Grinding method for circular arc end face teeth with respect to an annular master plate

Abstract

Realize the digital high-precision machining of arc end face teeth for a disk-shaped master plate. 【Solution means】Divide the gear teeth that need to be machined on the disk-shaped master plate into a plurality of groups. Each group contains K pairs of gear teeth at equal intervals. Complete the grinding of the gear teeth in each group in sequence, and rotate the disk-shaped master plate material by a small angle α every time the machining of the last pair of gear teeth in each group is completed. Here, before grinding, introduce a neural network to optimize the large and small angles for rotating the disk-shaped master plate material. During the grinding process, if the position of the disk-shaped master plate material is displaced, adjust the position of the disk-shaped master plate material. In the present invention, by using the uniform distribution method of errors, the cumulative value of the angular error is reduced. The wear of the grinding wheel is accurately modeled, measured, the grinding wheel is dressed / replaced by "thermal-mechanical coupling", and local quantitative re-grinding of the disk-shaped master plate is performed by comparing the surface defects of the arc end face teeth of the disk-shaped master plate with the discrete normal deviation of the surface of the theoretical arc end face teeth, thereby realizing the digital high-precision machining of the arc end face teeth of the disk-shaped master plate.

Description

Technical Field

[0001] The present invention belongs to the technical field of manufacturing of a disk-shaped mother board used for a turbo machine rotor, and specifically relates to a digital precision grinding method for arc end faces teeth on a disk-shaped mother board. Annular master plate (hereinafter sometimes referred to as " Disk-shaped mother board ") and more specifically to a digital precision grinding method for arc end face teeth on a disk-shaped mother board.

Background Art

[0002] Arc end face teeth are precise end face splines, used for connection at the shaft end, playing roles of positioning and torque transmission, and are an important technology for the core of a gas turbine. The processing ability of arc end face teeth is one of the main capabilities in the gas turbine manufacturing process. Also, the accuracy requirements for arc end face teeth are very high. It is necessary that the center lines of each convex tooth and concave tooth all face the center. Each tooth is evenly distributed along the circumference. The elements of the convex and concave teeth need to fit tightly and have equal curvature and high uniformity so that each tooth receives equal stress when transmitting torque. The main difficulty in processing arc end face teeth lies mainly in its detection standard tool. That is, the manufacturing of the disk-shaped mother board. The disk-shaped mother board is required to be processed based on designed relevant parameters, and the requirements for its geometric tolerances are very high, and the processing difficulty is very high. Furthermore, in the conventional method for processing arc end face teeth on a disk-shaped mother board, every time the grinding wheel finishes grinding a pair of gear teeth, the material of the disk-shaped mother board rotates 360° / Z times (Z is the number of gear teeth of the disk-shaped mother board), and the problem of cumulative angle error due to non-uniform angular distribution is not considered, nor is the error due to vibration considered. Therefore, there is a need for a method for processing arc end face teeth on a disk-shaped mother board that can reduce vibration and cumulative angle error.

Summary of the Invention

[0003] The object of the present invention is to provide a digital precision grinding method for arc end face teeth on a disk-shaped mother board in order to overcome the drawbacks of the prior art.

[0004] To achieve the above object, the present invention adopts the following technical solutions.

[0005] The digital precision grinding method of arc end face teeth on a disk-shaped mother board in the present invention is as follows: Replace the fixture on the rotary table of the numerically controlled machine tool with a four-jaw chuck, clamp the horizontally arranged disk-shaped mother board material with the four-jaw chuck, attach the grinding wheel to the spindle of the machine tool, and attach a three-dimensional point cloud device to the machine tool. Next, create a three-dimensional coordinate system of the machine tool with the reference point of the numerically controlled machine tool as the origin, and determine the center coordinate point of the rotary table, the center coordinate point of the grinding wheel, and the center coordinate point of the three-dimensional point cloud device in step 1; Identify the center of the disk-shaped mother board material by the three-dimensional point cloud device, translate the disk-shaped mother board material by the four locking claws of the four-jaw chuck, and make the center of the disk-shaped mother board material coincide with the center of the rotary table in the vertical direction. Subsequently, set the relative positional relationship between the grinding wheel and the disk-shaped mother board material during grinding. Then, drive the spindle by the numerically controlled machine tool so that the grinding wheel moves parallel, and make the relative position between the grinding wheel and the disk-shaped mother board material satisfy a predetermined relative material positional relationship, and make the grinding wheel located above the disk-shaped mother board material in step 2; Step 3 of controlling a grinding wheel by a numerically controlled machine tool to perform grinding on a disk-shaped mother board material. The grinding process includes: a first step of controlling the descent of a grinding wheel rotated by a numerically controlled machine tool to grind a pair of gear teeth on the disk-shaped mother board material; a second step of controlling, by the numerically controlled machine tool, the grinding wheel to rise to its original position after completing the grinding of a pair of gear teeth on the disk-shaped mother board material, and controlling a rotary table so that the disk-shaped mother board material rotates by a large angle β in conjunction with the rising, and then repeating the first step; a third step of evaluating the maximum wear depth on the surface of the grinding wheel after grinding and the maximum stress during the grinding process, and if the maximum wear depth or the maximum stress is greater than the corresponding predetermined discard value, replacing it with a new grinding wheel, and if not, performing the following process. When the maximum wear depth or the maximum stress is less than the corresponding predetermined grinding value, continue to execute, and if not, remove the grinding wheel, acquire the surface shape and wear characteristic data of the grinding wheel, regrind the grinding wheel, then reinstall the grinding wheel, and then repeat the second step until the grinding of one group of gear teeth is completed. Then, control the grinding wheel by the numerically controlled machine tool to rise to its original position. Here, one group of gear teeth refers to all the gear teeth that can be processed within a range not exceeding 360° when the disk-shaped mother board material is sequentially rotated by a large angle β. A fourth step of controlling the rotary table by the numerically controlled machine tool to rotate the disk-shaped mother board material by a small angle α, and then repeating the first to third steps; and a fifth step of repeating the fourth step until the grinding of all the gear teeth on the disk-shaped mother board material is completed. Before grinding the disk-shaped mother board material, a neural network is introduced to optimize the large angle β and the small angle α for rotating the disk-shaped mother board material. The large angle β for each rotation of the disk-shaped mother board material is made the same and an initial value is given, and the small angle α for each rotation is designed to be variable. During the process of grinding the disk-shaped mother board material, if the position of the disk-shaped mother board material is displaced after completing the grinding of a pair of teeth on the disk-shaped mother board material, adjust the position of the disk-shaped mother board material. Step 3 Step 4 of inspecting whether there are defects on the surface of each arc end face tooth of the gear teeth obtained by grinding the disk-shaped mother board material, and if there are defects, regrinding the defective part.

