A high-precision monitoring system for tooth deviation in a gear cutting process

Abstract

The present application relates to the technical field of numerical control machining monitoring, in particular to a high-precision monitoring system for tooth deviation in the process of gear shaving, comprising an online measurement module, a dynamic compensation module, a deviation calculation module and a result output module connected with the signal of the numerical control system of the gear shaving machine, the online measurement module, the dynamic compensation module, the deviation calculation module and the result output module being integrated in a processing element; the processing element is used to obtain the design parameters of the workpiece and build a theoretical workpiece model, the online measurement module is used to collect tooth surface point cloud data in real time during the machining process, the dynamic compensation module is used to correct the vibration error and temperature error of the point cloud data, the deviation calculation module is used to quantify the tooth surface deviation value, and the result output module is used to generate an early warning signal, a dynamic correction instruction and execute system self-calibration. The present application forms a closed loop of machining, monitoring, correction and re-machining with the numerical control system, and finally realizes high-precision real-time monitoring and dynamic correction of tooth deviation.

Description

Technical Field

[0001] This invention relates to the field of CNC machining monitoring technology, specifically to a high-precision monitoring system for tooth deviation during tooth cutting. Background Technology

[0002] Gear refinishing is a finishing process for gear teeth. Through the meshing motion of a refinishing cutter and the gear, minute cuts are made to the tooth surface to improve the gear's precision and surface quality. During gear refinishing, factors such as machine tool vibration, temperature changes, tool wear, and fluctuations in machining parameters can easily lead to tooth deviations, such as pitch deviation, profile deviation, and direction deviation. These tooth deviations directly affect the meshing accuracy, transmission smoothness, and service life of the formed gear, and in severe cases, may cause the entire transmission system to fail. Therefore, high-precision monitoring of tooth deviations during gear refinishing is crucial.

[0003] Existing technologies commonly employ contact-based measurement techniques, such as coordinate measuring machines (CMMs). However, these techniques require offline measurement of the workpiece, which not only affects processing continuity and reduces production efficiency but also can lead to measurement accuracy issues due to probe wear and environmental factors. Furthermore, they cannot monitor deviations during processing in real time. Commonly used non-contact measurement techniques, such as structured light, can acquire full-area data of the tooth surface online, but suffer from insufficient clarity of the light stripe under coolant interference, resulting in measurement blind spots. They also have limited accuracy in point cloud acquisition for critical areas such as the tooth root and lack dynamic environmental error compensation mechanisms, making it difficult to meet the real-time monitoring requirements of high-precision tooth-grinding machining.

[0004] To address the issues of insufficient accuracy and poor real-time performance in existing tooth deviation monitoring technologies for tooth scraping, a tooth deviation monitoring system integrating online measurement, dynamic compensation, high-precision deviation calculation, and intelligent correction is proposed for the tooth deviation monitoring process during tooth scraping. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a high-precision tooth deviation monitoring system during tooth cutting, which improves the accuracy and real-time performance of tooth deviation monitoring.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: a high-precision monitoring system for tooth deviation during tooth removal processing, comprising an online measurement module, a dynamic compensation module, a deviation calculation module, and a result output module connected to the CNC system of the tooth removal machine tool, wherein the online measurement module, the dynamic compensation module, the deviation calculation module, and the result output module are integrated into a processing element;

[0007] The processing element is used to acquire workpiece design parameters and build a theoretical workpiece model. The online measurement module is used to collect tooth surface point cloud data in real time during the processing. The dynamic compensation module is used to correct vibration and temperature errors in the point cloud data. The deviation calculation module is used to quantify tooth surface deviation values. The result output module is used to generate early warning signals, dynamic correction commands and execute system self-calibration.

[0008] Furthermore, the online measurement module includes a composite light source unit and an image acquisition unit, which are linked by synchronous control logic. The composite light source unit uses blue-green band composite line structured light to project a cross light plane onto the tooth surface, and penetrates the coolant fog barrier through a dual-wavelength intelligent switching algorithm. The image acquisition unit is used to synchronously acquire light stripe images and process them to generate standardized PCD format point cloud data.

[0009] Furthermore, the composite light source unit includes a dual-path projection assembly and a synchronous control subunit. The dual-path projection assembly consists of two sets of line lasers, which are mounted on an adjustable cantilever bracket and equipped with a stepper motor drive assembly with a harmonic reducer. The angle adjustment accuracy is ±0.01°, and the angle between the laser optical axis and the machine tool spindle axis projected onto the XY plane is 30°~45°, with an effective measurement area of ​​tooth surface overlap ≥60%. The synchronous control subunit receives signals from the spindle encoder and uses the formula:

[0010]

[0011] in, The trigger time of the line laser. When the main spindle encoder signal is received, This is the exposure delay compensation value;

[0012] To achieve synchronous exposure of the laser and the image acquisition unit.

