Automatic polishing method for electron beam selective area additive rotary body part

Abstract

An automatic polishing method for an electron beam selective area additive rotary body part comprises the steps that three-dimensional scanning point cloud data serves as input, after the data size is reduced and noisy points are removed, an ICP algorithm is used for point cloud matching, and the point cloud data are matched to template point cloud; extracting a region of interest required by path point calculation, obtaining path points by using an octree, and then removing abnormal points by using a shared bus equation for fitting to finally obtain the path points; performing convolution region segmentation, non-convolution feature recognition and equal-interval layering between path layers on the obtained path points to obtain a planned processing path and a matched parameter packet; a positioning rotating tool and a force feedback module are used for monitoring, feeding back, adjusting and precisely executing preset contact force in real time, so that a grinding tool follows the surface contour change of a workpiece in real time, meanwhile, the force and the stroke of the grinding tool are displayed in real time, and follow-up quantitative grinding of a part is completed; and the problem that the post-processing efficiency and precision of existing additive manufacturing control surface products are limited is solved.

Description

Technical Field

[0001] The invention relates to an automatic grinding method for electron beam selected area additive rotating body parts, belonging to the field of additive manufacturing post-processing. Background Art

[0002] Electron Beam Selective Melting (EBSM) is an additive manufacturing process that uses electron beam scanning, melting powder materials, and layer-by-layer deposition to manufacture three-dimensional metal parts. It is suitable for the forming and manufacturing of refractory high-performance metal materials such as titanium alloys and titanium-aluminum-based alloys. Compared with laser selective melting additive, electron beam selective melting additive has the following advantages: ① The vacuum environment reduces air pollution and is particularly suitable for the forming of active metals such as titanium; ② Higher energy efficiency, suitable for the manufacturing of refractory metals and high-temperature alloys; ③ Higher scanning speeds allow innovative preheating and melting strategies, which is essential for the forming of brittle metal materials (such as TiAl intermetallic alloys); ④ Reduce residual stress, reduce post-processing processes, and do not require dimensional support.

[0003] However, the prominent limiting factor of electron beam selective melting additive manufacturing is the poor surface integrity and surface finish of the formed parts. Patent CN202211597855.8 proposes an in-situ surface treatment device and a treatment method for electron beam fuse additive manufacturing. By rotating the tilting platform to adjust the facing position and angle of the printed part relative to the surface treatment electron beam gun, the surface treatment electron beam gun can in-situ treat the inner and outer surfaces of the printed part to complete the in-situ surface treatment and in-situ heat treatment. Patent CN202310645436.5 discloses a printing method for improving the surface quality of electron beam additive manufacturing, the method comprising: slicing parameter optimization, design of contour dotted line scanning strategy, circumference change and contour parameter adjustment, and the use of electron beam in-situ remelting to achieve the flattening and homogenization of the electron beam scanning traces, and improve the surface quality of the vertical workpiece side surface through slicing optimization and refined control of the contour melt channel size-morphology. Patent CN202010548795.5 relates to a method for suppressing powder splash in electron beam 3D printing of a powder bed, wherein the powder on the forming area of ​​the powder bed is preheated by a second round of electron beam scanning, and the powder on the preset area after the second round of electron beam scanning preheating is melted, and a third round of electron beam scanning preheating is performed after melting. There is basically no powder splashing during the 3D printing process, and the component after 3D printing has a high density and a small surface roughness. Patent CN202011355229.9 discloses a process for improving the surface quality of electron beam additively manufactured parts, which is achieved by accurately remelting the outer contour formed by electron beam melting of the added auxiliary material in the melting process of additive manufacturing by laser again, and then entering the cooling process of additive manufacturing for manufacturing to obtain the required surface quality of the part, and the roughness of the surface of the additively manufactured part after accurate remelting does not exceed Ra12.

[0004] Some of the above methods are targeted at electron beam fuse additive manufacturing, and have poor adaptability to electron beam selective melting additive manufacturing. Moreover, they are all for optimizing the parameters of the electron beam additive itself, and do not involve post-processing content. The traditional manual grinding method is inefficient, difficult to guarantee quality control, and the product quality is uneven. Therefore, a technical method for robot grinding electron beam selective melting additive products with flexible control, high precision, and convenient operation is urgently needed. Summary of the invention

[0005] The technical problem solved by the present invention is: to overcome the shortcomings of the prior art, to provide an automatic grinding method for electron beam selective additive manufacturing rotating body parts, and to improve the problems of limited post-processing efficiency and precision of existing additive manufacturing rudder products.