[0006] Preferably, in the step 2, as a step of identifying the center of the disk-shaped mother board material by the three-dimensional point group device, a three-dimensional point group of the disk-shaped mother board material is collected using the three-dimensional point group device, and an inner contour circle or an outer contour circle of the upper surface of the disk-shaped mother board material is obtained. Any two points on the inner contour circle or the outer contour circle are taken as a chord, and then any other two points on the inner contour circle (the circle forming the hole wall of the center hole provided in the disk-shaped mother board material) or the outer contour circle (the circle forming the outer wall of the disk-shaped mother board material) are taken as a chord. The perpendicular bisectors of the two chords are created, and the intersection point of the two perpendicular bisectors is taken as the center of the disk-shaped mother board material.

[0007] Preferably, in the step 2, the relative positional relationship between the grinding wheel and the disk-shaped mother board material during grinding is set, and the position coordinates of two ideal machining points of the grinding wheel and the disk-shaped mother board material during grinding are set. Here, the ideal machining point is defined as the intersection point of a circle with a radius equal to the average value of the inner circle radius and the outer circle radius of the grinding wheel and a circle with a radius equal to the average value of the inner circle radius and the outer circle radius of the disk-shaped mother board material when the relative position between the grinding wheel and the disk-shaped mother board material satisfies the relative positional relationship of a predetermined material.

[0008] More preferably, the relative positional relationship between the grinding wheel and the disk-shaped mother board material during grinding satisfies the following relational expression.

Equation

[0009] Preferably, the optimization process of the small rotation angle α of the disk-shaped mother board material is as follows: (a) First step: Set the feature vectors input to the neural network. The feature vectors include the operation a t and the processing state S t in the processing process of the disk-shaped mother board material. The operation a t is defined as a t =(β, α), where α is a matrix composed of each small rotation angle α of the disk-shaped mother board material, The processing state S t is

Number

[0010] Preferably, in the step 3, in the grinding process of each pair of gear teeth on the disk-shaped mother board material, the vibration signal value of the disk-shaped mother board material is detected by a vibration sensor fixed to the disk-shaped mother board material. When the detected vibration signal value exceeds a predetermined value, it is determined that a displacement has occurred in the position of the disk-shaped mother board material after the grinding process of the pair of gear teeth is completed. The position adjustment process of the disk-shaped mother board material is as follows. After the processing of the pair of gear teeth is completed, the position coordinates of the actual processing points of these two gear teeth are calculated from the image taken by the camera, and further, the position error between these two actual processing points and the corresponding ideal processing points is obtained. The disk-shaped mother board material is translated by the four locking claws of the four-jaw chuck to align the positions of these two actual processing points with the positions of the corresponding ideal processing points.

[0011] Preferably, as the decision process for replacing or regrinding the grinding wheel in the step 3, the grinding amount, working time, and working conditions on the surface of the grinding wheel are input into a prediction simulation model, the maximum wear depth on the surface of the grinding wheel after grinding and the maximum stress in the grinding process are evaluated. When the maximum wear depth or the maximum stress is greater than the corresponding predetermined discard value, it is replaced with a new grinding wheel. Otherwise, it is judged as follows. When the maximum wear depth or the maximum stress is less than the corresponding predetermined grinding value, it continues to be executed. Otherwise, the grinding wheel is removed, and the surface shape and wear characteristic data of the grinding wheel are obtained using laser triangulation measurement, and the surface of the grinding wheel after processing is reground.

[0012] More preferably, the construction process of the prediction simulation model is The first step of constructing a grinding simulation model for a disk-shaped mother board material and a grinding wheel, in which a three-dimensional simulation model of the disk-shaped mother board material and the grinding wheel is constructed by Solidworks, a three-dimensional CAD software, and the three-dimensional simulation model of the disk-shaped mother board material and the grinding wheel is introduced into ABAQUS, a finite element analysis software. At this time, the material properties of the disk-shaped mother board and the grinding wheel are input, and the material properties include elastic modulus, Poisson's ratio and wear coefficient. Next, mesh division is performed on the three-dimensional simulation model of the disk-shaped mother board material and the grinding wheel. Then, the boundary conditions of the three-dimensional simulation model of the disk-shaped mother board material and the grinding wheel are set to obtain the grinding simulation model of the disk-shaped mother board material and the grinding wheel, the first step. Set the working conditions, where the working conditions include load, rotational speed, working time and working temperature. Next, run ABAQUS to perform grinding simulation, obtain the wear depth and stress distribution of the three-dimensional simulation model of the grinding wheel after grinding simulation processing, and draw the wear depth distribution diagram and stress distribution diagram of the three-dimensional simulation model of the grinding wheel after grinding simulation processing by the post-processing of ABAQUS, the second step. Change the grinding amount on the surface of the grinding wheel and repeat the first step and the second step, the third step, and Based on the grinding amount on the surfaces of different grinding wheels, different working times, different working conditions and the corresponding distribution diagrams of the wear depth of the grinding wheel and stress distribution diagram data obtained by simulation, create a training data set and a verification data set, and perform training and verification on the constructed prediction simulation model for evaluating the maximum wear depth on the surface of the grinding wheel after processing and the maximum stress in the processing process, the fourth step.