[0013] Furthermore, the image acquisition unit includes a CMOS camera, a polarization filter assembly, and an adaptive exposure unit; the polarization filter assembly suppresses tooth surface reflection by using a motor-driven rotatable polarizer, and the adaptive exposure unit is based on the formula:

[0014]

[0015] Where G represents the CMOS camera gain. Here, k represents the initial gain of the CMOS camera, and k is the adaptive coefficient. These are the statistical values ​​of the grayscale histogram of the light stripe image;

[0016] The camera gain and exposure time are dynamically adjusted; the image acquisition unit processes the image through multi-frame adaptive median filtering and gray-scale centroid subpixel extraction, and generates point cloud data with ≥40,000 points per tooth by combining the principle of triangulation.

[0017] Furthermore, the dynamic compensation module includes a vibration compensation unit and a temperature drift unit, both of which are connected to environmental sensors. The environmental sensors include a triaxial accelerometer array mounted on the machine tool spindle and temperature sensors mounted on the front end of the spindle, the workpiece clamping seat, and the cutting zone. The dynamic compensation module receives point cloud data from the online measurement module, corrects for vibration and temperature errors, and outputs it to the deviation calculation module.

[0018] Furthermore, the vibration compensation unit includes a triaxial MEMS accelerometer array and a phase synchronization subunit; the triaxial MEMS accelerometer array has a sampling frequency of 100Hz, and the vibration data is processed by Kalman filtering based on the formula:

[0019]

[0020] in, This represents the point cloud displacement vector caused by vibration during the tooth-grinding process. This is the time-domain vibration acceleration vector;

[0021] The displacement vector is calculated, and the point cloud is rigidly transformed and corrected; the phase synchronization subunit realizes the time domain alignment of vibration data with the spindle rotation angle through phase-locked loop technology.

[0022] Furthermore, the temperature drift unit includes a high-precision temperature sensor and a finite element thermal analysis auxiliary model; the temperature sensor has a measurement accuracy of ±0.1℃ and a sampling frequency of 50Hz, based on the formula:

[0023]

[0024] in, The change in workpiece length caused by temperature. Indicates the original length of the workpiece feature. This is the coefficient of linear expansion of the material. Temperature changes in key areas;

[0025] The thermal expansion is calculated, and the thermal expansion is converted into the normal deformation deviation of the tooth surface using an auxiliary model. The point cloud after vibration compensation is then corrected.

[0026] Furthermore, the deviation calculation module includes a point cloud registration unit and a deviation mapping unit. The point cloud registration unit aligns the point cloud with the theoretical workpiece model through a two-step method of coarse registration and fine registration. The coarse registration is based on fitting the end face plane using a random sampling consensus algorithm and determining the workpiece axis through cylindrical fitting. The fine registration adopts an improved ICP algorithm that introduces tooth surface curvature weights, with a registration error ≤0.5μm.

[0027] Furthermore, the deviation mapping unit uses NURBS surface reconstruction technology to construct the actual tooth surface, through the formula:

[0028]

[0029] in, These are the actual point coordinates. Let n be the theoretical point coordinates, and n be the theoretical tooth surface normal vector.

[0030] Calculate the normal deviation at each point on the tooth surface, automatically classify and statistically analyze the tooth pitch deviation, tooth profile deviation, and tooth direction deviation, and generate a deviation heatmap and standardized deviation data file.

[0031] Furthermore, the result output module includes an edge contact early warning unit, a dynamic correction unit, and a reference calibration component. The edge contact early warning unit triggers an early warning based on threshold comparison and generates a local shaping strategy. The dynamic correction unit converts the shaping strategy into G-code correction instructions. The reference calibration component uses an involute template with an accuracy ≤1μm, and is defined by the formula:

[0032]

[0033] in, The system parameters to be calibrated include CMOS camera intrinsic parameters and structured light plane equations. To calibrate the coordinates of the measured points on the template, To calibrate the coordinates of the theoretical points on the template;

[0034] Perform system self-calibration regularly.

[0035] The principle behind the above scheme is as follows:

[0036] Centered on online data acquisition, dynamic correction, deviation quantification, and closed-loop calibration, this system achieves high-precision monitoring of tooth deviation through the collaborative efforts of various modules. First, the processing element constructs a theoretical model based on the workpiece design parameters. The online measurement module uses blue-green composite structured light to project a cross-beam surface, resisting coolant interference, and simultaneously acquires light stripe images, which are then denoised and sub-pixel extracted to generate point clouds. The dynamic compensation module uses accelerometers and temperature sensors to correct vibration displacement and thermal deformation errors, respectively. The deviation calculation module quantifies tooth surface deviation by coarsely and finely registering the point cloud with the theoretical model. The result output module generates real-time warnings and correction commands, and also periodically self-calibrates using a high-precision template, forming a closed loop with the CNC system to efficiently complete the monitoring and adjustment of tooth deviation during tooth cutting.

