熔丝沉积成型几何计算关键技术研究

发布时间：2018-02-03 09:18

本文关键词： 3D打印熔丝沉积成型最优放置角度模型分割装箱问题　出处：《西北大学》2016年博士论文　论文类型：学位论文

【摘要】：3D打印是目前的热点研究领域,熔丝沉积成型作为3D打印中最常用的技术之一,以其低成本、易于维护等特性受到人们的广泛关注和重视。本文针对熔丝沉积成型中制造精度不高、制造需添加额外支撑等突出问题,研究其几何计算关键技术,旨在提高打印精度及物品可用性、节约打印耗材和打印时间。主要研究进展包括：(1)提出了模型最优放置角度计算方法,分析受放置角度影响的熔丝沉积成型中的六类制造指标,根据需求建立无约束和带约束的最优放置角度目标函数,利用改进的powell方法求解目标函数。实验结果表明,该算法可有效计算各种需求条件下的物体最优放置角度,优化了制造物体表面精确度,节约了打印耗材和打印时间,避免部分后处理对模型造成的损害。(2)提出一种基于“熔丝成型”的支撑结构生成算法,针对影响熔丝成型的因素,建立四项熔丝打印成型约束,构造最优化目标函数在模型上计算各类支撑结构的最小支撑区域,建立代价最小生成树将支撑结构连接生成完整外部支撑,优化了现有支撑结构生成算法或耗材多或打印过程中不稳固的问题。与传统算法相比,充分考虑成型的最基本要素“熔丝成型”,不仅支撑区域更小,且支撑效果更好,可确保模型的每条打印熔丝均完好成型。实验结果表明,算法生成的支撑结构在打印过程中能稳固支撑模型,在耗材和耗时上均优于传统算法。(3)针对过大物体无法直接放入打印空间的问题,提出一种基于集束搜索的去外部支撑模型分割算法。对模型表面进行分区,根据各区域法向分布计算一组切面划分模型,采用集束搜索方式将所有划分构造成树结构,迭代划分直到所有子模型均为锥体,通过搜索树得到最优划分。针对家具等模型实验验证表明,算法将超出打印空间的模型划分为符合打印空间的分块,均可实现无外部支撑打印与分块组装使用。(4)提出一种去内部支撑的模型分割算法,针对空心物体直接打印时内部存在冗余填充物导致其不能使用的问题,算法通过区域生长在模型表面搜寻不需要内部支撑的分区作为候选分区,利用蒙特卡洛和深度剪枝生成树两种搜索方法,将候选分区不断组合分化,获得最优的无内部支撑分割方案。实验结果表明,算法运用在容器、花瓶、陶俑等模型上,打印的空心物品无内部填充,节省了打印材料与打印时间,可用性大大提高。(5)提出了紧凑低耗的装箱智能优化算法,针对大规模生产制造多模型同时打印需求,将装箱问题解用一组旋转向量和位移向量表示,利用动态邻域的局部学习粒子群算法求解该问题,获得包围盒最小或外部支撑最小的最优装箱方案,用户可一次向打印空间放入更多模型,减少人机交互和额外能耗。针对零件、文物碎片批量打印的实验验证表明,算法能将多模型紧凑装箱置于打印空间内,可有效提高打印空间利用率。
[Abstract]:3D printing is a hot research field at present. As one of the most commonly used technology in 3D printing, fuse deposition molding has low cost. The characteristics such as easy maintenance have been paid more attention to. In this paper, the key technology of geometric calculation is studied in view of the outstanding problems such as the low manufacturing precision and the need to add extra support in the fabrication of fuse deposition molding. The aim of this paper is to improve the accuracy of printing and the availability of objects, and to save printing consumables and printing time. The main research progress includes: (1) A method for calculating the optimal placement angle of the model is proposed. This paper analyzes six kinds of manufacturing indexes in fusing deposition molding influenced by placement angle, and establishes the optimal placement angle objective function of unconstrained and constrained according to the requirement. The improved powell method is used to solve the objective function. The experimental results show that the algorithm can effectively calculate the optimal placement angle of the object under various requirements and optimize the surface accuracy of the manufacturing object. This paper saves printing consumables and printing time, and avoids damage caused by partial post processing to the model. (2) A support structure generation algorithm based on "fuse forming" is proposed, aiming at the factors that affect fuse forming. Four fuse printing constraints are established, and the optimal objective function is constructed to calculate the minimum support region of all kinds of support structures on the model. The minimum spanning tree is established to link the support structure to generate the complete external support. This paper optimizes the existing algorithms for the generation of support structures or the problem of more consumables or instability in the printing process. Compared with the traditional algorithms, the most basic element of forming is fully considered "fuse forming", which not only has a smaller support area. The experimental results show that the support structure generated by the algorithm can support the model stably in the process of printing. In terms of consumables and time consuming, it is better than the traditional algorithm. 3) aiming at the problem that too large objects can not be directly put into print space. A cluster search based model segmentation algorithm is proposed. The surface of the model is partitioned and a set of tangent partition models are calculated according to the normal distribution of each region. All partitions are constructed into tree structure by cluster search, and all submodels are divided into cones iteratively. The optimal partition is obtained by searching the tree. The experimental results for furniture and other models show that. The algorithm divides the model beyond the printing space into blocks that accord with the printing space, and can realize printing and assembling without external support. (4) A model segmentation algorithm without internal support is proposed. In order to solve the problem that the hollow object can not be used because of redundant fillers when printing directly, the algorithm uses regions to search for candidate partitions on the surface of the model that do not require internal support. Monte Carlo and deep pruning tree are used to divide candidate partitions and obtain the optimal segmentation scheme without internal support. The experimental results show that the algorithm is applied to containers and vases. On models such as terracotta warriors and other models, the printed hollow objects have no internal filling, which saves printing materials and printing time, and greatly improves the usability. 5) A compact and low cost intelligent optimization algorithm for packing is proposed. Aiming at the requirement of multi-model printing in mass production, the packing problem is represented by a set of rotation vector and displacement vector, and the local learning particle swarm optimization algorithm of dynamic neighborhood is used to solve the problem. The optimal packing scheme with minimum bounding box or minimum external support is obtained. The user can put more models into print space at a time to reduce man-machine interaction and extra energy consumption. The experimental results of batch printing of fragments show that the algorithm can effectively improve the efficiency of printing space by putting the multi-model compact packing in the printing space.
【学位授予单位】：西北大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP391.73

【参考文献】