[0006] The technical solution of the present invention is: an automatic grinding method for electron beam selective area additive rotating body parts, comprising:

[0007] Taking 3D scanning point cloud data as input, downsampling and filtering are used to reduce the data volume and remove noise points, and then the ICP algorithm is used for point cloud matching, and the point cloud data is matched to the template point cloud; the cross-section method and envelope block are used to extract the region of interest required for path point calculation, and the octree is used to obtain the path points. Then, the shared busbar equation is used to remove abnormal points for fitting, and finally the path points are obtained;

[0008] The obtained path points are segmented into convolutional areas, identified as non-convolutional features, and layered with equal spacing between path layers to obtain the planned processing path and supporting parameter package;

[0009] Using the positioning rotary tooling and force feedback module, the preset contact force can be monitored, fed back, adjusted and accurately executed in real time, so that the grinding tool can follow the changes in the surface contour of the workpiece in real time. At the same time, the force and stroke of the grinding tool can be displayed in real time to complete the quantitative grinding of the parts.

[0010] Preferably, the downsampling subdivides the three-dimensional space of the rotating body part into three-dimensional grid voxel units of uniform size, and the centroid of the body or the point closest to the centroid is used as the sampling retention point.

[0011] Preferably, when the shared bus equation is estimated using the RANSAC algorithm:

[0012] After randomly initializing the model assumptions, the deviation of each point in the data set to the model is evaluated, and the internal and external points are distinguished according to the preset tolerance threshold; through repeated iterations, the maximum consensus set is identified, and then the least squares method is applied to fine-tune the model parameters to best fit this core point set.

[0013] Preferably, the convolution region segmentation is specifically as follows:

[0014] The workpiece is divided at the minimum rotation overlap angle. The smallest repeating unit after division is the spiral grinding area unit. The spiral area unit is further divided according to the busbar. When the curvature of the busbar changes suddenly, the grinding spiral unit is divided.

[0015] Preferably, when identifying non-convolutional features, the workpiece is rotated according to the minimum rotation overlap angle, and local features that cannot be overlapped are identified as non-convolutional features.

[0016] Preferably, when the path layers are equally spaced, the workpiece is divided based on a series of plane clusters, and the boundary between the plane and the surface contour of the workpiece is defined as a set of path points at this level.

[0017] Preferably, the positioning and rotating fixture enables the workpiece to be positioned and rotated according to the minimum rotation overlap angle, with a format diameter of ≥650mm and a repeat positioning accuracy of ≤0.005° / mm.

[0018] Preferably, the force feedback module includes a grinding force sensor, a micro-power actuator and an integrated force control unit, wherein:

[0019] The grinding force sensor monitors, feedbacks, adjusts and accurately executes the preset contact force in real time to ensure the accuracy of the surface contact force under any posture of the grinding tool;

[0020] The micro-actuator can follow and compensate the changes of the workpiece surface contour in real time to ensure the stable following of the tool and the workpiece surface;

[0021] The integrated force control unit displays the force and stroke measured by the grinding force sensor in real time, and sends instructions to the micro-power actuator to control the grinding tool and the automated operation of peripheral products.

[0022] Compared with the prior art, the present invention has the following advantages:

[0023] (1) The present invention solves the problem of surface defects of electron beam selected additive products through post-processing means, which is conducive to the promotion of new forming manufacturing processes and large-scale production.

[0024] (2) The present invention makes full use of the characteristics of the gyroscope parts, the grinding scheme is convenient and efficient, avoids repetitive programming work, improves grinding efficiency, better improves quality control, and ensures uniform product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of workpiece convolution region segmentation and non-convolution feature recognition according to an embodiment of the present invention;

[0026] Figure 2 This is a complete point cloud processed by the embodiment of the present invention.

