point cloud data. The calibration parameters are used for spatial error correction of the point cloud data, and calibration point cloud data is obtained; based on a deep learning model, abnormal point elimination and curved surface fitting processing are performed, and a target measurement curved surface is generated. Further, a reference plane is constructed by using a traditional geometric algorithm, flatness deviation between the target measurement curved surface and the reference plane is calculated, and finally, flatness results and three-dimensional point cloud coordinate results are output. The application realizes high-precision continuous measurement of plane flatness, and effectively improves the stability, anti-interference capability and three-dimensional visual analysis capability of the measurement results.

Description

Technical Field

[0001] This application relates to the field of industrial automation testing technology, and in particular to a high-precision continuous measurement method and system for the flatness of a point laser array plane. Background Technology

[0002] In related technologies, with the rapid development of precision manufacturing, semiconductor processing, electronic equipment assembly, large-scale sheet metal processing, and automated inspection technologies, flatness inspection has become an important technical means to ensure product processing accuracy, assembly consistency, and structural stability. This type of inspection system typically uses contact measuring devices, laser displacement sensors, structured light equipment, or machine vision systems to collect height, contour, and surface morphology data of the measured plane. Then, it combines these data with geometric fitting algorithms to calculate and analyze flatness deviations, determining whether the measured plane has warping, depressions, protrusions, or localized deformation, thereby providing data support for product quality inspection, error compensation, and subsequent processing.

[0003] However, existing flatness measurement methods still face the challenge of balancing accuracy and efficiency in large-scale continuous measurements. For contact-based measurement methods such as coordinate measuring machines (CMMs) and dial indicators, data acquisition typically requires point-by-point contact with probes, resulting in low measurement efficiency, limited detection range, and potential damage to the measured surface. For non-contact measurement methods such as line laser scanning, structured light projection, and single-point laser scanning, encoder triggering or fixed-point sampling on a motion platform is usually required. This can lead to accumulated stitching errors, increased motion synchronization errors, and excessively long detection cycles in large-scale planar inspections. Furthermore, traditional machine vision measurement methods are susceptible to changes in lighting, differences in material reflectivity, and environmental noise interference, resulting in insufficient measurement stability.

[0004] In summary, the technical problems existing in the relevant technologies need to be improved. Summary of the Invention

[0005] The main objective of this application is to propose a high-precision continuous measurement method and system for the flatness of a point laser array plane, so as to achieve high-precision continuous measurement of the flatness of the plane and effectively improve the stability, anti-interference ability and three-dimensional visualization analysis capability of the measurement results.

[0006] To achieve the above objectives, one aspect of this application proposes a high-precision continuous measurement method for the flatness of a point laser array plane, the method comprising the following steps: Acquire the height measurement data and motion axis position data corresponding to the point laser array; Based on a preset calibration benchmark, the point laser array is subjected to collinear calibration to obtain the corresponding calibration parameters; Based on the height measurement data and the motion axis position data, the corresponding point cloud data is constructed; Based on the calibration parameters, spatial error correction processing is performed on the point cloud data to obtain the corresponding calibration point cloud data; Based on the calibration point cloud data, anomaly removal and surface fitting are performed using a deep learning model to obtain the corresponding target measurement surface. Based on the target measurement surface, a reference plane is constructed using traditional geometric algorithms, and the flatness deviation between the target measurement surface and the reference plane is calculated. Based on the flatness deviation, the corresponding flatness result and three-dimensional point cloud coordinate result are output.

[0007] In some embodiments, the height measurement data includes the real-time height value, sampling timestamp information, and sensor identification information corresponding to the laser displacement sensor at each point; the motion axis position data includes the position coordinate information, motion speed information, and position sampling time information corresponding to the servo motion axis.

[0008] In some embodiments, the collinear calibration process of the point laser array based on a preset calibration benchmark to obtain corresponding calibration parameters includes: Control the point laser array to move to the preset calibration position and place the standard plane in the corresponding measurement area; Based on the standard plane, collect the reference height data corresponding to the laser displacement sensor at each point; Based on the reference height data, collinear calibration processing is performed on the laser displacement sensors at each point to obtain the corresponding sensor calibration data. Based on the sensor calibration data, zero-point compensation correction is performed on the laser displacement sensors at each point to obtain the corresponding calibration parameters.

[0009] In some embodiments, constructing corresponding point cloud data based on the height measurement data and the motion axis position data includes: Based on the height measurement data and the motion axis position data, a corresponding spatial coordinate mapping relationship is established; Based on the spatial coordinate mapping relationship, the height values ​​corresponding to the laser displacement sensors at each point are processed by spatial coordinate transformation to obtain the corresponding discrete point cloud data. Based on the discrete point cloud data, continuous stitching processing is performed according to the direction of motion to obtain the corresponding continuous point cloud data; Based on the continuous point cloud data, coordinate unification processing is performed to obtain the corresponding point cloud data.

[0010] In some embodiments, performing spatial error correction processing on the point cloud data based on the calibration parameters to obtain corresponding calibration point cloud data includes: Based on the calibration parameters, spatial offset correction processing is performed on the point cloud data to obtain the corresponding initial corrected point cloud data; Based on the initial corrected point cloud data, coordinate compensation processing is performed on the installation deviation of the laser displacement sensor at each point to obtain the corresponding coordinate corrected point cloud data. Based on the coordinate-corrected point cloud data, error compensation processing is performed on the trajectory deviation of the motion axis to obtain the corresponding error-compensated point cloud data. Based on the error-compensated point cloud data, coordinate unification processing is performed to obtain the corresponding calibration point cloud data.

[0011] In some embodiments, the step of performing outlier removal and surface fitting processing using a deep learning model based on the calibration point cloud data to obtain the corresponding target measurement surface includes: Based on the calibration point cloud data, the corresponding spatial coordinate features and height change features are extracted to obtain the corresponding surface feature data; Based on the surface feature data, outlier detection and removal are performed to obtain the corresponding valid point cloud data. Based on the effective point cloud data, a continuous curve fitting process is performed using a deep learning model to obtain the corresponding curve point cloud data. Based on the curve point cloud data, surface fitting processing is performed to obtain the corresponding target measurement surface.