[0013] Preferably, in the process of step 4, it is inspected whether there are defects on the surface of each arc end face tooth by using an eddy current detection method. For the defects, the coordinates of each discrete point in the defect area are obtained by using a three-dimensional point cloud device, the normal deviation between each discrete point and the corresponding point coordinates on the surface of the arc end face tooth of the theoretical disk-shaped mother board is calculated, and when the variance of the normal deviation between each discrete point in the defect area and the corresponding point coordinates on the surface of the theoretical arc end face tooth is greater than a predetermined value, the defect area on the surface of the arc end face tooth is reground.

Advantages of the Invention

[0014] The present invention has the following beneficial effects over the prior art. 1. By using the method of uniform distribution of errors, the present invention can reduce the cumulative value of angular errors. Specifically, in the present invention, the gear teeth that need to be processed on the disk-shaped mother board are divided into a plurality of groups, each group contains K pairs of gear teeth arranged at equal intervals, the angle between two adjacent gear tooth pairs in each group is β, and in the processing process, the grinding of the gear teeth in each group is completed in sequence, and every time the processing of the last pair of gear teeth in each group is completed, the disk-shaped mother board material is rotated by a small angle α, and every time the processing of other tooth pairs in each group is completed, the disk-shaped mother board material is rotated by an angle β, so that the angular distribution in the processing process becomes more uniform, the cumulative angular error is evenly distributed among K groups, the cumulative angular error is reduced, and the processing efficiency is improved. The present invention takes into account the influence of the rotation accuracy of the rotary table of the numerical control machine tool on the processing accuracy of the disk-shaped mother board, and by introducing a neural network before grinding the disk-shaped mother board material to optimize the large angle β and the small angle α for rotating the disk-shaped mother board material, reasonable large angle β and each small angle α for rotation can be quickly calculated and obtained, ensuring the processing accuracy and improving the processing efficiency. Furthermore, the present invention considers the problem that the position of the disk-shaped mother board material is displaced due to the large vibration generated when the grinding wheel contacts the disk-shaped mother board material, and by adjusting the position of the disk-shaped mother board material after its position is displaced, the processing accuracy of the disk-shaped mother board is further ensured. Therefore, the present invention can realize the digital high-precision processing of the arc end face teeth of the disk-shaped mother board.

[0015] 2. The present invention evaluates the maximum wear depth of the grinding wheel surface after grinding and the maximum stress in the grinding process. If the maximum wear depth or the maximum stress is greater than the corresponding discard value, the grinding wheel is replaced with a new one. Otherwise, the following judgment is made. If the maximum wear depth or the maximum stress is less than the corresponding predetermined grinding value, the operation continues. Otherwise, the grinding wheel is removed, the surface shape and wear characteristic data of the grinding wheel are obtained, the grinding wheel is reground, and then the grinding wheel is reinstalled. Thereby, the machining accuracy of the next disk-shaped mother board is ensured, and the inspection time of the grinding wheel is shortened.

[0016] 3. In the present invention, after the machining of the disk-shaped mother board is completed, the surface of the arc end face teeth on the disk-shaped mother board is inspected. It is inspected whether there are defects on the surface of the arc end face teeth on the disk-shaped mother board. If there are defects on the surface of the arc end face teeth, regrinding is performed on the defective part of the surface of the arc end face teeth to improve the machining accuracy of the surface of the arc end face teeth. Here, in the present invention, by using the eddy current detection method to inspect the surface of the arc end face teeth, the inspection effects of high sensitivity, high efficiency, and high speed can be realized. In addition, the defect area is discretized by using the point cloud technology to form discrete point coordinates, and the normal deviation between the discrete points on the actual surface of the arc end face teeth and the corresponding points on the ideal surface of the arc end face teeth is calculated. When the variance of the normal deviation between each discrete point in the defect area and the corresponding point coordinates on the theoretical surface of the arc end face teeth is greater than a predetermined value, regrinding is performed on the corresponding defect area to achieve the purpose of local quantitative regrinding.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Embodiments for Carrying out the Invention

[0018] The present invention will be described in detail below in conjunction with the drawings.

[0019] As shown in FIGS. 1, 2, 3, 4 and 5, the digital precision grinding method for the arc end face teeth of the disk-shaped mother board in the present invention includes the following four steps.

[0020] Step 1: Replace the jig on the rotary table of the numerically controlled machine tool (NC machine tool) with a four-jaw chuck (intelligent fixed jaw force control four-jaw chuck), clamp the material of the horizontally arranged disk-shaped mother board with the four-jaw chuck, attach the grinding wheel to the spindle of the machine tool, and attach a three-dimensional point cloud device (for example, a laser radar, a three-dimensional laser scanner) to the machine tool. Here, both the material of the disk-shaped mother board and the grinding wheel are in an overall circular ring shape. Next, establish the three-dimensional coordinate system of the machine tool with the reference point of the numerically controlled machine tool (hereinafter sometimes simply referred to as the "machine tool") as the origin, and determine the center coordinate point of the rotary table, the center coordinate point of the grinding wheel (the center coordinate point of the machine tool spindle), and the center coordinate point of the three-dimensional point cloud device. Here, the accuracy of the three-dimensional coordinate system of the machine tool depends on the grating scale for detecting linear displacement or angular displacement, and accurately feeds back the position information of the machine tool to the numerical control system, thereby realizing high-precision closed-loop control and improving the processing accuracy, efficiency, and stability.