[0037] Compared to existing technologies, the above solution has the following advantages:

[0038] 1. This solution uses an online measurement module to synchronously collect tooth surface data throughout the entire machining process, without interrupting machining or removing the workpiece. This ensures machining continuity, avoids repeated clamping errors, and improves production efficiency. Through a dual-wavelength intelligent switching algorithm and dynamic laser power adjustment, it effectively penetrates the coolant mist barrier; the cross-plane design and multi-camera collaborative acquisition ensure coverage of all areas of the tooth tip and root, eliminating measurement blind spots; combined with sub-pixel-level light stripe extraction and dynamic compensation, it improves point cloud data accuracy and reduces tooth deviation measurement errors.

[0039] 2. This solution constructs a coupled error correction model using a vibration compensation unit and a temperature drift unit to correct vibration displacement and thermal deformation deviations in real time. This effectively counteracts environmental interference with measurements, ensuring that deviation calculations are based on real tooth surface data and further improving the reliability of monitoring results. The result output module generates early warning signals and dynamic correction commands in real time, resulting in extremely low latency from deviation data acquisition to CNC system adjustment, allowing for timely correction of machining parameters. Simultaneously, based on the LSTM algorithm, deviation trends are predicted to proactively avoid deviation exceeding limits and reduce the defect rate.

[0040] 3. This solution automatically performs self-calibration periodically through the benchmark calibration component and optimizes system parameters using a high-precision involute template to ensure long-term stable measurement accuracy without frequent manual intervention, thereby reducing operation and maintenance costs and ensuring consistency of measurement results for different batches of workpieces. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the overall system framework according to an embodiment of the present invention;

[0042] Figure 2 This is a schematic diagram of the system processing flow according to an embodiment of the present invention;

[0043] Figure 3 This is a schematic diagram of the system module framework of an embodiment of the present invention. Detailed Implementation

[0044] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0045] Further details are provided through specific implementation methods:

[0046] Traditional tooth-grinding processes have several shortcomings in monitoring tooth deviation. Common contact measurement methods, such as using coordinate measuring machines (CMMs), can achieve a certain level of accuracy, but the process is cumbersome. It requires removing the workpiece from the machine tool, measuring under specific conditions, and then re-clamping the workpiece for subsequent tooth-grinding. This not only affects the continuity of tooth-grinding and reduces production efficiency, but the repeated clamping of the workpiece can also lead to tooth profile errors. Secondly, frequent contact between the probe and the tooth surface can easily cause probe wear, affecting measurement accuracy, and the probe may also cause minor wear on the tooth surface. Thirdly, this type of measurement method is highly sensitive to environmental conditions; even slight changes in environmental factors such as temperature, humidity, and vibration can lead to deviations in the measurement results.

[0047] While existing non-contact measurement technologies such as visual measurement and laser scanning can acquire full-area data of the tooth surface online, their accuracy still falls short of the requirements for high-precision tooth resurfacing. For example, their sensitivity in detecting minute tooth deviations is insufficient, especially when processing complex tooth profiles. Limitations in measurement algorithms make it difficult to accurately distinguish and quantify minute tooth deviations, failing to provide precise basis for timely adjustments to processing parameters. Furthermore, existing non-contact measurement technologies often have poor real-time performance, unable to synchronously output tooth deviation monitoring results during processing. This makes it difficult for operators to promptly detect anomalies during tooth resurfacing, leading to defective products and increased production costs.

[0048] For existing automated production lines using gear-cutting machine tools, this embodiment utilizes a monitoring system to perform high-precision monitoring of tooth deviation during the gear-cutting process. The high-precision tooth deviation monitoring system during gear-cutting (hereinafter referred to as the "monitoring system") has the following framework: Figure 1 As shown, it includes an online measurement module, a dynamic compensation module, a deviation calculation module, and a result output module that are connected to the CNC system of the gear cutting machine. The online measurement module, dynamic compensation module, deviation calculation module, and result output module are integrated into the processing element. In this embodiment, the processing element is an industrial PC, which is integrated into the machine tool worktable and integrates a Windows system or a Linux system.

[0049] Before the workpiece gear-cutting process begins, during the preparation stage (workpiece clamping and setting or acquiring machining parameters), the workpiece blank is mounted on the gear-cutting machine tool. Relevant design parameters of the workpiece, including module, number of teeth, and tooth profile angle, are acquired via an industrial PC. In this embodiment, the relevant design parameters are directly acquired from the CNC system via the industrial PC. In some embodiments, the relevant design parameters are input via the industrial PC, and various modules within the industrial PC acquire these parameters. The industrial PC then constructs a theoretical workpiece model based on these design parameters. In some embodiments, the industrial PC can also be a microprocessor, a general-purpose processor, or other processing element.

[0050] The following is a detailed description of each module:

[0051] The online measurement module serves as the data acquisition front-end of the tooth deviation monitoring system during tooth machining. It connects the CNC system with the subsequent dynamic compensation module, acquiring real-time 3D tooth surface data during the machining process using non-contact structured light measurement technology. Its core functions include anti-interference light projection, synchronous image acquisition, high-precision light stripe processing, and tooth surface point cloud generation, as detailed below:

[0052] The online measurement module consists of a composite light source unit and an image acquisition unit. These two units work together through synchronized control logic to complete the entire process from light signal projection to point cloud output. The composite light source unit projects a highly penetrating cross-beam light onto the tooth surface, overcoming coolant fog interference and ensuring complete coverage of the light stripes on the tooth surface (including the tooth tip and root). The image acquisition unit synchronously acquires images of the light stripes modulated by the tooth profile. Through noise reduction and sub-pixel-level light stripe extraction, it transforms the two-dimensional image information into three-dimensional point cloud data, providing raw data support for subsequent dynamic compensation.