[0012] In some embodiments, the step of performing continuous curve fitting processing using a deep learning model based on the effective point cloud data to obtain the corresponding curve point cloud data includes: Based on the effective point cloud data, point cloud sequence partitioning is performed according to the direction of motion to obtain the corresponding local point cloud data. Based on the local point cloud data, the corresponding height change features and spatial continuity features are extracted to obtain the corresponding curve feature parameters; Based on the curve feature parameters, a deep learning model is used to perform continuous curve fitting to obtain the corresponding initial curve data. Based on the initial curve data, curve smoothing compensation processing is performed to obtain the corresponding continuous curve data; Based on the continuous curve data, the height information corresponding to any position in the spatial coordinate system is obtained, and the corresponding curve point cloud data is obtained.

[0013] In some embodiments, the step of constructing a reference plane based on the target measurement surface using conventional geometric algorithms and calculating the flatness deviation between the target measurement surface and the reference plane includes: Based on the target measurement surface, extract the corresponding surface point cloud data; Based on the surface point cloud data, a corresponding reference plane is constructed using a least squares fitting algorithm; Based on the reference plane, the distance deviation of each point in the curved surface point cloud data is calculated to obtain the corresponding plane deviation data. Based on the plane deviation data, determine the corresponding maximum and minimum deviation values; Based on the maximum deviation value and the minimum deviation value, the corresponding flatness deviation is calculated.

[0014] In some embodiments, calculating the distance deviation corresponding to each point in the surface point cloud data based on the reference plane to obtain the corresponding plane deviation data includes: Based on the surface point cloud data, obtain the spatial coordinate data corresponding to each point; Based on the spatial coordinate data and the reference plane, calculate the normal distance value corresponding to each point; Based on each of the aforementioned normal distance values, corresponding distance deviation data is generated; Based on the distance deviation data, deviation distribution statistical processing is performed to obtain the corresponding planar deviation data.

[0015] To achieve the above objectives, another aspect of this application proposes a high-precision continuous measurement system for the planar flatness of a point laser array, the system comprising: The data acquisition module is used to acquire the height measurement data and motion axis position data corresponding to the point laser array; The collinear calibration module is used to perform collinear calibration on the point laser array based on a preset calibration benchmark to obtain the corresponding calibration parameters; The point cloud construction module is used to construct corresponding point cloud data based on the height measurement data and the motion axis position data; An error correction module is used to perform spatial error correction processing on the point cloud data based on the calibration parameters to obtain the corresponding calibration point cloud data. The first calculation module is used to perform outlier removal and surface fitting processing based on the calibration point cloud data using a deep learning model to obtain the corresponding target measurement surface. The second calculation module is used to construct a reference plane based on the target measurement surface using traditional geometric algorithms, and to calculate the flatness deviation between the target measurement surface and the reference plane. The result output module is used to output the corresponding flatness result and three-dimensional point cloud coordinate result based on the flatness deviation.

[0016] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method.

[0017] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0018] To achieve the above objectives, another aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0019] The embodiments of this application include at least the following beneficial effects: This application provides a method and system for high-precision continuous measurement of the flatness of a point laser array. This scheme acquires the height measurement data and motion axis position data corresponding to the point laser array, and combines collinear calibration processing and spatial error correction processing to achieve unified coordinate mapping and high-precision spatial compensation of multi-point laser measurement data. This effectively reduces the impact of point laser array installation deviation, motion axis jitter error, and sensor zero-point drift on the measurement results, thereby improving the spatial consistency and measurement stability of point cloud data. Furthermore, by constructing point cloud data and using a deep learning model to perform outlier removal and surface fitting processing, it is possible to adaptively identify and optimize discrete noise points, defect points, and local distortion data generated during the measurement process, improving the continuous surface reconstruction capability and data robustness in complex surface scenarios. Simultaneously, by combining traditional geometric algorithms to construct a reference plane and calculating the flatness deviation between the target measurement surface and the reference plane, it is possible to balance the nonlinear fitting capability of deep learning algorithms with the high interpretability of traditional geometric algorithms, achieving high-precision quantitative analysis of flatness results. Compared with traditional single-point discrete measurement methods, this invention has the advantages of high continuous measurement efficiency, high measurement accuracy, strong anti-interference ability, and strong three-dimensional visualization ability, and can be widely used in precision machining, industrial inspection, and high-precision planar quality assessment. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating a high-precision continuous measurement method for the planar flatness of a point laser array provided in an embodiment of this application. Figure 2 This is a schematic diagram of curve fitting after continuous curve fitting processing based on a deep learning model, provided in an embodiment of this application. Figure 3 This is a schematic diagram of the XYZ coordinates of the fitted surface grid points generated based on a deep learning model, provided in an embodiment of this application. Figure 4 This is a depth map of the fitted surface provided in the embodiments of this application; Figure 5 This is a schematic diagram of the three-dimensional fitting of the target measurement surface and the reference plane provided in the embodiments of this application; Figure 6 This is a schematic diagram of a module for a high-precision continuous measurement method for the flatness of a point laser array plane provided in an embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0022] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0023] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0025] Before providing a detailed description of the embodiments of this application, some of the nouns and terms involved in the embodiments of this application will be explained first. The nouns and terms involved in the embodiments of this application are subject to the following interpretations.

[0026] A point laser array is a laser measurement array formed by arranging multiple point laser displacement sensors at a preset interval. It is used to simultaneously measure the height of multiple positions on the plane being measured, thereby improving the continuity and measurement efficiency of plane detection. A point laser displacement sensor is a sensor that performs non-contact measurement of height changes on a target surface based on the principle of laser reflection, and is used to obtain height information of the corresponding position on the measured plane; The Grubbs criterion is a statistical method for detecting outliers in sample data. Based on the sample mean and standard deviation, it calculates the deviation of the data point to be detected from the sample mean and compares it to a threshold value at a preset significance level to determine whether the data point is an outlier. This method is typically suitable for detecting single outliers in small datasets that follow a normal distribution. Deep learning models are data processing models built on multi-layer neural network structures, used to perform outlier identification, continuous curve fitting, and surface reconstruction on point cloud data. The least squares fitting algorithm is a mathematical algorithm that fits a reference plane by minimizing the sum of squared errors between the measurement points and the target fitting model.