[0027] Explanation of the accompanying drawings: 1-upper cylindrical surface outside the first area; 2-upper curved surface outside the second area; 3-lower curved surface outside the third area; 4-lower cylindrical surface outside the fourth area; 5-repeating boss in the fifth area; 6-upper cylindrical surface inside the sixth area; 7-upper curved surface inside the seventh area; 8-lower curved surface inside the eighth area; 9-lower cylindrical surface inside the ninth area; 10-first mounting platform; 11-second mounting platform; 12-XY plane. DETAILED DESCRIPTION

[0028] The present invention proposes an automatic grinding method for electron beam additive rotating body parts, which is characterized by obtaining workpiece contour detection technology based on three-dimensional scanning point cloud, planning processing paths and supporting parameter packages, and cooperating with positioning rotary tooling and force feedback modules to achieve shape-matching quantitative grinding of the workpiece.

[0029] The three-dimensional scanning point cloud acquisition workpiece contour detection technology uses three-dimensional scanning point cloud data as input, uses downsampling and filtering to reduce the data volume and eliminate noise points, then uses the ICP algorithm to match the point cloud data to the template point cloud, and uses the cross-section method and envelope block to extract the area of ​​interest required for path point calculation, uses the octree to obtain the path points, and then uses the shared bus equation to eliminate abnormal points for fitting, and finally obtains the path points.

[0030] The planned processing path includes convolutional area segmentation, non-convolutional feature recognition, and equidistant layering between path layers.

[0031] The matching parameter package includes different partition parameter packages and non-convolution characteristic parameter packages within each convolution unit, and different polishing convolution units repeatedly execute the combination of partition parameter packages within each convolution unit.

[0032] The positioning and rotating tooling comprises a clamping and locking device and an intermittent rotating device, wherein: the clamping and locking device is fixed on the bearing platform, the bearing platform is connected to the intermittent rotating device through a bearing, and the intermittent rotating device is connected to the system via a data interface; the positioning and rotating tooling enables the workpiece to be positioned and rotated according to the minimum rotation overlap angle, with a format diameter of ≥650mm and a repeat positioning accuracy of ≤0.005° / mm; in the embodiment of the present invention, the workpiece is positioned and rotated according to the minimum rotation overlap angle of 60°, with a format diameter of 700mm and a repeat positioning accuracy of 0.003° / mm.

[0033] The force feedback module comprises a grinding force sensor, a micro-power actuator and an integrated force control unit.

[0034] The downsampling subdivides the three-dimensional space of the part into three-dimensional grid voxel units of uniform size, and the centroid of the body or the point closest to the centroid is used as the sampling retention point.

[0035] The point cloud registration is to transform the converted workpiece point cloud into a standard coordinate system, and unify the points obtained from different perspectives into a fixed coordinate system to form a complete point cloud. Figure 2 shown.

[0036] The shared busbar equation is estimated using the RANSAC (Random Sampling Consensus) algorithm, which first randomly initializes the model assumptions, then evaluates the deviation of each point in the data set to the model, and distinguishes the inliers and inliers based on the preset tolerance threshold; through repeated iterations, the maximum consensus set is identified, and then the least squares method is used to fine-tune the model parameters to best fit this core point set. In this case, the straight line is selected as the fitting model, the random sampling strategy is used for data segmentation, and a reasonable error limit is set to obtain the mathematical expression of the busbar and its corresponding inliers index.

[0037] The gyroscopic area segmentation is to segment the workpiece at the minimum rotation overlap angle (60° in the embodiment of the present invention). The minimum repeating unit after segmentation is the gyroscopic grinding area unit. The gyroscopic area unit is further divided according to the busbar. When the curvature of the busbar changes suddenly, the grinding gyroscopic unit is divided into different areas. Figure 1 As shown, the outer surface is divided into the outer upper cylinder of the first area 1, the outer upper curved surface of the second area 2, the outer lower curved surface of the third area 3, the outer lower cylinder of the fourth area 4, and the repeated boss of the fifth area 5, and the inner surface is divided into the inner upper cylinder of the sixth area 6, the inner upper curved surface of the seventh area 7, the inner lower curved surface of the eighth area 8, and the inner lower cylinder of the ninth area 9.

[0038] The non-rotational feature recognition is to identify the local features that cannot be overlapped after the workpiece is rotated according to the minimum rotation overlap angle of 60° as the first mounting platform 10 and the second mounting platform 11.

[0039] The path layers are equally spaced and layered based on a series of paths parallel to the XY reference plane (such as Figure 1 The workpiece is divided into plane clusters (as shown in FIG. 12 ), and the boundary line between the plane and the surface contour of the workpiece is defined as a set of path points at this level. In the embodiment of the present invention, the part height is 250 mm, and the plane clusters are spaced 5 mm apart in the Z-axis direction.