[0053] Specifically, the composite light source unit is equipped with a dual-path projection assembly. This assembly (composed of two sets of line lasers) projects a blue-green band composite line structured light onto the tooth surface, creating an intersecting light plane. The line lasers are mounted on a machine tool cantilever bracket, which is fixed by a Z-axis guide rail slider and is adjustable vertically. Each laser set is equipped with a stepper motor drive assembly with a harmonic reducer, providing an angle adjustment accuracy of ±0.01° to ensure that the two laser beams form an effective measurement area of ​​≥60% in the overlapping region of the tooth surface. The projection angle between the laser optical axis and the machine tool spindle axis in the XY plane is set to 30°~45°, covering both the tooth tip (protruding area of ​​the workpiece) and the tooth root (recessed area of ​​the workpiece), avoiding measurement blind spots. An integrated lens group at the output end optimizes the uniformity of the light stripe. To address the issue of blurred light stripes caused by coolant fog during processing, the output power and phase of the dual-wavelength lasers are controlled via fiber optic coupling, employing an intelligent wavelength switching algorithm.

[0054] Lasers in the 550-580nm wavelength range are preferred due to their high penetration capability into liquid mist; combined with a laser intensity attenuation model:

[0055]

[0056] Where I represents the intensity of the laser light after it penetrates the coolant. This indicates the initial intensity of the laser beam. The absorption coefficient of the coolant is represented by d, and the laser penetration distance is represented by d.

[0057] The laser power is dynamically adjusted to ensure that the light stripe remains clearly visible even in a liquid mist environment.

[0058] The composite light source unit is also equipped with a synchronization control subunit. This subunit receives encoder signals from the gear-cutting machine spindle and uses a programmable logic controller (PLC) to synchronize the exposure of the line laser and the image acquisition unit, preventing image misalignment caused by spindle rotation. The formula for calculating the synchronization trigger time is:

[0059]

[0060] in, The trigger time of the line laser. When the main spindle encoder signal is received, This is the exposure delay compensation value (dynamically adjusted according to the spindle speed to adapt to different machining conditions).

[0061] Specifically, the image acquisition unit includes a CMOS camera and a polarization filter assembly. The polarization filter assembly consists of a rotatable polarizer and a drive motor. The polarizer is automatically rotated and adjusted by the drive motor, and combined with image grayscale statistical analysis, the optimal polarization angle is matched in real time to suppress metal reflection from the tooth surface and prevent light stripe breakage caused by reflection. The CMOS camera's acquisition rate is set to 100fps to ensure synchronization with the spindle rotation and laser triggering, and each frame can cover the entire tooth surface. The image acquisition unit is also equipped with an adaptive exposure unit. This unit dynamically adjusts the CMOS camera gain parameters based on the tooth surface reflection intensity to avoid overexposure or underexposure of the light stripe. The gain calculation formula is:

[0062]

[0063] Where G represents the CMOS camera gain. Here, k represents the initial gain of the CMOS camera, and k is the adaptive coefficient. The grayscale histogram of the light stripe image is used as the statistical value. By monitoring the grayscale distribution in real time, the exposure time (range 10~100μs) and gain are adjusted synchronously to ensure that the edges of the light stripe are clear.

[0064] The image acquisition unit performs two-step processing on the acquired light stripe image to improve the accuracy of the light stripe center positioning:

[0065] (1) Multi-frame adaptive median filtering with a window size of 5×5 removes coolant reflection noise, filters out isolated noise points, and preserves the continuous outline of the light stripe;

[0066] (2) Subpixel-level light stripe center extraction: The coordinates of the light stripe center are calculated based on the gray-scale centroid method. The formula is as follows:

[0067]

[0068] in, Represented as the x-coordinate of the center of the light stripe. This is represented by the grayscale value at position x. Combined with a deep learning sub-pixel localization algorithm, the center localization accuracy of the light stripe is improved to the 0.1 pixel level. Here, x is the pixel coordinate along the horizontal direction within the light stripe, used to identify the position of a point on the light stripe in the horizontal dimension. x iterates through all horizontal pixel coordinates within the light stripe region, combining the corresponding grayscale value... Finally, the horizontal coordinates of the center of the light stripe were calculated. This achieves sub-pixel-level center positioning of the light stripe.