[0027] This application provides a method and system for high-precision continuous measurement of the flatness of a point laser array. This method acquires the height measurement data and motion axis position data corresponding to the point laser array, and combines collinearity calibration and spatial error correction processing to achieve unified coordinate mapping and high-precision spatial compensation of multi-point laser measurement data. This effectively reduces the impact of point laser array installation deviation, motion axis jitter error, and sensor zero-point drift on the measurement results, thereby improving the spatial consistency and measurement stability of the point cloud data. Furthermore, by constructing point cloud data and using a deep learning model to perform outlier removal and surface fitting processing, it is possible to adaptively identify and optimize discrete noise points, defect points, and local distortion data generated during the measurement process, improving the continuous surface reconstruction capability and data robustness in complex surface scenarios. Simultaneously, by combining traditional geometric algorithms to construct a reference plane and calculating the flatness deviation between the target measurement surface and the reference plane, it can balance the nonlinear fitting capability of deep learning algorithms with the high interpretability of traditional geometric algorithms, achieving high-precision quantitative analysis of flatness results. Compared with traditional single-point discrete measurement methods, this invention has the advantages of high continuous measurement efficiency, high measurement accuracy, strong anti-interference ability, and strong three-dimensional visualization ability, and can be widely used in precision machining, industrial inspection, and high-precision planar quality assessment.

[0028] This application provides a high-precision continuous measurement method for the planar flatness of a point laser array, relating to the field of industrial automation testing technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or vehicle-mounted terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application implementing the high-precision continuous measurement method for the planar flatness of a point laser array, but is not limited to the above forms.

[0029] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0030] Figure 1 This is an optional flowchart of a high-precision continuous measurement method for the planar flatness of a point laser array provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S1 to S7: S1: Acquire the height measurement data and motion axis position data corresponding to the point laser array; the height measurement data includes the real-time height value, sampling timestamp information and sensor identification information corresponding to each point laser displacement sensor; the motion axis position data includes the position coordinate information, motion speed information and position sampling time information corresponding to the servo motion axis.

[0031] In this embodiment, before measurement, the plane to be measured is fixed on a carrier plate or measuring platform, and the plane is kept stable during measurement by lifting, limiting, or clamping structures to avoid the workpiece shaking affecting the height sampling results. After measurement is started, the servo motion axis drives the point laser array to move at a constant speed along a preset motion direction. The point laser displacement sensor array performs continuous sampling without triggering according to a preset sampling frequency and moves continuously at a constant speed along the preset direction. The movement speed can be set to 100 mm / s. The sampling frequency can be set to no less than 10 kHz. Since the sampling process does not rely on encoder fixed-point triggering or platform stop triggering, the sensor array can continuously output height measurement data in continuous motion, thereby improving the detection efficiency of large-size planes.

[0032] The height measurement data includes real-time height values ​​collected by each laser displacement sensor at each sampling time, sampling timestamp information, and sensor identification information. The real-time height value represents the distance change between the corresponding laser displacement sensor and a local area of ​​the measured plane; the sampling timestamp information identifies the time when the height value was generated; and the sensor identification information distinguishes the installation positions of different laser displacement sensors within the array. For arrays consisting of multiple sensors, the identification information of each sensor can be pre-bound to its lateral installation coordinates within the array, allowing for the subsequent conversion of height values ​​collected by different sensors into point cloud data in the same spatial coordinate system.

[0033] Simultaneously, the servo motion axis outputs position coordinate information, motion speed information, and position sampling time information in real time. The position coordinate information is used to indicate the current position of the sensor array in the motion direction, the motion speed information is used to determine whether the current scanning process maintains a uniform and stable speed, and the position sampling time information is used to synchronize and match with the sensor sampling timestamp.

[0034] S2: Based on the preset calibration benchmark, the point laser array is collinearly calibrated to obtain the corresponding calibration parameters; Among them, based on a preset calibration benchmark, the point laser array is subjected to collinear calibration to obtain the corresponding calibration parameters, including: The control point laser array is moved to the preset calibration position, and the standard plane is placed in the corresponding measurement area; Based on a standard plane, collect reference height data corresponding to the laser displacement sensor at each point; Based on the reference height data, collinear calibration processing of the laser displacement sensors at each point is performed to obtain the corresponding sensor calibration data. Based on the sensor calibration data, zero-point compensation correction is performed on the laser displacement sensors at each point to obtain the corresponding calibration parameters.

[0035] In this embodiment, the measuring equipment is first initialized, and the servo motion axis is controlled to drive the point laser array to a preset calibration position, while a standard plane is placed in the corresponding measurement area. The standard plane is preferably a grade 0 marble standard plane to ensure high planar accuracy of the calibration reference. After the point laser array moves to the calibration station, the laser measurement points of each point laser displacement sensor fall on the standard plane, ensuring that all point laser displacement sensors perform calibration based on the same height reference. Subsequently, the calibration sampling process is initiated, continuously collecting the reference height data corresponding to each point laser displacement sensor. The reference height data includes the corresponding height measurement value, sampling time information, and sensor identification information.

[0036] Because the installation height, angle, and zero-point drift of each laser displacement sensor may vary during installation, the height measurements output by different sensors may differ even when facing the same standard plane. Therefore, collinear calibration is performed on each laser displacement sensor based on reference height data. Specifically, using a unified height reference corresponding to the standard plane as a reference, the reference height data collected by different laser displacement sensors are compared and analyzed to determine the height offset of each sensor relative to the unified reference plane. Corresponding sensor calibration data is then generated based on this offset. This sensor calibration data may include zero-point offset values, installation height compensation values, and array position compensation values ​​to achieve consistency and uniformity in the measurement reference of each laser displacement sensor.

[0037] Subsequently, based on the sensor calibration data, zero-point compensation correction was performed on the laser displacement sensors at each point. This ensured that the original height measurements output by each laser displacement sensor during subsequent measurements could be corrected in real time based on the corresponding compensation parameters, thereby eliminating zero-point errors and installation errors between different laser displacement sensors. After zero-point compensation correction, the data output by each laser displacement sensor was unified to the same collinear reference, ultimately yielding the corresponding calibration parameters. This provides a unified measurement benchmark for subsequent point cloud construction, spatial error correction, and flatness calculation.