[0021] Step 2: Identify the center of the disk-shaped mother board material by the three-dimensional point group device, translate the disk-shaped mother board material by the four locking claws of the four-jaw chuck, and align (make coincide) the center of the disk-shaped mother board material and the center of the rotary table in the vertical direction. Here, the process of identifying the center of the disk-shaped mother board material by the three-dimensional point group device is as follows. Obtain the upper surface contour circle of the disk-shaped mother board material using three-dimensional point group technology, take any two points on the circle as chords, then take any other two points on the circle as chords, and create the perpendicular bisectors of the two chords. The intersection of the two perpendicular bisectors is taken as the center of the disk-shaped mother board material.

[0022] Next, set the relative positional relationship between the grinding wheel during processing and the disk-shaped mother board material, and set the position coordinates of the two ideal processing points of the grinding wheel and the disk-shaped mother board material during grinding. Here, the ideal processing point is defined as the intersection of a circle with a radius equal to the average value of the inner diameter radius and the outer diameter radius of the ring-shaped grinding wheel and a circle with a radius equal to the average value of the inner circle radius and the outer circle radius of the disk-shaped mother board material when the relative position between the grinding wheel and the rotary grinding disk material satisfies a predetermined relative positional relationship between the materials.

[0023] Next, drive the spindle by the numerically controlled machine tool so that the grinding wheel moves parallel, and make the relative position between the grinding wheel and the disk-shaped mother board material satisfy a predetermined relative positional relationship between the materials, and make the grinding wheel located above the disk-shaped mother board material in the initial state. Here, in the grinding process of the disk-shaped mother board material, since the processing position is changed by controlling the lifting and lowering of the grinding wheel and the rotation of the disk-shaped mother board material, the displacement of the grinding wheel in the Z direction is not considered. When determining the relative positional relationship between the grinding wheel and the disk-shaped mother board material during grinding, make the relative position between the grinding wheel and the disk-shaped mother board material satisfy the following relational expression.

Number

[0024] Step 3: As shown in FIG. 6, control the grinding wheel by a numerically controlled machine tool to perform grinding on the disk-shaped base plate. The grinding process includes: a first step of controlling the descent of the rotating grinding wheel by a numerically controlled machine tool to grind a pair of gear teeth on the disk-shaped base plate; a second step of controlling, by the numerically controlled machine tool, the grinding wheel to rise to the original position after completing the grinding of a pair of gear teeth on the disk-shaped base plate, and controlling the rotary table so that the disk-shaped base plate rotates by a large angle β in conjunction with the rise, and then repeating the first step; a third step of evaluating the maximum wear depth on the surface of the grinding wheel after grinding and the maximum stress during the grinding process, and if the maximum wear depth or the maximum stress is greater than the corresponding predetermined discard value, replacing it with a new grinding wheel, and if not, performing the following processing. When the maximum wear depth or the maximum stress is less than the corresponding predetermined grinding value, continue to execute, otherwise remove the grinding wheel, acquire the surface shape and wear characteristic data of the grinding wheel, regrind the grinding wheel, then reinstall the grinding wheel, and then repeat the second step until the grinding of one group of gear teeth is completed. Control the grinding wheel by the numerically controlled machine tool to rise to the original position. Here, one group of gear teeth refers to all the gear teeth that can be processed within a range not exceeding 360° when the disk-shaped base plate is sequentially rotated by a large angle β. A fourth step of controlling the rotary table by the numerically controlled machine tool to rotate the disk-shaped base plate by a small angle α, and then repeating the first to third steps; and a fifth step of repeating the fourth step until the grinding of all the gear teeth on the disk-shaped base plate is completed.

[0025] Considering the influence of the rotational accuracy of the rotary table of the numerically controlled machine tool on the machining accuracy of the disk-shaped master plate, the small angle α may not reach the designed angle (i.e., the ideal small angle) between two adjacent gear teeth on the disk-shaped master plate. Therefore, it is necessary to introduce a neural network before grinding the disk-shaped master plate material (i.e., before executing Step 3) to optimize the large angle β and the small angle α for rotating the disk-shaped master plate material. Here, the large angle β for each rotation of the disk-shaped master plate material is designed to be the same (however, it is necessary to optimize the specific value of β), the small angle α for each rotation is designed to be variable (the specific value of α may alternately repeat large values and small values), and in the initial state, it is assumed that β = 360° / K. Here, K is the number of pairs of gear teeth included in each group, and both β and K are integers. The optimization process is as follows.

[0026] (a) Set the feature vector input to the neural network. The feature vector of the neural network includes the operation a t and the machining state S t in the machining process of the disk-shaped master plate material. The operation a t is defined as a t =(β, α). α is a matrix composed of each small angle α when the disk-shaped master plate material rotates. The machining state S t is

Number

[0027] For example, if it is necessary to machine 88 teeth on the disk-shaped blank material and the number of pairs of gear teeth K = 4 for each group of teeth, when machining the first pair of gear teeth with a grinding wheel, S t =(88, 1, 0, 0, 0), and when machining the second pair of gear teeth with a grinding wheel, S t =(88, 2, β, 0, 0), and when machining the fifth pair of gear teeth with a grinding wheel, St = (88, 5, 3β + α, α, ε).