[0069] The image acquisition unit processes the acquired light stripe image using the two steps described above, and then, in conjunction with the CMOS camera system calibration parameters, generates a 3D point cloud of the tooth surface using the principle of triangulation. The specific steps are as follows:

[0070] First, the intrinsic parameters (focal length, distortion coefficient) of the CMOS camera and the equation of the structured light plane are obtained using a checkerboard calibration board, serving as the benchmark for point cloud coordinate calculation. Then, based on the light stripe phase information and calibration parameters, the three-dimensional coordinates of each point on the tooth surface are calculated using a triangulation algorithm, with the following formula:

[0071]

[0072] In the formula, In the image of the tooth surface of the workpiece Phase value at coordinates, The intrinsic parameters of the CMOS camera are calibrated. The generated point cloud data has approximately 40,000 points per tooth, and the density meets the requirements for detecting minor deviations on the tooth surface. In some scenarios, multiple CMOS cameras can be used for collaborative acquisition (field of view overlap ≥60%) to further cover the tooth tip and root edge areas and completely eliminate measurement blind spots.

[0073] The online measurement module ultimately outputs standardized PCD format tooth surface point cloud data, which is transmitted in real time to the dynamic compensation module via industrial Ethernet. This provides high-precision point cloud data for subsequent vibration-temperature coupling error correction and simultaneously feeds back the acquisition status to the CNC system, such as "acquisition normal" and "blurred light stripe alarm," ensuring that machining and measurement are synchronized in a closed loop.

[0074] The dynamic compensation module serves as the central hub for point cloud error correction. Its core function is to eliminate the interference of vibration and temperature in the machining environment on the original point cloud; it corrects the error-contaminated point cloud data into high-precision actual tooth surface point cloud data, providing a true data foundation for subsequent deviation calculations. Specifically:

[0075] The dynamic compensation module consists of a vibration compensation unit and a temperature drift unit. The vibration compensation unit corrects point cloud displacement deviations caused by machine tool spindle and table vibrations by calculating displacement using acceleration data, thus achieving rigid transformation correction. The temperature drift unit corrects workpiece thermal deformation caused by machining heat (cutting heat, ambient temperature changes) by using temperature data to correct tooth surface normal deviations. Vibration and temperature data are acquired through environmental sensors, including a triaxial accelerometer array mounted on the machine tool spindle and a temperature sensor mounted on the workpiece clamping / cutting zone. The dynamic compensation module synchronously receives PCD format point cloud data output from the online measurement module.

[0076] Specifically, the vibration compensation unit includes a triaxial MEMS accelerometer array and a phase synchronization subunit. The triaxial MEMS accelerometer array has an accuracy of ±0.01g and a sampling frequency of 100Hz. It is fixed to the end face of the machine tool spindle tail via a fixture to capture key spindle vibrations, and data is transmitted via a low-noise coaxial cable. The phase synchronization subunit receives the machine tool spindle rotation angle signal and uses phase-locked loop technology to align the vibration data with the spindle rotation angle in the time domain, avoiding correction deviations caused by phase misalignment. The vibration error correction process is as follows:

[0077] First, the raw vibration acceleration data is processed using the Kalman filter algorithm to remove high-frequency interference (such as cutting impact noise) and obtain a clean time-domain vibration acceleration vector. Then, the acceleration is converted into a point cloud displacement vector through a second integral, as shown in the formula:

[0078]

[0079] in, This represents the point cloud displacement vector caused by vibration during the tooth-grinding process. Then, the displacement vector is applied to the original point cloud data, and the three-dimensional coordinates of each point are corrected through matrix transformation to eliminate the point cloud offset caused by vibration, such as the tooth surface position deviation caused by the radial runout of the spindle.

[0080] Specifically, the temperature drift unit includes a high-precision temperature sensor and an auxiliary model. The high-precision temperature sensor has a measurement accuracy of ±0.1℃ and a sampling frequency of 50Hz. In this embodiment, three high-precision temperature sensors are installed at the front end of the spindle, the workpiece clamping base, and next to the cutting zone, respectively, to collect real-time temperature changes in key areas. The auxiliary model is based on the finite element thermal analysis model of the workpiece material, with a preset material linear expansion coefficient. For example, the coefficient of linear expansion of 45# steel is The process for correcting thermal deformation errors is as follows:

[0081] First, calculate the thermal expansion. Based on the thermal expansion model, calculate the change in the workpiece's feature length using the following formula:

[0082]

[0083] in, The change in workpiece length caused by temperature. This represents the original length of the workpiece feature. Then, it is combined with the finite element thermal analysis model to... The deviation is converted into the normal deformation of each point on the tooth surface, and the point cloud after vibration compensation is corrected a second time. Then, the thermal correction weight is dynamically adjusted according to the machining parameters (such as spindle speed and cutting depth).

[0084] The dynamic compensation module ultimately outputs standardized PCD point cloud data after vibration error correction and thermal deformation error correction, which is transmitted to the deviation calculation module in real time via industrial Ethernet. At the same time, it feeds back the error correction status to the CNC system, such as "vibration correction completed" and "temperature exceeds threshold alarm", to ensure that the deviation calculation is based on real tooth surface data without environmental interference.