[0038] S3: Construct the corresponding point cloud data based on height measurement data and motion axis position data; Specifically, based on height measurement data and motion axis position data, corresponding point cloud data is constructed, including: Based on height measurement data and motion axis position data, establish the corresponding spatial coordinate mapping relationship; Based on the spatial coordinate mapping relationship, the height values ​​corresponding to the laser displacement sensors at each point are transformed into spatial coordinates to obtain the corresponding discrete point cloud data. Based on discrete point cloud data, continuous stitching processing is performed according to the direction of motion to obtain the corresponding continuous point cloud data; Based on continuous point cloud data, coordinate unification processing is performed to obtain the corresponding point cloud data.

[0039] In this embodiment, the collected height measurement data is first matched with the motion axis position data to establish a spatial coordinate mapping relationship. This mapping relationship associates the height value of each laser displacement sensor at a specific sampling time point with the corresponding motion axis position coordinates. Simultaneously, combined with the sensor's lateral installation position in the array, a preliminary coordinate system for each sampling point in three-dimensional space is formed. This mapping ensures the precise position of each height measurement value in the X (motion direction), Y (array width direction), and Z (height direction) three-dimensional coordinates.

[0040] Subsequently, based on the established spatial coordinate mapping relationship, the height value corresponding to each point laser displacement sensor is processed by spatial coordinate transformation, converting it into discrete point cloud data. Specifically, the X-axis coordinate is determined by the real-time position of the motion axis, the Y-axis coordinate is determined by the sensor's installation position in the array, and the Z-axis coordinate is obtained by adding calibration deviation correction to the height measurement value. Each sampling point generates a three-dimensional spatial coordinate, thus forming a discrete point cloud set covering the measured plane.

[0041] Next, based on the discrete point cloud data, the sampling points are continuously stitched together according to the direction of motion to form continuous point cloud data, enabling the continuous representation of the height change of the measured plane along the scanning direction. Finally, coordinate unification processing is performed on the continuous point cloud data, including unified adjustments to the array spacing, motion axis deviation, and zero-point correction, so that the entire point cloud data is in a unified spatial coordinate system, ultimately obtaining complete three-dimensional point cloud data, providing a foundation for subsequent spatial error correction, surface fitting, and flatness calculation.

[0042] S4: Based on the calibration parameters, perform spatial error correction processing on the point cloud data to obtain the corresponding calibration point cloud data; Specifically, based on calibration parameters, spatial error correction processing is performed on the point cloud data to obtain the corresponding calibration point cloud data, including: Based on the calibration parameters, spatial offset correction processing is performed on the point cloud data to obtain the corresponding initial corrected point cloud data; Based on the initial corrected point cloud data, coordinate compensation processing is performed on the installation deviation of the laser displacement sensor at each point to obtain the corresponding coordinate corrected point cloud data. Based on the coordinate-corrected point cloud data, error compensation processing is performed on the trajectory deviation of the motion axis to obtain the corresponding error-compensated point cloud data; Based on the error-compensated point cloud data, coordinate unification processing is performed to obtain the corresponding calibration point cloud data.

[0043] In this embodiment, calibration parameters are first invoked to perform spatial offset correction processing on the point cloud data. Since different point laser displacement sensors may have zero-point offsets, initial height offsets, and array installation position errors during installation, the height coordinates and planar coordinates in the point cloud data are uniformly corrected based on the corresponding zero-point compensation parameters and array position parameters. This ensures that the point cloud data collected by different point laser displacement sensors are mapped to a unified spatial reference, thereby obtaining the corresponding initially corrected point cloud data. This processing eliminates the overall spatial offset problem caused by different installation references among multiple sensors.

[0044] Subsequently, based on the initial corrected point cloud data, coordinate compensation processing was performed on the installation deviations corresponding to the laser displacement sensors at each point. Specifically, since there may be installation tilt angle errors, lateral spacing errors, and vertical installation deviations between the laser displacement sensors at each point, compensation and correction were performed on the X, Y, and Z coordinates in the point cloud data in conjunction with the corresponding sensor installation parameters to improve the spatial continuity between the multi-sensor point clouds. At the same time, consistency correction was performed on the boundary overlapping areas between different sensor acquisition areas to reduce coordinate abrupt changes at the array splicing positions, thereby obtaining the corresponding coordinate-corrected point cloud data.

[0045] Next, based on the coordinate-corrected point cloud data, error compensation processing is performed on the trajectory deviation of the motion axis. Since the servo motion axis may experience speed fluctuations, guide rail vibrations, or trajectory offsets during continuous motion, dynamic error compensation is performed on the point cloud coordinates along the motion direction, combining the motion axis position data and motion speed information, to reduce the impact of mechanical motion errors on the point cloud accuracy. Finally, based on the error-compensated point cloud data, coordinate unification processing is performed, uniformly mapping the coordinate reference of the entire point cloud data so that all point cloud data are in a unified three-dimensional spatial coordinate system. This ultimately yields the corresponding calibrated point cloud data, providing a high-precision data foundation for subsequent deep learning surface fitting and flatness calculations.

[0046] S5: Based on the calibration point cloud data, use a deep learning model to perform outlier removal and surface fitting to obtain the corresponding target measurement surface; Specifically, based on the calibration point cloud data, a deep learning model is used to perform outlier removal and surface fitting to obtain the corresponding target measurement surface, including: Based on the calibration point cloud data, the corresponding spatial coordinate features and height change features are extracted to obtain the corresponding surface feature data; Based on surface feature data, outlier detection and removal are performed to obtain the corresponding valid point cloud data; Based on the effective point cloud data, a deep learning model is used to perform continuous curve fitting processing to obtain the corresponding curve point cloud data. Based on the curve point cloud data, surface fitting processing is performed to obtain the corresponding target measurement surface.

[0047] Specifically, based on valid point cloud data, a continuous curve fitting process is performed using a deep learning model to obtain the corresponding curve point cloud data, including: Based on the effective point cloud data, point cloud sequence partitioning is performed according to the direction of motion to obtain the corresponding local point cloud data. Based on local point cloud data, the corresponding height change features and spatial continuity features are extracted to obtain the corresponding curve feature parameters; Based on the curve feature parameters, a deep learning model is used to perform continuous curve fitting to obtain the corresponding initial curve data. Based on the initial curve data, curve smoothing compensation is performed to obtain the corresponding continuous curve data; Based on continuous curve data, the height information corresponding to any position in the spatial coordinate system is obtained, and the corresponding curve point cloud data is obtained.