[0028] (b) Construct a neural network architecture. Employ the particle swarm algorithm as the optimization algorithm for the neural network, and set the learning factor and inertia weight of the neural network, and set the maximum number of generated particle swarms, the number of iterations, and the step size. Here, the optimization process in the particle swarm algorithm is as follows. (1) Input the training samples. The parameters in the training samples include the number of teeth Z of the disk-shaped blank, the large angle β in the initial state, the ideal small angle α1, and each small angle α of the disk-shaped blank material in the initial state. (2) Within a predetermined range of the small angle α (i.e., within the allowable range of the local error ε), the optimal value P best of the small angle α corresponding to the individual and the optimal value G bestUpdate it. Here, the optimal value P of the small angle α corresponding to the individual best is the optimal solution experienced by a single particle in its iterative process, and the optimal value G of the small angle α corresponding to the particle swarm best is the most preferable one among the optimal solutions of the corresponding small angle α for all particles in the current iterative process. The update process is as follows. (i) Calculate the fitness. For each individual S i , calculate the fitness f(X i ) corresponding to the current value X of each small angle α. The fitness function is a function inversely proportional to εmax. i (ii) Compare the fitnesses. Compare each calculated fitness f(X i ) with the fitness f(P best ) calculated based on the optimal value of the corresponding small angle α of the individual up to the current time (in the initial state, each initial small angle α is taken as the optimal value of each small angle α for the individual). If f(X i ) > f(P best ), update the optimal value of the small angle α corresponding to the individual to P best = X i , and at the same time update its fitness value to f(P best ) = f(X i ). (3) After updating all individuals, compare each updated fitness value f(P best ) with the fitness f(G best ) calculated based on the optimal value G of the small angle α corresponding to the particle swarm (in the initial state, the initial small angle α is taken as the optimal value of the small angle α of the particle swarm). If f(P best ) > f(G best ), update the optimal value of the small angle α corresponding to the particle swarm to G best = P best , and at the same time update its fitness value to f(G best ) = f(P best ). best (4) Update the processing state S of each individual t . ​​When εmax is smaller than a predetermined value, output the optimized small angle α, large angle β, and the number of iterations of the disk-shaped mother board material; otherwise, update the large angle β to the next value in descending order within the selectable values (the selectable values of the large angle β are calculated based on β = 360° / K and are integers, and it is necessary to ensure that the number of gear teeth pairs K in each group after updating is an integer), reset each small angle α of the disk-shaped mother board material to the initial value, and then return to (2).

[0029] In this embodiment, set the step size to the minimum rotation angle of the numerically controlled machine tool, and set the learning factors of the neural network to c1 = c2 = 2. Depending on the magnitude of the inertial weight, it is determined whether the neural network has excellent global optimization ability or excellent local optimization ability. Since the dynamic inertial weight is more likely to find an optimization result superior to the fixed value, set the inertial weight w within the range of 0.4 to 0.9.

[0030] Furthermore, considering that when the grinding wheel comes into contact with the disk-shaped mother board material, a large vibration may occur, which may cause a (slight) deviation in the position of the disk-shaped mother board material, in order to ensure the processing accuracy of the disk-shaped mother board, after the grinding process of a pair of teeth on the disk-shaped mother board material is completed, if the position of the disk-shaped mother board material is deviated, the position of the disk-shaped mother board material is adjusted. Of course, it should be noted that the deviation is small and the processing accuracy of the pair of teeth still meets the requirements. Only to avoid the continuous accumulation of subsequent processing errors, when it is considered that a slight deviation has occurred, the position adjustment of the disk-shaped mother board material is immediately performed. Specifically, in the grinding process of each pair of gear teeth on the disk-shaped mother board material, the vibration signal value of the disk-shaped mother board material is detected by a vibration sensor fixed to the disk-shaped mother board material. When the detected vibration signal value exceeds a predetermined value, it is determined that a (slight) deviation has occurred in the position of the disk-shaped mother board material after the grinding process of the pair of gear teeth is completed. After the processing of the pair of gear teeth is completed, the position coordinates of the actual processing points of these two gear teeth (the intersection of the currently processed gear tooth on the disk-shaped mother board material after grinding and the circle with the average value of the inner circle radius and the outer circle radius on the disk-shaped mother board material as the radius) are calculated by the image taken by a camera (for example, a microscope camera). Furthermore, the position error between these two actual processing points and the corresponding ideal processing points is obtained, and the disk-shaped mother board material is translated by the four locking claws of the four-jaw chuck to align the positions of these two actual processing points with the positions of the corresponding ideal processing points.

[0031] Furthermore, the decision-making process for the replacement or re-grinding of the grinding wheel in Step 3 is as follows. Input the grinding amount, working time, and working conditions on the grinding wheel surface into the prediction simulation model to evaluate the maximum wear depth of the grinding wheel surface after grinding and the maximum stress during the grinding process. If the maximum wear depth or the maximum stress is greater than the corresponding predetermined discard value, replace it with a new grinding wheel. Otherwise, make the following judgment. If the maximum wear depth or the maximum stress is less than the corresponding predetermined grinding value, continue the execution. Otherwise, remove the grinding wheel, and use laser triangulation to obtain the surface shape and wear characteristic data of the grinding wheel, and re-grind the surface of the grinding wheel after processing. Here, the process of obtaining the surface shape of the grinding wheel by laser triangulation is as follows. Irradiate the grinding wheel surface with a laser beam emitted from the laser. Due to the unevenness and curvature changes on the grinding wheel surface, the beam projected due to the change in the contour position fluctuates, and the image formed on the PSD (Position Sensitive Detector) detection element also undergoes a corresponding displacement change. The PSD detection element collects the image formed by the reflected light, performs filtering and smoothing processing on the collected image, and based on the position correspondence relationship between the imaging point and the measurement point, the height value of the grinding wheel surface can be obtained. By performing interpolation and reconstruction on the scattered sampling data, the three-dimensional coordinate data of the grinding wheel surface is obtained, and using the drawing software, a three-dimensional shape diagram of the grinding wheel surface is drawn using the three-dimensional coordinate data of the collected grinding wheel surface, and data analysis is performed to obtain the wear characteristic data of the grinding wheel surface. Here, add a standard sphere to adjust the sensor head so that the emitted light ray is always perpendicular to the intersection line of the grinding wheel surface and the triangular plane of the light, and eliminate the measurement deviation caused by the tilt angle.

[0032] Furthermore, in order to save the detection time of the maximum wear depth of the grinding wheel surface after grinding and the maximum stress during the grinding process, a prediction simulation model for the maximum wear depth of the grinding wheel surface after grinding and the maximum stress during the grinding process is constructed. Then, during the grinding process of the disk-shaped mother board material, the maximum wear depth of the grinding wheel surface after grinding and the maximum stress in the grinding process are predicted by the prediction simulation model. Here, the construction process of the prediction simulation model is as follows.

