【作者】 唐先智；

【导师】 刘飞；

【作者基本信息】 重庆大学 ， 机械制造及其自动化， 2013， 博士

【摘要】 汽车覆盖件是构成汽车车身的重要零部件。通过逆向工程建立的汽车覆盖件数字模型，应用于车身结构设计、车身刚度分析以及误差检测等方面，将传统的从CAD图纸到实物的设计模式改变为基于测量建模的数字化设计模式，极大地提高了产品设计效率，缩短了产品的设计周期，为产品开发设计提供了一条新的途径。然而，一方面逆向工程测量设备的数字化、自动化和测量精度的不断提高，导致了模型的测量数据量的快速增长；另一方面，由于汽车覆盖件存在面积大、厚度薄、产品形状复杂等特征，导致逆向工程的测量建模设计有更高的难度和复杂性，使得还有不少问题需要深入研究。为此，本论文对汽车复杂形体覆盖件逆向工程中的几项关键技术进行了研究。主要研究内容如下：①针对汽车覆盖件的特点，分析逆向工程中不同种类扫描设备的优劣性，对汽车覆盖件数据采集的扫描方法进行了综合研究；并对汽车覆盖件数据预处理技术及网格化技术相关理论进行研究。②汽车覆盖件存在面积大、产品厚度薄等特点带来点云数据采集过程中可能会存在层叠点云现象，若将大量层叠的点云数据处理转化为曲面会消耗大量的时间，同时产生比较大的形状误差。为此，本文提出了一种汽车覆盖件的点云数据分层处理的新的切片算法，该算法首先确定层叠点云数据的位置，然后沿着模型的三维坐标对其进行多层切片，接着对每层切片点云进行采样，确定采样范围内的波峰与波谷点，通过其产生层叠采样点云的分割线（也成为中线），并根据最佳线性的方法确定倒角位置的分隔线段，从而自动将层叠点云数据分离。③汽车覆盖件点云数据获取过程中，因测量原理不同、测量软件的限制以及人为因素等原因会出现点云数据孔洞缺陷等现象，导致点云数据特征不全从而影响后期建立的三维模型失真。对此，本文提出了一种汽车覆盖件点云数据孔洞缺陷修补算法，该算法是在扫描所获得的二维图像和三维空间网格点云数据之间建立映射关系，通过孔洞边界提取方法提取出网格点云数据孔洞的边缘数据，将边缘点映射到二维图片上，再通过二维图片的像素值以及灰度值将映射点分层，从而将三维空间网格点云数据孔洞缺陷进行修复。④复杂的汽车覆盖件建模过程需要协同设计支持，如数据提供方与客户的在线交互、客户对数据信息的全程跟踪以及及时反馈建议信息等。对此，建立了一种基于Internet的远程汽车覆盖件逆向工程中设计网络化共享服务平台。应用该平台能在测量建模过程中实现人机交互，能对整个数字化过程进行在线评估，进一步快速、高效获取优化的点云数据，实现与客户网络化协同设计。最后对本文的工作进行了总结，并展望了下一步的研究方向。上述研究成果，在为东风汽车，力帆汽车等大中型企业服务中得到了验证，取得了明显的经济效益和社会效益。更多还原

【Abstract】 Complex automotive body panels constitute the important parts of the automobilebody structure. Created through the reverse engineering of complex automotive bodypanels digital model used in the design of the body structure, body stiffness analysis anderror detection, the traditional kind of design model from a CAD drawing to change thedigital design model for measurement-based modeling, greatly improve the efficiencyof product design and shorten the product design cycle, and provides a new way forproduct development and design. However, on the one hand the continuousimprovement of reverse engineering digitization, automation and measurement accuracyof the measuring equipment, led to the rapid growth of the model for measuring theamount of data; other hand, due to the presence of auto cover a large area, thin, complexshape products, the measurement modeling using reverse engineering design have ahigher degree of difficulty and complexity, and therefore, there are a lot of issues needto delve into; such, the model has a very complex geometry and topology measurementmodel the existence of the scan point cloud data stacked and the presence of certainareas of the point cloud data missing, as well as the complexity of the modeling processrequires collaborative design support. The papers faimed at reverse engineering of theautomobile cover the keyconduct further research, The main contents are as follows:①Contrary to the characteristics of the auto cover, analysis of the pros and consof the different types of scanning equipment in reverse engineering, conducted acomprehensive study of the auto cover data acquisition scan method;and auto cover datapreprocessing technology and grid technologyrelated theory.②Auto cover large area, thin products, so that it may exist in the point cloud dataacquisition process stacked point cloud phenomenon, while producing the large stack ofpoint cloud data processing will consume a lot of time into curved the relatively largeshape errors. Hierarchical point cloud data preprocessing car cover is particularlyimportant. In this paper, a car cover hierarchical processing point cloud data slicingalgorithm.The algorithm first determines the position of the the cascading point clouddata, and then along the model of the three-dimensional coordinates of its multi-layeredslice, then each slicing points cloud is sampled to determine the point of the crests andtroughs in the sampling range, and generates a point cloud of the dividing line of thelaminate sample through its （also become midline）, and the method according to the best linear determine chamfered position separator segment, which will automaticallythe cascading point cloud data separation.③The auto cover point cloud data acquisition process, because of the differentmeasurement principle, measurement limitations of the software as well as humanfactors because there will be holes in the point cloud data defects phenomenon, leadingto incomplete point cloud data features late to establish a three-dimensional model ofdistortion. In this regard, this paper proposes an automobile cover the hole defects ofpoint cloud data supplement algorithm, the algorithm is to establish the mappingbetween the scan head, two-dimensional images and three-dimensional grid of pointcloud data, through the hole boundary extraction method extracted data of the edge ofthe holes of the grid point cloud data, the edge point is mapped to a two-dimensionalpicture, by a two-dimensional picture of the pixel value and a gradation value of themapping point stratified, so that the point cloud data of the three-dimensional spatialgrid repair.④Complex auto cover modeling process needs to be collaborative design support,such as the data side online interaction with customers, tracking customer datainformation and timely feedback suggestions information. In this regard, theestablishment of an Internet-based remote car cover reverse engineering design networkshared services platform, Application of the platform to achieve human-computerinteraction measurement modeling process, customers can assess the entire digitizationprocess online, to achieve the purpose of the collaboration between enterprises andproduct design.Finally, this work is a summary and outlook of future research directions.The content of the above-mentioned research, validation services for the large andmedium-sized enterprises of Dongfeng Motor, Lifan cars, and achieved remarkableeconomic and social benefits.更多还原