[0048] In this embodiment, spatial coordinate features and height variation features are first extracted from the calibration point cloud data to obtain the corresponding surface feature data. The spatial coordinate features include the coordinate distribution relationships of each point cloud data point in the direction of motion, array width, and height. The height variation features include the height difference between adjacent sampling points, local height fluctuation trends, and the height continuity between different laser displacement sensors. Through the above feature extraction, the subsequent deep learning model can not only identify the height value of a single sampling point but also determine the true undulation state of the measured plane by combining the spatial continuity relationship of the surrounding point cloud.

[0049] Subsequently, outlier detection and removal are performed based on the surface feature data to obtain the corresponding valid point cloud data. Specifically, due to factors such as dust obstruction, surface reflection, instantaneous sensor jumps, or motion vibrations during the measurement process, some point cloud height values ​​may significantly deviate from the continuous trend of the surrounding point clouds. Based on the local height distribution, the height difference between adjacent points, and the degree of statistical dispersion, abrupt changes are identified, and point cloud data that clearly does not conform to the spatial continuity characteristics are removed. After outlier removal, the remaining valid point cloud data can more accurately reflect the actual shape of the measured plane, avoiding interference from outliers in subsequent continuous curve fitting and surface fitting.

[0050] Specifically, the Grubbs criterion is used to remove outliers, filter out gross errors and jump values ​​in the measurement data, and avoid outliers from contaminating the accuracy of surface fitting.

[0051] Calculate the sample mean using the discrete point height data Z after coordinate transformation: ; In the formula, This represents the sample mean of discrete height data, i.e., the average of all measured height values; n represents the total number of measurement points involved in the statistical calculation. Indicates the first The height values ​​corresponding to each measurement point.

[0052] And calculate the sample standard deviation: ; In the formula, The standard deviation represents the sample size and is used to characterize the dispersion of height data across all measurement points. This indicates the degree of freedom correction term; Indicates the first The deviation of the height value of each measurement point from the sample mean.

[0053] The Grubbs statistic is as follows: ; In the formula, Indicates the first The Grubbs statistic corresponding to each measurement point is used to determine whether the point is an outlier.

[0054] Given a significance level (which needs to be adjusted appropriately according to the actual measured plane). Zagrubbs critical value : like : Identified as an outlier, removed; like : Retain valid measurement points.

[0055] Furthermore, based on the effective point cloud data, point cloud sequence segmentation is performed according to the direction of motion to obtain the corresponding local point cloud data. Specifically, the effective point cloud collected by the laser displacement sensor at the same point during continuous motion can be divided into a set of local point cloud sequences extending along the direction of motion, and their height change features and spatial continuity features are extracted to obtain the corresponding curve feature parameters. Subsequently, the curve feature parameters are input into a deep learning model, which learns the nonlinear mapping relationship between the motion direction coordinates and height values, and performs continuous curve fitting processing on the discrete point cloud sequence to obtain the corresponding initial curve data.

[0056] refer to Figure 2 As shown, Figure 2 This is a schematic diagram of the curve fitting process after continuous curve fitting based on a deep learning model in this application. It is mainly used to illustrate the process of forming a continuous curve after the discrete height data collected by the point laser displacement sensor is processed by the deep learning model. Figure 2The red dots represent discrete point cloud data acquired by the original point laser displacement sensor, with each discrete point corresponding to the height measurement value at different spatial locations; the blue curve represents a continuous curve fitted based on a deep learning model, used to characterize the continuous height change trend of the measured plane in the direction of motion; the green markers represent the maximum and minimum points in the fitted curve, used to reflect the convex and concave positions of the local area.

[0057] Furthermore, Figure 2 The continuous curves shown not only enable the transformation from discrete points to continuous curves but also provide height information for any position in the spatial coordinate system. Compared to traditional polynomial fitting or interpolation algorithms, continuous curves obtained based on deep learning models exhibit stronger robustness to noise, outliers, and complex nonlinear variations, effectively reducing the impact of local anomalies on curve continuity. Based on the continuous curve data corresponding to multiple point laser displacement sensors, further multi-line surface fitting processing can be performed to construct the corresponding target measurement surface, providing a foundation for subsequent flatness deviation calculation and 3D point cloud reconstruction.

[0058] After obtaining the initial curve data, curve smoothing compensation is performed to reduce the impact of local noise, sampling jitter, and small-scale abnormal fluctuations on the curve continuity, resulting in corresponding continuous curve data. Based on the continuous curve data, the height information corresponding to any movement position in the spatial coordinate system can be obtained, thus obtaining the corresponding curve point cloud data. Finally, the curve point cloud data corresponding to multiple point laser displacement sensors are comprehensively fitted along the array width direction, and multiple continuous curves are further formed into a continuous measurement surface, ultimately obtaining the corresponding target measurement surface. This target measurement surface not only preserves the overall undulation shape of the measured plane, but also improves the fitting accuracy of complex surfaces through deep learning models, providing a data foundation for subsequent reference plane construction and flatness deviation calculation.

[0059] In this embodiment, the deep learning model employs a multilayer perceptron (MLP) or fully connected neural network structure, using the spatial coordinates and height variation features of effective point cloud data as input, and outputting continuous height prediction values ​​for corresponding locations. Model construction first standardizes the input data to eliminate the influence of different scales; then, network layers are defined, for example, the input layer receives (x,y) coordinate information, and a nonlinear activation function ReLU is introduced through multiple hidden layers to model the nonlinear relationships of complex surfaces. During network training, the mean squared error (MSE) loss function is used to minimize the error between the model output and the true height value, and the backpropagation algorithm is used to iteratively update the weights and bias parameters, enabling the model to learn the spatial continuous mapping relationship of discrete point cloud data. After training, the model can output the corresponding predicted height for any continuous coordinate in the input domain, realizing the transformation from discrete point clouds to continuous curves and surfaces, and possessing robustness to noise and local anomalies, thus providing an accurate continuous data foundation for the generation of the target measurement surface.

[0060] refer to Figure 3 As shown, Figure 3 This is a schematic diagram of the XYZ coordinates of the fitted surface grid points generated based on the deep learning model in this application. It is mainly used to illustrate the target measurement surface formed after the discrete point cloud data collected by the point laser array is processed by continuous curve fitting and surface fitting. Figure 3 The horizontal axis represents the spatial position of the measured plane in the direction of movement, the vertical axis represents the spatial position in the direction of array width, and the color and height changes represent the changes in the height value at the corresponding position. Figure 3 The grid points are continuous surface point cloud data calculated by a deep learning model. Multiple discrete sampling points are transformed into a regular three-dimensional spatial surface after curve compensation and surface reconstruction, which can intuitively reflect the overall shape changes of the measured plane.