[0033] Step 1: Construct a grinding simulation model of the disk-shaped mother board material and the grinding wheel. Use Solidworks to construct a three-dimensional simulation model of the disk-shaped mother board material and the grinding wheel, and import the three-dimensional simulation models of the disk-shaped mother board material and the grinding wheel into ABAQUS. At this time, input the material properties of the disk-shaped mother board and the grinding wheel. Here, the material properties include the elastic modulus, Poisson's ratio, and wear coefficient. Next, perform mesh division on the three-dimensional simulation models of the disk-shaped mother board material and the grinding wheel. Here, hexahedral virtual elements can be used for mesh division. Also, for the wear area of the three-dimensional simulation model of the grinding wheel, appropriately increase the mesh density to ensure that the mesh density is high enough to capture the details of wear. For the non-wear area of the three-dimensional simulation model of the grinding wheel, the mesh density can be appropriately decreased, which can reduce the operating time of the computer. Then, set the boundary conditions of the three-dimensional simulation models of the disk-shaped mother board material and the grinding wheel to obtain the grinding simulation model of the disk-shaped mother board material and the grinding wheel.

[0034] The second stage: Set the working conditions. The working conditions include load, rotational speed, working time, and working temperature, ensuring that the simulated environment is consistent with the actual processing environment. Next, execute ABAQUS to perform the grinding simulation, obtain the wear depth and stress distribution of the three-dimensional simulation model of the grinding wheel after the grinding simulation process, and draw the wear depth distribution diagram and stress distribution diagram of the three-dimensional simulation model of the grinding wheel after the grinding simulation process by using the post-processing function of ABAQUS.

[0035] The third stage: Change the grinding amount on the surface of the grinding wheel and repeat the first and second stages.

[0036] The fourth stage: Based on the distribution diagrams and stress distribution diagram data of the corresponding grinding wheel wear depths obtained from different grinding amounts on the surface of the grinding wheel, different working times, different working conditions, and simulations, create a training dataset and a validation dataset, and perform training and validation on the constructed prediction simulation model for evaluating the maximum wear depth on the surface of the grinding wheel after processing and the maximum stress in the processing process.

[0037] Step 4: Check whether there are defects on the surface of each arc end face tooth of the disk-shaped mother board obtained by grinding the disk-shaped mother board material. If there are defects on the surface of the arc end face teeth of the disk-shaped mother board, re-grind the defective parts on the surface of each arc end face tooth of the disk-shaped mother board to improve the machining accuracy of the disk-shaped mother board. Here, use the eddy current flaw detection method to check whether there are defects on the surface of each arc end face tooth of the disk-shaped mother board. Apply a changing magnetic field around the disk-shaped mother board and at the same time pass an alternating current through the detection coil to generate an alternating magnetic field perpendicular to the disk-shaped mother board. By approaching the detection coil to the arc end face teeth of the disk-shaped mother board, eddy currents are induced on the surface of the arc end face teeth of the disk-shaped mother board. At the same time, a magnetic field opposite to the original magnetic field is generated, and a part of the original magnetic field is cancelled. As a result, changes are brought about in the resistance and inductance of the detection coil. If there are defects on the surface of the arc end face teeth, the intensity and distribution of the eddy current field change, the coil impedance changes, and by analyzing these changes, it is possible to determine whether there are surface defects on the arc end face teeth of the disk-shaped mother board. Also, adjust the correction coefficient used according to the conductivity of the disk-shaped mother board material. When the conductivity of the disk-shaped mother board material is high, select a high correction coefficient; otherwise, select a low correction coefficient.

[0038] Use a three-dimensional point cloud device to obtain the coordinates of each discrete point in the surface defect area of the arc end face teeth of the disk-shaped mother board. Calculate the normal deviation between each discrete point and the corresponding point coordinates on the surface of the theoretical arc end face teeth of the disk-shaped mother board. The normal deviation value is an index that reflects the difference between the actual surface of the arc end face teeth and the ideal surface of the arc end face teeth. If the variance of the normal deviation between each discrete point in the defect area and the corresponding point coordinates on the surface of the theoretical arc end face teeth is greater than a predetermined value, re-grind the defect area on the surface of the arc end face teeth.