[0085] The deviation calculation module is the core of the monitoring system's data processing. Its core function is to align the corrected point cloud data with the theoretical workpiece model, and quantify the deviation values ​​at various positions on the tooth surface, such as tooth pitch deviation, tooth profile deviation, and tooth direction deviation, through surface reconstruction and normal comparison. This provides the result output module with deviation data that can be directly used for adjustment.

[0086] The deviation calculation module consists of a point cloud registration unit and a deviation mapping unit. The point cloud registration unit is used to ensure alignment accuracy through a two-step method of coarse registration and fine registration, solving the spatial alignment problem between the corrected point cloud data and the theoretical model. The deviation mapping unit is used to calculate the normal deviation of each point on the tooth surface based on the aligned point cloud data and the theoretical model through surface reconstruction, generating a deviation heatmap and a deviation parameter table. The corrected point cloud data comes from the output of the state compensation module, and the theoretical model comes from the pre-stored data in the industrial PC (constructed based on workpiece design parameters such as module and number of teeth).

[0087] Specifically, the point cloud registration unit is used for coarse registration based on the random sampling consensus algorithm and principal component analysis, using the workpiece end face and axis as feature references. First, the end face point set is extracted from the corrected point cloud data. Then, the end face plane is fitted to the theoretical end face using the random sampling consensus algorithm. The fitted end face plane equation is as follows:

[0088]

[0089] Where a, b, c, and d are plane equation coefficients, and x, y, and z are end face point cloud coordinates. Then, the root circle / tip circle point set is extracted from the point cloud, and the actual workpiece axis is determined using a cylindrical fitting algorithm, aligning it with the theoretical axis. The initial deviation between the point cloud and the theoretical model is reduced to within 10 μm, laying the foundation for fine registration. The point cloud registration unit is used for fine registration based on the improved ICP algorithm. First, if the point cloud density in the root region is below a threshold, radial basis function interpolation is used to refine the density, ensuring registration accuracy in key areas; the weight of the corresponding points is calculated based on the tooth surface curvature, as shown in the following formula:

[0090]

[0091] in, For weight adjustment parameters, This is the reference curvature value of the tooth surface. The curvature is the actual point curvature. Regions with greater curvature (such as the tooth root transition area) have higher weights and are aligned first; then, the objective function is to minimize the weighted distance, as follows:

[0092]

[0093] in, These are weighting coefficients based on tooth surface curvature. Represented as the measured and corrected point cloud coordinate vector. This represents the coordinate vector of the point corresponding to the theoretical workpiece model. Then, the optimal rotation matrix and translation vector are calculated iteratively to ensure the final registration error is ≤0.5μm; the iterative calculation formula is as follows:

[0094]

[0095] in, Let R be the optimal rotation matrix and translation vector of the tooth surface, respectively. R is the rotation matrix and t is the translation vector.

[0096] Specifically, the deviation mapping unit is used to calculate the deviation values ​​at various positions on the tooth surface using NURBS surface reconstruction. First, based on the aligned point cloud data, the actual tooth surface is constructed using NURBS surface formulas:

[0097]

[0098] Where u and v are both surface parameters. All are spline basis functions. As a weighting factor, To control vertex coordinates, a pre-stored theoretical tooth surface NURBS surface (built based on design parameters) is used, ensuring consistency in parametric references between the two. Then, for each point on the actual tooth surface, the distance between the actual point and the theoretical point is calculated along the normal direction of the theoretical tooth surface, using the following formula:

[0099]

[0100] in, These are the actual point coordinates. Here are the theoretical point coordinates, and n is the theoretical tooth surface normal vector. Then, based on the calculation results, the system automatically classifies and statistically analyzes tooth pitch deviation (the difference between deviations of adjacent tooth surfaces), tooth profile deviation (the maximum deviation in the tooth profile direction), and tooth direction deviation (the change in deviation in the tooth width direction), generating a single tooth deviation parameter table.

[0101] The deviation calculation module ultimately outputs a deviation heatmap and standardized deviation data. The deviation heatmap visually displays the deviation distribution in each region of the tooth surface, while the standardized deviation data includes the normal deviation value and classification deviation parameters for each point. The output deviation heatmap and standardized deviation data are transmitted to the result output module via industrial Ethernet, and the registration status is simultaneously fed back to the dynamic compensation module, such as "registration successful" or "insufficient point cloud density alarm".

[0102] The output module is the closed-loop control terminal of the monitoring system. Its core function is to generate early warning signals and dynamic correction instructions based on the deviation calculation results, simultaneously realize system self-calibration, convert deviation data into adjustment parameters that can be executed by the machine tool, and finally realize the machining closed loop of measurement-calculation-adjustment.

[0103] The output module consists of an edge contact early warning unit, a dynamic correction unit, and a reference calibration component. The edge contact early warning unit monitors out-of-limit points in the deviation data in real time, triggers alarms, and generates a local correction strategy. The dynamic correction unit converts the correction strategy into machine tool-executable cutting tool feed correction parameters and outputs them to the CNC system. The reference calibration component performs periodic system self-calibration to ensure long-term measurement accuracy. Deviation data originates from the deviation calculation module, and machine tool operating parameters originate from the CNC system.