[0061] Figure 3 The XYZ coordinates of the grid points shown are the fitted surface point cloud data, where each grid point corresponds to a predicted height value at a spatial location. Preferably, regularized point cloud data can be generated according to a preset grid density, for example, dividing the X and Y directions into 100 sampling points each, thereby forming 100×100 surface grid point data to improve the accuracy of subsequent flatness calculations.

[0062] Furthermore, Figure 3The target measurement surface shown not only reflects the overall height change trend of the measured plane, but also obtains the height information corresponding to any position in the spatial coordinate system, providing a foundation for subsequent reference plane construction and flatness deviation calculation. Compared with traditional polynomial fitting or interpolation methods, the continuous surface obtained based on the deep learning model has stronger nonlinear fitting ability and noise resistance, effectively reducing the impact of local outliers, random noise, and complex surface changes on the fitting results. Furthermore, through the regularized grid point output method, it is possible to further generate 3D point cloud maps, surface depth maps, and flatness distribution maps for subsequent product quality analysis, error compensation, and defect location.

[0063] S6: Based on the target measurement surface, a reference plane is constructed using traditional geometric algorithms, and the flatness deviation between the target measurement surface and the reference plane is calculated; Specifically, based on the target measurement surface, a reference plane is constructed using traditional geometric algorithms, and the flatness deviation between the target measurement surface and the reference plane is calculated, including: Based on the target measurement surface, extract the corresponding surface point cloud data; Based on surface point cloud data, a corresponding reference plane is constructed using a least squares fitting algorithm. Based on the reference plane, the distance deviation of each point in the surface point cloud data is calculated to obtain the corresponding plane deviation data. Based on the plane deviation data, determine the corresponding maximum and minimum deviation values; Calculate the corresponding flatness deviation based on the maximum and minimum deviation values.

[0064] Specifically, based on the reference plane, the distance deviation corresponding to each point in the surface point cloud data is calculated to obtain the corresponding planar deviation data, including: Based on curved point cloud data, obtain the spatial coordinate data corresponding to each point; Based on spatial coordinate data and a reference plane, calculate the normal distance value corresponding to each point; Based on each normal distance value, generate corresponding distance deviation data; Based on the distance deviation data, deviation distribution statistical processing is performed to obtain the corresponding planar deviation data.

[0065] In this embodiment, the corresponding surface point cloud data is first extracted based on the target measurement surface. Specifically, the target measurement surface can be sampled according to a preset grid density to obtain multiple regularly distributed spatial coordinate points. Each point includes coordinates of the motion direction, coordinates of the array width direction, and height coordinates. This surface point cloud data is used to characterize the actual shape of the measured plane after outlier removal and deep learning fitting compensation. Compared with the original discrete measurement points, its data continuity and noise resistance are stronger, and it can serve as effective input data for subsequent reference plane construction.

[0066] Subsequently, based on the surface point cloud data, a corresponding reference plane is constructed using a least-squares fitting algorithm. Specifically, the spatial coordinates of each point in the surface point cloud data are input into a traditional geometric fitting algorithm. By minimizing the sum of squared distances from each point to the fitted plane, the optimal plane parameters are obtained, thereby generating a reference plane for evaluating flatness. This reference plane represents the overall trend reference surface of the measured plane, avoiding the direct use of local convex or concave areas as evaluation benchmarks, and improving the stability and consistency of the flatness calculation results.

[0067] After obtaining the reference plane, spatial coordinate data corresponding to each point is acquired based on the surface point cloud data, and the normal distance value corresponding to each point is calculated in conjunction with the reference plane. Specifically, for each spatial point in the surface point cloud data, the vertical distance from that point to the reference plane is calculated, and the corresponding positive or negative deviation direction is determined according to whether the point is above or below the reference plane, thus obtaining distance deviation data. Subsequently, deviation distribution statistical processing is performed on all distance deviation data to generate corresponding planar deviation data. This planar deviation data can be used to reflect the convexity, concavity, and overall undulation of different areas of the measured plane relative to the reference plane.

[0068] Furthermore, based on the planar deviation data, the corresponding maximum and minimum deviation values ​​are determined. The maximum deviation value represents the deviation corresponding to the highest protrusion position on the target measurement surface relative to the reference plane, and the minimum deviation value represents the deviation corresponding to the lowest concave position on the target measurement surface relative to the reference plane. Finally, based on the difference between the maximum and minimum deviation values, the corresponding flatness deviation is calculated. This flatness deviation can serve as an overall flatness evaluation result for the measured surface, used to determine whether the measured product meets the preset quality requirements, and can provide a basis for subsequent defect location, processing error analysis, and process compensation.

[0069] refer to Figure 4 As shown, Figure 4 This is a schematic diagram of the fitted surface depth map in this application, mainly used to illustrate the distribution of the planar deviation of the target measurement surface relative to the reference plane. Figure 4The horizontal axis represents the spatial position of the measured plane in the direction of movement, and the vertical axis represents the spatial position in the direction of array width. Different colors indicate the magnitude of the distance deviation of the corresponding position relative to the reference plane. Specifically, red areas represent positive deviations, indicating that the corresponding position is above the reference plane, i.e., there is a local bulge; the darker the color, the more bulges there are. Blue areas represent negative deviations, indicating that the corresponding position is below the reference plane, i.e., there is a local depression; the darker the color, the more depressions there are. White or light-colored areas indicate that the corresponding position is close to the reference plane, with smaller flatness deviations. The maximum and minimum deviation points in the figure correspond to the highest bulge area and the lowest depression area in the target measured surface, respectively, and are used for subsequent flatness deviation calculations.

[0070] The depth map shows the overall undulation of the fitted surface, which presents a "wave-like" distribution rather than random and irregular bumps. This indicates that the deviation is most likely due to systematic errors in the processing technology or deformation errors in the material itself, rather than completely random measurement noise. This suggests that the effect of the fitted surface is relatively consistent with the actual situation.