Claims

1. A method for grinding arc end faces of teeth with respect to an annular master plate, comprising: replacing a jig on a rotary table of a numerically controlled machine tool with a four-jaw chuck, clamping a horizontally disposed annular master plate material between the four-jaw chuck, mounting an annular grinding wheel on a spindle of the numerically controlled machine tool, and mounting a three-dimensional point cloud device on the numerically controlled machine tool; then creating a three-dimensional coordinate system of the numerically controlled machine tool with a reference point of the numerically controlled machine tool as the origin, and determining a center coordinate point of the rotary table, a center coordinate point of the grinding wheel, and a center coordinate point of the three-dimensional point cloud device in Step 1; identifying the center of the annular master plate material by the three-dimensional point cloud device, horizontally translating the annular master plate material by the four locking claws of the four-jaw chuck, aligning the center of the annular master plate material with the center of the rotary table in the vertical direction, then setting a relative positional relationship between the grinding wheel and the annular master plate material during grinding, and then driving the spindle by the numerically controlled machine tool so that the grinding wheel translates horizontally, and making the relative position between the grinding wheel and the annular master plate material satisfy the relative positional relationship and the grinding wheel is located above the annular master plate material in Step 2; Step 3 of controlling the grinding wheel by the numerical control machine tool to perform grinding on the annular mother board material. The grinding process includes: controlling the descent of the grinding wheel rotated by the numerical control machine tool, and performing grinding on the annular mother board material at each of two locations where the grinding wheel and the annular mother board material intersect so that a pair of gear teeth are formed, which is the first step; controlling by the numerical control machine tool to make the grinding wheel rise to its original position after the grinding of the pair of gear teeth on the annular mother board material is completed, and controlling the rotary table so that the annular mother board material rotates by a large angle β in conjunction with the rise, and then repeating the first step, which is the second step; evaluating the maximum wear depth on the surface of the grinding wheel after grinding and the maximum stress during the grinding process, and if the maximum wear depth or the maximum stress is greater than the corresponding predetermined discard value, replacing it with a new grinding wheel, and if not, performing the following process, which is the third step. When the maximum wear depth or the maximum stress is less than the corresponding predetermined grinding value, continue to execute, and if not, remove the grinding wheel, acquire the surface shape and wear characteristic data of the grinding wheel, regrind the grinding wheel, then reinstall the grinding wheel, and then repeat the second step until the grinding of one group of gear teeth is completed. Controlling the grinding wheel by the numerical control machine tool to rise to its original position. Here, one group of gear teeth refers to all the gear teeth that can be processed within a range not exceeding 360° when the annular mother board material is sequentially rotated by the large angle β, which is the third step; controlling the rotary table by the numerical control machine tool to rotate the annular mother board material by a small angle α, and then repeating the first step to the third step, which is the fourth step; and repeating the fourth step until the grinding of all the gear teeth on the annular mother board material is completed, which is the fifth step. Here, before grinding the annular mother board material, an initial value is given to the large angle β, and a neural network is introduced to optimize the large angle β and the small angle α for rotating the annular mother board material so that the large angle β for each rotation of the annular mother board material is the same as each other and the small angle α for each rotation is variable. During the process of grinding the annular mother board material, after the grinding of the pair of gear teeth on the annular mother board material is completed,If the position of the annular mother board material is displaced, step 3 of adjusting the position of the annular mother board material, grinding the annular master plate material to obtain gear teeth with arc end faces, and inspecting whether there are defects on the surface of each arc end face tooth. If there are defects, re-grinding the defective parts in Step 4. A method for grinding arc end faces of teeth with respect to an annular master plate, characterized by comprising the above steps.

2. In Step 2, as a step of identifying the center of the annular master plate material by the three-dimensional point cloud device, collecting a three-dimensional point cloud of the annular master plate material using the three-dimensional point cloud device, obtaining an inner contour circle or an outer contour circle of the upper surface of the annular master plate material, taking any two points on the inner contour circle or the outer contour circle as a chord, then taking any other two points on the inner contour circle or the outer contour circle as a chord, creating perpendicular bisectors of the two chords, and taking the intersection point of the two perpendicular bisectors as the center of the annular master plate material. The method for grinding arc end faces of teeth with respect to an annular master plate according to Claim 1, characterized by the above steps.

3. In the step 2, the relative positional relationship between the grinding wheel and the annular mother plate material during the grinding process is set, and the position coordinates of two ideal machining points of the grinding wheel and the annular mother plate material during the grinding process are set. Here, when the relative position between the grinding wheel and the annular mother plate material satisfies the relative positional relationship, the ideal machining point is defined as the intersection point between a circle with the average value of the inner circle radius and the outer circle radius of the grinding wheel as the radius and a circle with the average value of the inner circle radius and the outer circle radius of the annular mother plate material as the radius. The method for grinding the arc end face teeth of the annular mother plate according to claim 1, characterized in that.

4. The relative positional relationship between the grinding wheel and the annular mother plate material during the grinding process is 【Number 1】 Satisfies the following relational expression, where γ is the tooth space angle, which is the angle between the straight line connecting one of the ideal machining points and the center of the annular mother plate material when grinding the annular mother plate material with the grinding wheel, and the straight line connecting the center of the grinding wheel and the center of the annular mother plate material, N ｘ is the number of teeth between two of the ideal machining points when grinding the annular mother plate material with the grinding wheel, Z is the number of teeth of the annular mother plate, S is the distance between the center of the annular mother plate material and the center of the grinding wheel, R ｍｌ is the average value of the inner circle radius and the outer circle radius of the grinding wheel, and the method for grinding the arc end face teeth for the annular mother plate according to claim 3, characterized in that

5. The optimization process of the small angle α of the annular mother plate material is (a) A feature vector setting step of setting a feature vector to be input to a neural network, where the feature vector is operation a in the processing process of the annular mother board material ｔ and processing state S ｔ and includes operation a ｔ is defined as a = (β, α), where α is a matrix composed of each of the small angles α at which the annular mother board material rotates, and processing state S ｔ is ｔ to 【Number 2】 is defined, where i ｔ is the tooth number of the i-th pair of gear teeth at the current grinding position, and A ｃ is the cumulative rotation angle of the current annular mother board material, and A ｓ is the sum of each of the small angles α by which the annular mother board material has rotated, and εmax is the maximum value of the absolute value of the angle difference between the ideal small angle α, which is the design angle between two adjacent tooth angles and two adjacent gear teeth, within any one of the large angle β ranges １ and a feature vector setting step that is the maximum value of the absolute value of the angle difference between them (b) An architecture construction process for constructing a neural network architecture, which includes adopting a particle swarm algorithm as the optimization algorithm of the neural network, setting the learning factor and the inertia weight of the neural network, and setting the maximum number, the number of iterations, and the step size of the generated particle swarm. The optimization process in the particle swarm algorithm is (1) Inputting a training sample, wherein the parameters in the training sample include the number of gear teeth Z of the annular mother plate, the large angle β in the initial state, the ideal small angle α, １ and each of the small angles α of the annular mother plate material in the initial state. Within a predetermined range of the small angle α, the optimal value P of each small angle α corresponding to an individual ｂｅｓｔ and the optimal value G of each small angle α of the particle swarm ｂｅｓｔ are updated, and the update process is (i) Calculating the fitness, where the fitness function is a function inversely proportional to εmax, for each individual S ｉ with respect to the current value X of each of the small angles α ｉ calculate the fitness f(X ｉ ) corresponding thereto, (ii) Comparing the fitness values, where each calculated fitness value f(X ｉ ) is compared with the fitness value f(P ｂｅｓｔ ) calculated based on the optimal value of the corresponding small angle α of the individual up to the present, and if f(X ｉ ) > f(P ｂｅｓｔ ), then updating the optimal value of the small angle α corresponding to the individual to P ｂｅｓｔ = X ｉ , and at the same time updating its fitness value to f(P ｂｅｓｔ ) = f(X ｉ ), including this. (3) After updating all individuals, for each fitness f(P ｂｅｓｔ ), compare it with the fitness f(G ｂｅｓｔ ) obtained by calculating based on the optimal value G ｂｅｓｔ of the small angle α corresponding to the particle swarm. If f(P ｂｅｓｔ ) > f(G ｂｅｓｔ ), then update the optimal value of the small angle α corresponding to the particle swarm to G ｂｅｓｔ = P ｂｅｓｔ , and at the same time update its fitness to f(G ｂｅｓｔ ) = f(P ｂｅｓｔ ). (4) Update the processing state S of each individual, and ｔ ​ (5) When εmax is smaller than a predetermined value, output the small angle α, the large angle β, and the number of iterations of the optimized annular mother plate material. Otherwise, update the large angle β to the next value in descending order within the selectable values, reset each small angle α of the annular mother plate material to the initial value, and then return to (2). The selectable value of the large angle β is calculated based on β = 360° / K, and is an integer, and it is ensured that the number K of pairs of gear teeth in each group after the update is an integer. The method for grinding the arc end face teeth of the annular mother plate according to claim 1, characterized in that it includes.