[0104] Specifically, the edge contact early warning unit is used to preset a deviation threshold based on workpiece tolerance requirements. For each point in the deviation data, a threshold comparison is performed, using the following formula:

[0105]

[0106] in, As a warning marker, when When the warning is triggered, an alert is activated, and the location of the out-of-limit point is marked on the industrial PC interface. The edge contact warning unit is also used in threshold-based and region-growing-based out-of-limit zone identification algorithms and expert system-based local shaping strategies. First, adjacent out-of-limit points are divided into single shaping regions using a region-growing algorithm. Then, the shaping amount is calculated based on the maximum deviation value of that region, using the following formula:

[0107]

[0108] in, For correction factor, The maximum deviation value within the region is used; then, a local shaping strategy is generated, which includes the coordinates of the shaping region and the shaping amount.

[0109] Specifically, the dynamic correction unit converts the radial profile adjustment amount into a radial feed correction value for the cutting tool. It adjusts the smoothing coefficient of the correction parameters based on the machine tool spindle speed and cutting force (to avoid vibration caused by sudden feed changes), converts the correction parameters into G-code instructions recognizable by the CNC system, and compares the corrected feed amount with the machine tool's safe range. If it exceeds the range, it pauses output and issues an alarm. The G-code instructions are transmitted to the CNC system via industrial Ethernet. The unit also receives the instruction execution status from the CNC system; if no feedback is received within 100ms, the instruction is resent to ensure the adjustment takes effect. Furthermore, it uses an LSTM algorithm to predict the deviation trend of the next tooth based on historical deviation data and outputs pre-correction instructions in advance.

[0110] Specifically, the benchmark calibration components include a high-precision involute template and a dedicated fixture, which are mounted on the machine tool table. In this embodiment, the system self-calibrates every 24 hours; in some embodiments, it can also be triggered when the system issues an accuracy degradation alarm. The self-calibration process is as follows:

[0111] First, the fixture automatically positions the involute template to the measurement reference position, and the industrial PC identifies the template outline. Then, the online measurement module and dynamic compensation module perform a complete measurement of the template, generating actual point cloud data of the template. Next, the actual point cloud data of the template is compared with the theoretical model of the template, and the system parameters are optimized using the least squares method, with the following formula:

[0112]

[0113] in, The system parameters to be calibrated (in this embodiment, these include CMOS camera intrinsic parameters and structured light plane equations). To calibrate the coordinates of the measured points on the template, To calibrate the theoretical point coordinates of the template, the system parameters were calibrated through measurement and analysis of the template, and the template was remeasured to verify the accuracy of the monitoring system.

[0114] The output module ultimately outputs a warning signal, a dynamic correction command (G-code command), and a calibration report. The warning signal alerts the operator to address any out-of-limit issues; the dynamic correction command drives the CNC system to adjust the cutter feed parameters; and the calibration report records the self-calibration results.

[0115] Throughout the machining process, the monitoring system also acquires real-time operating parameters of the gear cutting machine from the CNC system, including spindle speed, feed rate, and cutting force. By analyzing these operating parameters and combining them with tooth deviation data, the system can further determine the cause of tooth deviation. For example, if a sudden increase in cutting force is detected, along with abnormal tooth profile deviation, it may indicate severe tool wear or interference during the cutting process. The system will promptly issue an alarm to allow operators to check and adjust the tool in a timely manner.

[0116] After the workpiece is machined and remains stationary, the monitoring system performs a comprehensive measurement of the entire workpiece again, generating a complete tooth deviation report. The report details the measurement results of parameters such as tooth pitch deviation, tooth profile deviation, and tooth direction deviation for each tooth surface, and compares them with the design values ​​in the theoretical workpiece model. Operators can use the data in the report to assess whether the workpiece's machining quality meets the requirements. If individual tooth deviations still exceed the tolerance range, operators can use the data provided by the system to perform targeted corrective machining on the workpiece, or adjust the machining process parameters for subsequent batches of workpieces to improve machining quality.