[0071] Furthermore, Figure 4 The fitted surface depth map shown can be used not only to calculate flatness deviation, but also to analyze local processing errors, material deformation trends, and overall surface undulation patterns of the measured plane. By statistically analyzing the difference between the maximum and minimum deviation values, the corresponding flatness result can be obtained. Simultaneously, by observing the spatial distribution of the red and blue areas, it is possible to determine whether the deviation exhibits continuity, regularity, or local concentration, thus providing a basis for subsequent processing optimization, error compensation, and defect location. Compared to traditional single-point measurement methods, this application, through a combination of continuous sampling with a point laser array, deep learning surface fitting, and traditional geometric plane fitting, can more accurately reflect the true spatial morphology of large-size planes.

[0072] S7: Based on the flatness deviation, output the corresponding flatness result and 3D point cloud coordinate result.

[0073] In this embodiment, a flatness result is generated based on the flatness deviation. This flatness result may include a flatness value, maximum deviation value, minimum deviation value, maximum protrusion location, maximum depression location, and a measurement pass / fail judgment result. If the flatness deviation is less than or equal to a preset flatness threshold, the measured surface can be determined to meet the corresponding testing requirements; if the flatness deviation is greater than the preset flatness threshold, the measured surface can be determined to have a flatness anomaly, and the abnormal area can be further located based on the maximum protrusion location and the maximum depression location.

[0074] Simultaneously, the system outputs 3D point cloud coordinate results. These results may include raw point cloud data, calibration point cloud data, and regularized point cloud data generated based on the target measurement surface. The raw point cloud data is used to trace the sensor acquisition status, the calibration point cloud data reflects the actual measurement data after spatial error correction, and the regularized point cloud data represents the continuous morphology of the target measurement surface. By outputting 3D point cloud coordinate results, data support can be provided for subsequent product quality analysis, local defect location, processing error compensation, and 3D visualization.

[0075] Furthermore, flatness distribution maps, surface depth maps, or 3D surface display results are generated based on the deviations of each point relative to the reference plane. In the visualization results, the local convexities, depressions, and overall fluctuation trends of the measured plane can be displayed through the height deviation distribution of different areas, allowing inspectors to intuitively determine whether the changes in the plane's shape exhibit systematic processing error characteristics. Finally, the flatness results, 3D point cloud coordinate results, and visualization results are simultaneously output to the terminal interface or stored in a database for online production line inspection, quality sampling, and subsequent error correction.

[0076] refer to Figure 5 As shown, Figure 5 This is a schematic diagram of the three-dimensional fitting between the target measurement surface and the reference plane in this application. It is mainly used to illustrate the spatial deviation relationship between the target measurement surface obtained by fitting based on the deep learning model and the reference plane constructed by the traditional geometric algorithm. Figure 5 The light-colored semi-transparent surface represents the target measurement surface after fitting, the black grid plane represents the standard reference plane constructed using the least squares fitting algorithm, the green dots represent the normal point cloud data participating in the fitting calculation, and the red dots represent outliers that were removed after detection. Figure 5 The maximum and minimum deviation points correspond to the highest convex position and the lowest concave position of the target measurement surface relative to the reference plane, respectively, and are used for subsequent flatness deviation calculation.

[0077] Furthermore, by performing deviation distribution statistical processing on all distance deviation data, the corresponding maximum and minimum deviation values ​​can be determined, and the corresponding flatness deviation can be calculated based on the difference between the maximum and minimum deviation values. Figure 5 The maximum deviation point shown indicates the location of the most obvious local bulge, and the minimum deviation point indicates the location of the most obvious local depression. The difference between the two is the final output flatness result. This three-dimensional fitting method can intuitively display the overall spatial undulation of the measured plane, and simultaneously observe the distribution of abnormal points, the surface fitting effect, and the flatness change trend, providing a basis for subsequent product quality analysis, processing error location, and process compensation.

[0078] Please see Figure 6This application also provides a high-precision continuous measurement system for the flatness of a point laser array plane, the system comprising: The data acquisition module is used to acquire the height measurement data and motion axis position data corresponding to the point laser array; The collinearity calibration module is used to perform collinearity calibration on the point laser array based on a preset calibration benchmark to obtain the corresponding calibration parameters; The point cloud construction module is used to construct corresponding point cloud data based on height measurement data and motion axis position data; The error correction module is used to perform spatial error correction processing on the point cloud data based on the calibration parameters to obtain the corresponding calibration point cloud data. The first calculation module is used to perform outlier removal and surface fitting processing based on calibration point cloud data using a deep learning model to obtain the corresponding target measurement surface. The second calculation module is used to construct a reference plane based on the target measurement surface using traditional geometric algorithms, and to calculate the flatness deviation between the target measurement surface and the reference plane. The results output module is used to output the corresponding flatness results and 3D point cloud coordinate results based on the flatness deviation.

[0079] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0080] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0081] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0082] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0083] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0084] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0085] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0086] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0087] This application provides a method and system for high-precision continuous measurement of the flatness of a point laser array. This method acquires the height measurement data and motion axis position data corresponding to the point laser array, and combines collinearity calibration and spatial error correction processing to achieve unified coordinate mapping and high-precision spatial compensation of multi-point laser measurement data. This effectively reduces the impact of point laser array installation deviation, motion axis jitter error, and sensor zero-point drift on the measurement results, thereby improving the spatial consistency and measurement stability of the point cloud data. Furthermore, by constructing point cloud data and using a deep learning model to perform outlier removal and surface fitting processing, it is possible to adaptively identify and optimize discrete noise points, defect points, and local distortion data generated during the measurement process, improving the continuous surface reconstruction capability and data robustness in complex surface scenarios. Simultaneously, by combining traditional geometric algorithms to construct a reference plane and calculating the flatness deviation between the target measurement surface and the reference plane, it can balance the nonlinear fitting capability of deep learning algorithms with the high interpretability of traditional geometric algorithms, achieving high-precision quantitative analysis of flatness results. Compared with traditional single-point discrete measurement methods, this invention has the advantages of high continuous measurement efficiency, high measurement accuracy, strong anti-interference ability, and strong three-dimensional visualization ability, and can be widely used in precision machining, industrial inspection, and high-precision planar quality assessment.

