I. Introduction Traditional product design is generally a process of “going from scratch.†Designers first conceive the product's appearance, performance, and general technical parameters, then use CAD to build a three-dimensional digital model of the product, and finally transfer the model to the manufacturing process. Complete the entire design and manufacturing cycle of the product. Such a process can be called "forward design." Reverse engineering is a process of “from being to nothingâ€. According to the existing product model, the design data of the product is reversed, including design drawings and digital models. The reverse engineering of professional software Surfacer, ICEM, CopyCAD and RapidForm, etc., these software is very suitable for processing a large number of scanned point cloud data. For example, the laser scanning of a car's appearance can get about 300,000 measurement points. The digital model can be obtained through professional Surfacer software and achieve the desired effect. At the same time, we have also explored the application of UG in reverse engineering and gained some experience in the process. The following details are as follows. Second, the input of data points With UG software for reverse engineering, most of the measuring equipment used are contact-type manual three-line striping machines, mainly for profile, contour and characteristic line measurements, measuring data points are not many, UG processing is also relatively easy. However, there are as many as 300,000 data points measured by laser scanning in the vehicle model in this paper. It is very difficult to input UG for so many data points. Therefore, we have eliminated the noise and sparsity of point cloud data in the Surfacer software. . In order to accurately maintain the original feature points and contour points, we generally construct the outline and feature lines, and import the UG with point cloud data, as shown in FIG. 1 . Third, through the point of construction curve 1. In the process of connection, it is generally connected to the characteristic line point first and then the profile point. Before connecting, there should be a reasonable plan to determine how to divide the surface according to the shape and characteristics of the car, in order to determine which points should be connected, and be aware of the method of the subsequent facet, the error of the connection is generally controlled at 0.4mm the following. 2. Commonly used are straight lines, arcs, and splines, of which splines are the most commonly used. The “through point†method is generally selected, and the order is preferably 3 steps, because the higher the order, the poorer the flexibility is, that is, the deformation is difficult, and the subsequent processing speed is slow, and the data exchange is difficult. 3. Because there are errors in the measurement and the outer surface of the model is not smooth, it is necessary to adjust even when the spline is not smooth, otherwise the curved surface is not smooth. One of the commonly used methods for adjustment is Edit Spline. Commonly used Edit pole options include moving, adding control points, and controlling pole movement in a certain direction to facilitate editing of splines. In addition, the curves are divided and bridged. (bridge) and Smooth spline are also often used. In short, a lot of adjustment work needs to be done before the plane is generated, and the curvature can be used to analyze the transition line to ensure the quality of the curve, as shown in FIG. 2 . Fourth, the construction surface Because the car body requires a smooth shape and a smooth outer surface, when constructing a curved surface, it must be divided into a number of curved surfaces, in particular, to ensure that the surface and the surface can be tangentially continuous or continuous in curvature, so as to form a seamless mark. Surface. In addition, when constructing a curved surface, it is necessary to select an appropriate construction method according to the specific situation. 1. The method of constructing the surface (1) The most common construction method is the Through Curve Mesh, which not only guarantees the continuity of curvature at the boundary of the surface, but also controls the curvature (peripheral) of the boundary around it. The Through curves can only guarantee the curvature of the two sides. (2) The use of more nxn command, you can dynamically display the surface is being created, you can also increase or decrease the definition of the curve string, and the surface will also change. Similarly, G0, G1, and G2 with adjacent faces can also be kept continuous. (3) When constructing a surface, you often encounter three-sided surfaces and five-sided surfaces. Generally, a curve is formed by converting a three-sided surface into a four-sided surface, or extending a boundary line, and converting the five-sided surface into a four-sided surface to reconstruct the surface. Among them, on the surface, the curve on the surface and the trim are two commands commonly used, as shown in FIG. 3 . (4) After the outer surface is constructed, mirror processing is performed. At the center of the surface, there is often a bulge, which appears to be a surface that is not smooth. Generally, the center of the surface is cut off and smoothed by bridging. (5) When constructing a curved surface, there is often a "crease" between the two surfaces, and the curved surface is not smooth, mainly because the two surfaces are tangent to each other. To solve this problem, you can use the Thick curve mesh to set the boundary tangency continuous option. You can also select the match edge (NX3) option after the surface is constructed so that the boundaries of the two surfaces can be matched to make the curvature continuous. In addition, the match edge command automatically extends the boundary of one surface to the boundary of another surface even if the two surfaces do not meet. (6) When constructing single and relatively flat surfaces, using point cloud directly will be more convenient and accurate. Sometimes the gap between faces needs to bridge to ensure a smooth transition of the surface. When the surfaces intersect, filleting also makes the two surfaces smoothly transition. 