6. In the step 3, in the grinding process of each pair of gear teeth on the circular ring-shaped mother board material, the vibration signal value of the circular ring-shaped mother board material is detected by a vibration sensor fixed to the circular ring-shaped mother board material. When the detected vibration signal value exceeds a predetermined value, it is determined that a displacement has occurred in the position of the circular ring-shaped mother board material after the grinding process of the pair of gear teeth is completed. After the processing of the pair of gear teeth is completed, the position coordinates of the actual processing points of the two gear teeth are calculated from an image taken by a camera, and further, the position error between the two actual processing points and the corresponding ideal processing points is obtained. The circular ring-shaped mother board material is horizontally translated by the four locking claws of the four-jaw chuck, and the position adjustment process of the circular ring-shaped mother board material is performed to align the positions of the two actual processing points with the positions of the corresponding ideal processing points. The method for grinding circular arc end face teeth on a circular ring-shaped mother board according to claim 3, characterized in that the above-mentioned position adjustment process is performed.

7. As the determination process for replacing or re-grinding the grinding wheel in the step 3, the grinding amount, working time, and working conditions on the surface of the grinding wheel are input into a prediction simulation model, the maximum wear depth on the surface of the grinding wheel after grinding and the maximum stress in the grinding process are evaluated. When the maximum wear depth or the maximum stress is greater than the corresponding predetermined discard value, it is replaced with a new grinding wheel. Otherwise, when the maximum wear depth or the maximum stress is smaller than the corresponding predetermined grinding value, the process continues. Otherwise, the grinding wheel is removed, and the surface shape and wear characteristic data of the grinding wheel are obtained using laser triangulation measurement, and the surface of the grinding wheel after processing is re-ground. The method for grinding circular arc end face teeth on a circular ring-shaped mother board according to claim 1, characterized in that the above is performed.

8. The construction process of the prediction simulation model is The first step of constructing the grinding simulation model of the annular mother board material and the grinding wheel, in which a three-dimensional simulation model of the annular mother board material and the grinding wheel is constructed by three-dimensional CAD software, the three-dimensional simulation models of the annular mother board material and the grinding wheel are introduced into finite element analysis software, and at this time, the material properties of the annular mother board and the grinding wheel including elastic modulus, Poisson's ratio and wear coefficient are input. Next, mesh division is performed on the three-dimensional simulation models of the annular mother board material and the grinding wheel, and then the boundary conditions of the three-dimensional simulation models of the annular mother board material and the grinding wheel are set to obtain the grinding simulation model of the annular mother board material and the grinding wheel. Set the working conditions including load, rotational speed, working time and working temperature. Next, execute the finite element analysis software to perform grinding simulation, obtain the wear depth and stress distribution of the three-dimensional simulation model of the grinding wheel after grinding simulation processing, and draw the wear depth distribution diagram and stress distribution diagram of the three-dimensional simulation model of the grinding wheel after grinding simulation processing by the post-processing of the finite element analysis software. The second step. The third step of changing the grinding amount on the surface of the grinding wheel and repeating the first step and the second step, and A training data set and a verification data set are created based on the grinding amount on the surface of different grinding wheels, different working times, different working conditions and the corresponding wear depth distribution diagrams and stress distribution diagram data obtained by simulation, and the maximum wear depth on the surface of the grinding wheel after processing and the maximum stress in the processing process are evaluated. The fourth step of training and verifying the constructed prediction simulation model is included. The method for grinding the arc end face teeth on the annular mother board according to claim 7, characterized in that it comprises the above steps.

9. The process of the step 4 includes inspecting whether there are defects on the surfaces of the arc end faces of each tooth using an eddy current detection method. For the defects, all the coordinates of the points with three-dimensional coordinates in the defect region, which are discrete points in the defect region obtained by the three-dimensional point cloud device, are acquired using the three-dimensional point cloud device. The normal deviation between each discrete point and the corresponding point coordinates on the surface of the arc end face tooth of the theoretical circular ring-shaped master plate is calculated. When the variance of the normal deviation between each discrete point in the defect region and the corresponding point coordinates on the surface of the theoretical arc end face tooth is greater than a predetermined value, re-grinding the defect region on the surface of the arc end face tooth, which is characterized in that it includes the grinding method for the arc end face tooth of the circular ring-shaped master plate according to claim 1.

Images