[0117] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A high-precision monitoring system for tooth deviation during tooth cutting, characterized in that, It includes an online measurement module, a dynamic compensation module, a deviation calculation module, and a result output module that are connected to the CNC system of the gear cutting machine tool. The online measurement module, dynamic compensation module, deviation calculation module, and result output module are integrated into the processing element. The processing element is used to acquire workpiece design parameters and construct a theoretical workpiece model. The online measurement module is used to collect tooth surface point cloud data in real time during the processing. The dynamic compensation module is used to correct the vibration error and temperature error of the point cloud data. The deviation calculation module is used to quantify the tooth surface deviation value. The result output module is used to generate early warning signals, dynamic correction instructions and execute system self-calibration. The online measurement module includes a composite light source unit and an image acquisition unit, which are linked by synchronous control logic. The composite light source unit uses blue-green band composite line structured light to project a cross light plane onto the tooth surface and penetrates the coolant fog barrier through a dual-wavelength intelligent switching algorithm. The image acquisition unit is used to synchronously acquire light stripe images and process them to generate standardized PCD format point cloud data. The dynamic compensation module includes a vibration compensation unit and a temperature drift unit, both of which are connected to environmental sensors. The environmental sensors include a triaxial accelerometer array mounted on the machine tool spindle and temperature sensors mounted on the front end of the spindle, the workpiece clamping seat, and the cutting zone. The dynamic compensation module receives point cloud data from the online measurement module, corrects for vibration and temperature errors, and outputs it to the deviation calculation module. The deviation calculation module includes a point cloud registration unit and a deviation mapping unit. The point cloud registration unit aligns the point cloud with the theoretical workpiece model through a two-step method of coarse registration and fine registration. The coarse registration is based on fitting the end face plane with a random sampling consensus algorithm and determining the workpiece axis through cylindrical fitting. The fine registration adopts an improved ICP algorithm that introduces tooth surface curvature weights, and the registration error is ≤0.5μm. The result output module includes an edge contact early warning unit, a dynamic correction unit, and a reference calibration component. The edge contact early warning unit triggers an early warning based on threshold comparison and generates a local shaping strategy. The dynamic correction unit converts the shaping strategy into G-code correction instructions. The reference calibration component uses an involute template with an accuracy ≤1μm, and uses the formula: ； in, The system parameters to be calibrated include CMOS camera intrinsic parameters and structured light plane equations. To calibrate the coordinates of the measured points on the template, To calibrate the coordinates of the theoretical points on the template; Perform system self-calibration regularly.

2. The high-precision tooth deviation monitoring system during tooth cutting according to claim 1, characterized in that, The composite light source unit includes a dual-path projection assembly and a synchronous control subunit. The dual-path projection assembly consists of two sets of line lasers, each mounted on an adjustable cantilever bracket and equipped with a stepper motor drive assembly with a harmonic reducer. The angle adjustment accuracy is ±0.01°, and the angle between the laser optical axis and the machine tool spindle axis projected onto the XY plane is 30°~45°. The effective measurement area of ​​tooth surface overlap is ≥60%. The synchronous control subunit receives signals from the spindle encoder and uses the formula: ； in, The trigger time of the line laser. When the main spindle encoder signal is received, This is the exposure delay compensation value; To achieve synchronous exposure of the laser and the image acquisition unit.

3. The high-precision monitoring system for tooth deviation during tooth cutting according to claim 1, characterized in that, The image acquisition unit includes a CMOS camera, a polarization filter assembly, and an adaptive exposure unit; the polarization filter assembly suppresses tooth surface reflection by using a motor-driven rotatable polarizer; the adaptive exposure unit is based on the formula: ； Where G represents the CMOS camera gain. Here, k represents the initial gain of the CMOS camera, and k is the adaptive coefficient. These are the statistical values ​​of the grayscale histogram of the light stripe image; The camera gain and exposure time are dynamically adjusted; the image acquisition unit processes the image through multi-frame adaptive median filtering and gray-scale centroid subpixel extraction, and generates point cloud data with ≥40,000 points per tooth by combining the triangulation principle.

4. The high-precision tooth deviation monitoring system during tooth cutting according to claim 1, characterized in that, The vibration compensation unit includes a triaxial MEMS accelerometer array and a phase synchronization subunit; the triaxial MEMS accelerometer array has a sampling frequency of 100Hz, and the vibration data is processed by Kalman filtering based on the formula: ； in, This represents the point cloud displacement vector caused by vibration during the tooth-grinding process. This is the time-domain vibration acceleration vector; The displacement vector is calculated, and the point cloud is rigidly transformed and corrected; the phase synchronization subunit realizes the time domain alignment of vibration data with the spindle rotation angle through phase-locked loop technology.

5. The high-precision monitoring system for tooth deviation during tooth cutting according to claim 1, characterized in that, The temperature drift unit includes a high-precision temperature sensor and a finite element thermal analysis auxiliary model; the temperature sensor has a measurement accuracy of ±0.1℃ and a sampling frequency of 50Hz, based on the formula: ； in, The change in workpiece length caused by temperature. Indicates the original length of the workpiece feature. is the coefficient of linear expansion of the material. Temperature changes in key areas; The thermal expansion is calculated, and the thermal expansion is converted into the normal deformation deviation of the tooth surface using an auxiliary model. The point cloud after vibration compensation is then corrected.

6. The high-precision monitoring system for tooth deviation during tooth cutting according to claim 1, characterized in that, The deviation mapping unit uses NURBS surface reconstruction technology to construct the actual tooth surface, using the formula: ； in, These are the actual point coordinates. Let n be the theoretical point coordinates, and n be the theoretical tooth surface normal vector. Calculate the normal deviation at each point on the tooth surface, automatically classify and statistically analyze the tooth pitch deviation, tooth profile deviation, and tooth direction deviation, and generate a deviation heatmap and standardized deviation data file.

Images