2. The construction of the surface should pay attention to a few issues (1) The most crucial aspect of the facet is to grasp the features of the sample, but also need to be concise, the surface area is as large as possible, and the number of sheets is not too much. In addition, we must also rationalize facets to improve modeling efficiency. (2) In the process of building a surface, it is sometimes necessary to add some lines in order to construct the surface, and the lines and facets need to be alternated frequently. After the surface is built, it is necessary to check the surface of the error, the general measurement point to surface error, the error should not exceed 1mm. (3) The order of the construction surface should be as small as possible, and it is generally recommended to be 3 steps. Because high-order slices make it less likely to successfully exchange data with other CAD systems, other CAD systems may not support high-order surfaces. When the order is high, the sheet is relatively rigid, and the surface is far from the pole, which is inconvenient when editing the surface at the pole. In addition, the low order is also conducive to increase some features such as rounded corners, slope and thickening, which is conducive to the next programming process, to increase the speed of the subsequent generation of NC machining tool rails. V. Structural entities If the outer surface is stitched into a whole body and then the bottom surface of the vehicle is a closed body, it becomes an entity. However, since the curvature of the bottom surface of the car changes greatly, the shelling command cannot be achieved. Therefore, first of all, it is necessary to first offset each sheet of the outer surface, and then construct a cross-section of the inner and outer surfaces. Finally, the resulting cross-section and the inner and outer surfaces are stitched together to make it a closed sheet, and thus automatically Converted into entities, this process generally includes the following four aspects. 1. The value of the surface Use the Offset command to select multiple faces or select all at the same time to improve efficiency. The bias value of each piece on the outer surface of the trolley is shown in Fig. 4. The arrow in FIG. 4 indicates the direction of the bias value. If the arrow is reversed, it is sufficient to input a negative value. It is not possible for any surface to achieve a bias value. Generally, there are several reasons why bias values ​​cannot be achieved: (2) The partial-value distance is too large and causes self-intersection after the partial value, which leads to the failure of the partial value (in this article, the partial bias value of the car body is 2 mm); (3) The quality of the biased surface is not good, and there are partial ripples. Only the modified surface can be modified and then the partial value can be adjusted. The above four situations may be encountered when constructing a curved surface. Learn to analyze why. 2. Surface stitching The surface after the offset value also needs to be trimmed or complemented, and the inner surface is built with various surface editing methods, and then the inner surface and the outer surface are sewn. When suturing, suturing often fails, generally the following possibilities. (1) When sewing, there are too many sutures. Only a few slices should be sewn at a time, requiring multiple sutures. (2) The stitching tolerance is less than the distance between adjacent edges of the two stitched surfaces. In the face of such problems, the suture is generally performed after the suture tolerance is increased. (3) After the two surfaces are extended, they cannot meet and the edge shapes do not match. If the sheet is not B-curved, the sheet must first be converted to a B-surface so that it matches the edge of the corresponding other sheet and then stitched. (4) There are subtle malformations or other geometric imperfections on the edges. Local magnification can be performed, surface analysis can be performed to check for geometric defects, and if a geometric defect does exist, the sheet is modified or reconstructed and then re-stitched. 3. The effectiveness of the suture Finally, it should be noted that although the suturing command was executed, the computer did not give an error message. It seems that the suture was successful. It may not be true. Some films will show bright or bright lines, and even gaps when zoomed in. Therefore, after suturing is completed, the effectiveness of the suture must be checked immediately. If there is a bright or bright line on the suture, it means that there is no suture in this area. The suture must be canceled and the suture must be re-stitched. Otherwise, it will bring difficulties to the subsequent solid modeling, but only Highlighting the surroundings indicates successful suturing, as shown in Figure 5. 4. Generate entity Combining the inner and outer surfaces and cross-sections into a closed body, the bodies will automatically be converted into solid bodies, as shown in Figure 6. Sixth, summary In short, using UG software to do reverse design is time consuming, and it is not as convenient as professional reverse software (such as Surfacer, Rapidform, etc.) in the processing and construction of the point cloud. However, UG does some ideas and methods for reverse design, such as how to structure lines and facets, how to plan according to product features, and how to divide products. It is also helpful to use reverse engineering of professional reverse software. In fact, this is also true. Do the key to reverse design. Straight Hose Insert,Hose Couplings,Hydraulic Hose Ferrule,Carbon Steel Hydraulic Ferrule Jiaxing Master Fitting CO.,LTD , https://www.master-fittings.com
Figure 1 Input data 
Figure 2 The constructed curve 
Figure 3 Constructed triangular face
After building the outer surface, you need to build a solid digital model. If the model is relatively simple and the curvature does not change much, stitch them together into a whole, then use the thickening instructions to create the entity, but in most cases it is impossible to achieve, especially for the models in this article. 
Figure 4 The offset of the surface
(1) Since the curvature of the curved surface itself is too large, the normal surface has a sudden change in normal line;
(4) There are still some surfaces that look good, but they can't be biased. In this case, you can use Extract Geometry to convert to B surface and then perform the bias value. 
Figure 5 Sewing of inner and outer surfaces and cross-sections 
Figure 6 Entity
It describes how to use UG to do reverse design in reverse engineering. In general, you enter the measured data point cloud first, connect the data points, and then construct the surface. Finally, the process and method of constructing the body as a solid body are introduced.