[0040] The polishing force sensor monitors, feeds back, adjusts and accurately executes the preset contact force in real time, ensuring that the accuracy of the surface contact force of the tool (robot) in any posture is ≤1N. In the embodiment of the present invention, the accuracy of the surface contact force is 0.5N.

[0041] The micro-power actuator compensates for changes in the surface contour of the workpiece in real time, ensuring stable tracking of the tool and the workpiece surface.

[0042] The integrated force control unit displays the force and stroke measured by the grinding force sensor in real time, and can send instructions to the micro-power actuator to control the grinding tool and automate the operation of peripheral products.

[0043] The described quantitative grinding has a grinding amount of 0.3 mm.

[0044] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0045] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in this field.

Claims

1. An automatic grinding method for electron beam selective area additive rotating body parts, characterized in that: include: Taking 3D scanning point cloud data as input, downsampling and filtering are used to reduce the data volume and remove noise points, and then the ICP algorithm is used for point cloud matching, and the point cloud data is matched to the template point cloud; the cross-section method and envelope block are used to extract the region of interest required for path point calculation, and the octree is used to obtain the path points. Then, the shared busbar equation is used to remove abnormal points for fitting, and finally the path points are obtained; The obtained path points are segmented into convolutional areas, identified as non-convolutional features, and layered with equal spacing between path layers to obtain the planned processing path and supporting parameter package; Using the positioning rotary tooling and force feedback module, the preset contact force can be monitored, fed back, adjusted and accurately executed in real time, so that the grinding tool can follow the changes in the surface contour of the workpiece in real time. At the same time, the force and stroke of the grinding tool can be displayed in real time to complete the quantitative grinding of the parts.

2. The method for automatic grinding of rotating body parts by electron beam selective additive manufacturing according to claim 1, characterized in that: The downsampling subdivides the three-dimensional space of the rotating body part into three-dimensional grid voxel units of uniform size, and the centroid of the body or the point closest to the centroid is used as the sampling retention point.

3. The method for automatic grinding of rotating body parts by electron beam selective additive manufacturing according to claim 1, characterized in that: When the shared bus equation is estimated using the RANSAC algorithm: After randomly initializing the model assumptions, the deviation of each point in the data set to the model is evaluated, and the internal and external points are distinguished according to the preset tolerance threshold; through repeated iterations, the maximum consensus set is identified, and then the least squares method is applied to fine-tune the model parameters to best fit this core point set.

4. The method for automatic grinding of rotating body parts by electron beam selective additive manufacturing according to claim 1, characterized in that: The specific segmentation of the convolution region is as follows: The workpiece is divided at the minimum rotation overlap angle. The smallest repeating unit after division is the spiral grinding area unit. The spiral area unit is further divided according to the busbar. When the curvature of the busbar changes suddenly, the grinding spiral unit is divided.

5. The method for automatically grinding a rotating body part by electron beam selective additive manufacturing according to claim 4, characterized in that: When identifying non-convolutional features, the workpiece is rotated according to the minimum rotation overlap angle, and the local features that cannot be overlapped are identified as non-convolutional features.

6. The method for automatic grinding of rotating body parts by electron beam selective additive manufacturing according to claim 1, characterized in that: When the path layers are equally spaced, the workpiece is divided based on a series of plane clusters, and the boundary between the plane and the surface contour of the workpiece is defined as the set of path points at this level.

7. The method for automatically grinding a rotating body part by electron beam selective additive manufacturing according to claim 1, characterized in that: The positioning and rotating fixture enables the workpiece to be positioned and rotated according to the minimum rotation overlap angle, with a format diameter of ≥650mm and a repeated positioning accuracy of ≤0.005° / mm.

8. The method for automatically grinding a rotating body part by electron beam selective additive manufacturing according to claim 1, characterized in that: The force feedback module includes a grinding force sensor, a micro-actuator and an integrated force control unit, including: The grinding force sensor monitors, feedbacks, adjusts and accurately executes the preset contact force in real time to ensure the accuracy of the surface contact force under any posture of the grinding tool; The micro-power actuator follows and compensates for changes in the workpiece surface contour in real time, ensuring stable tracking between the tool and the workpiece surface; The integrated force control unit displays the force and stroke measured by the grinding force sensor in real time, and sends instructions to the micro-power actuator to control the grinding tool and the automated operation of peripheral products.

Images