[0088] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0089] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0090] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0091] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0092] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for high-precision continuous measurement of the flatness of a point laser array plane, characterized in that, The method includes the following steps: Acquire the height measurement data and motion axis position data corresponding to the point laser array; Based on a preset calibration benchmark, the point laser array is subjected to collinear calibration to obtain the corresponding calibration parameters; Based on the height measurement data and the motion axis position data, the corresponding point cloud data is constructed; Based on the calibration parameters, spatial error correction processing is performed on the point cloud data to obtain the corresponding calibration point cloud data; Based on the calibration point cloud data, anomaly removal and surface fitting are performed using a deep learning model to obtain the corresponding target measurement surface. Based on the target measurement surface, a reference plane is constructed using traditional geometric algorithms, and the flatness deviation between the target measurement surface and the reference plane is calculated. Based on the flatness deviation, the corresponding flatness result and three-dimensional point cloud coordinate result are output.

2. The method according to claim 1, characterized in that, The height measurement data includes the real-time height value, sampling timestamp information, and sensor identification information corresponding to the laser displacement sensor at each point; the motion axis position data includes the position coordinate information, motion speed information, and position sampling time information corresponding to the servo motion axis.

3. The method according to claim 1, characterized in that, The collinear calibration process of the point laser array based on a preset calibration benchmark to obtain the corresponding calibration parameters includes: Control the point laser array to move to the preset calibration position and place the standard plane in the corresponding measurement area; Based on the standard plane, collect the reference height data corresponding to the laser displacement sensor at each point; Based on the reference height data, collinear calibration processing is performed on the laser displacement sensors at each point to obtain the corresponding sensor calibration data. Based on the sensor calibration data, zero-point compensation correction is performed on the laser displacement sensors at each point to obtain the corresponding calibration parameters.

4. The method according to claim 1, characterized in that, The construction of corresponding point cloud data based on the height measurement data and the motion axis position data includes: Based on the height measurement data and the motion axis position data, a corresponding spatial coordinate mapping relationship is established; Based on the spatial coordinate mapping relationship, the height values ​​corresponding to the laser displacement sensors at each point are processed by spatial coordinate transformation to obtain the corresponding discrete point cloud data. Based on the discrete point cloud data, continuous stitching processing is performed according to the direction of motion to obtain the corresponding continuous point cloud data; Based on the continuous point cloud data, coordinate unification processing is performed to obtain the corresponding point cloud data.

5. The method according to claim 1, characterized in that, The step of performing spatial error correction processing on the point cloud data based on the calibration parameters to obtain corresponding calibration point cloud data includes: Based on the calibration parameters, spatial offset correction processing is performed on the point cloud data to obtain the corresponding initial corrected point cloud data; Based on the initial corrected point cloud data, coordinate compensation processing is performed on the installation deviation of the laser displacement sensor at each point to obtain the corresponding coordinate corrected point cloud data. Based on the coordinate-corrected point cloud data, error compensation processing is performed on the trajectory deviation of the motion axis to obtain the corresponding error-compensated point cloud data. Based on the error-compensated point cloud data, coordinate unification processing is performed to obtain the corresponding calibration point cloud data.

6. The method according to claim 1, characterized in that, Based on the calibration point cloud data, a deep learning model is used to perform outlier removal and surface fitting to obtain the corresponding target measurement surface, including: Based on the calibration point cloud data, the corresponding spatial coordinate features and height change features are extracted to obtain the corresponding surface feature data; Based on the surface feature data, outlier detection and removal are performed to obtain the corresponding valid point cloud data. Based on the effective point cloud data, a continuous curve fitting process is performed using a deep learning model to obtain the corresponding curve point cloud data. Based on the curve point cloud data, surface fitting processing is performed to obtain the corresponding target measurement surface.

7. The method according to claim 6, characterized in that, The step of performing continuous curve fitting processing using a deep learning model based on the effective point cloud data to obtain the corresponding curve point cloud data includes: Based on the effective point cloud data, point cloud sequence partitioning is performed according to the direction of motion to obtain the corresponding local point cloud data. Based on the local point cloud data, the corresponding height change features and spatial continuity features are extracted to obtain the corresponding curve feature parameters; Based on the curve feature parameters, a deep learning model is used to perform continuous curve fitting to obtain the corresponding initial curve data. Based on the initial curve data, curve smoothing compensation processing is performed to obtain the corresponding continuous curve data; Based on the continuous curve data, the height information corresponding to any position in the spatial coordinate system is obtained, and the corresponding curve point cloud data is obtained.

8. The method according to claim 1, characterized in that, The step of constructing a reference plane based on the target measurement surface using traditional geometric algorithms and calculating the flatness deviation between the target measurement surface and the reference plane includes: Based on the target measurement surface, extract the corresponding surface point cloud data; Based on the surface point cloud data, a corresponding reference plane is constructed using a least squares fitting algorithm; Based on the reference plane, the distance deviation of each point in the curved surface point cloud data is calculated to obtain the corresponding plane deviation data. Based on the plane deviation data, determine the corresponding maximum and minimum deviation values; Based on the maximum deviation value and the minimum deviation value, the corresponding flatness deviation is calculated.

9. The method according to claim 8, characterized in that, The step of calculating the distance deviation corresponding to each point in the surface point cloud data based on the reference plane to obtain the corresponding plane deviation data includes: Based on the surface point cloud data, obtain the spatial coordinate data corresponding to each point; Based on the spatial coordinate data and the reference plane, calculate the normal distance value corresponding to each point; Based on each of the aforementioned normal distance values, corresponding distance deviation data is generated; Based on the distance deviation data, deviation distribution statistical processing is performed to obtain the corresponding planar deviation data.

10. A high-precision continuous measurement system for the flatness of a point laser array plane, characterized in that, The system includes: The data acquisition module is used to acquire the height measurement data and motion axis position data corresponding to the point laser array; The collinear calibration module is used to perform collinear calibration on the point laser array based on a preset calibration benchmark to obtain the corresponding calibration parameters; The point cloud construction module is used to construct corresponding point cloud data based on the height measurement data and the motion axis position data; An error correction module is used to perform spatial error correction processing on the point cloud data based on the calibration parameters to obtain the corresponding calibration point cloud data. The first calculation module is used to perform outlier removal and surface fitting processing based on the calibration point cloud data using a deep learning model to obtain the corresponding target measurement surface. The second calculation module is used to construct a reference plane based on the target measurement surface using traditional geometric algorithms, and to calculate the flatness deviation between the target measurement surface and the reference plane. The result output module is used to output the corresponding flatness result and three-dimensional point cloud coordinate result based on the flatness deviation.

Images