【作者】 邓冬梅；

【导师】 周来水；

【作者基本信息】 南京航空航天大学 ， 航空宇航制造工程， 2007， 博士

【摘要】 复合材料因其比强度高、比模量大、材料的刚度和强度可设计等一系列优点,在航空航天领域得到广泛应用,但高昂的价格成了复合材料应用的最大壁垒。国外的研究和应用成果表明数字化技术是降低复合材料构件制造成本、提高构件性能的有效途径。目前国内主要还以手工设计和手工制造为主、自动化程度不高,不仅浪费人力、物力,而且产品质量难以保证,因此有必要对复合材料数字化技术进行研究。优化排样是复合材料构件数字化生产过程中的重要环节。本文在研究各种排样算法的基础上,提出了新的矩形件排样算法、优化算法以及不规则样片的排样算法,并与复合材料铺层排样的特点相结合开发了复合材料铺层排样软件。主要研究内容和创新点如下:矩形件排样不仅适用于矩形样片的排放,也是不规则样片排样的基础。本文在建立矩形件排样数学模型的基础上,介绍了各种常见的定序列矩形件排样算法并分析其特点,提出了一种新的启发式排样算法——最低轮廓线搜索算法。该算法满足“最下最左”条件,克服了其他排样算法对某些排样图不能给出排列的缺点,实验结果表明该算法排样效果好于最低水平线算法和最下最左(BL)算法。利用该算法实现了大量不同规格图纸的集中出图,省时省力,节约纸张20~50%。矩形件排样问题具有图形运算和组合优化两方面的特性,单纯的排样算法只能解决图形运算问题,样片的排放顺序对排样结果同样重要。针对较小规模(一般少于100个图形)的矩形件排样问题,本文提出了模拟退火与最低轮廓线搜索算法相结合的综合优化算法。对于十多个图形的排样,该算法可短时间内求得最优解;对于近百个图形的排样,在排样效果相当的情况下,该算法比其他基于模拟退火的综合优化算法效率提高百倍以上。针对大规模矩形件排样问题,本文提出了蚁群算法与最低轮廓线搜索算法相结合的综合优化算法,该算法比模拟退火与最低轮廓线算法相结合的综合优化算法效率提高十倍以上。不规则图形排样是所有排样研究中的热点和难点。本文将不规则样片简化成多边形进行排样,提出了两种不同的解决方法:一是基于矩形的排样方法,二是直接对多边形进行排样。基于矩形求解不规则样片排样时,将图形运算、矩形件排样算法及交互调整相结合,提出了基于矩形的多边形综合排样算法。通过各种优化组合策略,对单一样片和多种样片进行组合求其最小包络矩形,从而将不规则形状样片排样转化为矩形件排样进行求解。直接排放多边形时,重点研究两个多边形的临界多边形(NFP)的求解。首先对基于倾斜图法的NFP求解法进行了改进和优化,完善了凹、凸两多边形NFP的求解,然后提出了适用于任意两多边形NFP求解的边界绕行法,该方法比基于倾斜图的求解方法适用范围广,计算简单、效率高。根据复合材料构件数字化生产的主要过程,分析总结了复合材料铺层排样的特点,并将优化排样算法与复合材料铺层排样的特点相结合,设计开发了复合材料构件铺层排样软件系统。更多还原

【Abstract】 Composite was widely used in aerospace because of its high strength, modulus, designable rigidity and strength. But high price has becoming an obstacle for its use. The overseas investigation shown that digital manufacture technology is effective for cost saving. However, composite parts are mainly made by hand domestically, it is very slow and quality is not able to be assured. So research on composites digital technology is very urgent. Optimized packing is a key process of composites digital manufacture, and it belongs to NP complete problem.Based on the study of various packing algorithms, the author proposed a new rectangular packing algorithm, a hybrid heuristic optimizing algorithm and a new irregular packing problem, and combined these algorithms with composites plies packing specialties to develop a composites packing software. The main achievements and creative ideas are given below.The rectangular packing problem not only deals with packing of 2D rectangular parts but also plays an important role in solving irregular packing problem. In order to solve the problem of rectangular parts optimal layout with dynamic constraints, a on-line packing algorithm for order-given parts is proposed——Lowest Outline Searching Algorithm based on its mathematical model. It satisfies the requirement of BL (bottom left) condition and overcome the shortcomings of other algorithm for some patterns. What’s more, according to the test compariation with other algorithm such as BL and Lowest herizatinal Line Algorithm, it shown that the algorithm is very effective. And this algorithm is used for many drawings packing and mass printing, which is helpful to save time and paper.Rectangular packing problem mainly consist two fields, which is geometry caculation and compound optimization. Single packing method is only for geometry calculation, but packing order is also very important to optimized layout. In this paper a hybrid optimization algorithm of simullated annealing and Lowest Outline Searching Algorithm was proposed to solve small scale rectangular packing problem and the ways to improve its efficiency was studied. It can get best layout for more than ten rectangles, while for about one hundard rectangles, it is more than 100 times faster than other Simullated annealing based optimization algorithm with almost same material usage.Also ant colony based optimization algorithm was put forward to solve large scale packing problem. The ant colony was used to search a good packing order while Lowest Outline Searching Algorithm is used get the pattern for the ordered parts. Many experiments results show that ant colony based optimization is much faster than simulated annealing based hybrid optimization alrightm; it is very suitable for large scale rectangular packing problem.The irregular packing problem is among the most difficult class of packing problem and has been studied for many years. This paper concentrate on solving the problem from two aspects,that is the method based on graphics calculation and the rectangular packing based on method. For the rectangular based one, graphics calculation, rectangular packing and interactive adjust technologies are combined together to get the enclosed rectangular of irregular parts and packing them. For graphics calculation method, the no-fit polygon algorithm was studied for calculating the relative position for two polygons to save the un-used margin. The existing method based on slope graph is mended and a new boundary slide algorithm was proposed to calculate NFP. The new one is suitable for any two polygons and it is easier than what have been used.Based on the main procedure of composites digital manufacture, the specialty of composite ply packing is analyzed and summed up. According to composites ply packing property and based on the study of theory and algorithms mentioned above,the author designed and developed a software for packing composite layers, which can be integrated with the composite digital manufacture system.更多还原