【作者】 田颖；

【导师】 王太勇；

【作者基本信息】 天津大学 ， 机械制造及其自动化， 2010， 博士

【摘要】 日益增加的生命终端EOL(end-of-life)产品给世界环境与资源带来前所未有的压力。围绕EOL产品的拆卸与回收技术正成为全球性的研究热点。而许多机电产品不能被很好的回收正是源于其没有得到很好的拆卸。本论文以生命终端机电产品为研究对象,对可拆卸性建模和多目标、多工况下的最优拆卸路径决策等问题进行了研究,具体做了以下几方面的工作:可拆卸性建模,即合理表达从产品到组件之间所有可行的拆卸路径。这在数学上属于NP-Complete型难题。当拆卸系统零件数稍微增加,普通建模算法便会产生组合爆炸问题。为此本文提出两种降低拆卸系统规模的方法。方法一:从零件的结构特征入手,对其重新分类,给出算法流程。提出可拆卸性建模就是确定限位结构件单元LSU间的拆卸层级关系。其他类型的零件均可作为附件压缩到限位结构件单元列表中。从而显著降低参与拆卸建模的零件数量。方法二:通过建立拆卸模块的方法来降低拆卸系统的复杂度。在具体实现上,以模糊聚类算法为理论依据,首先,利用工程上对各种形式连接强度的公式定义拆卸系统内部任意两个零件间的拆卸隶属度函数,进而生成拆卸隶属度矩阵。然后,依据定位连接件的拆卸层级关系定义修正系数,对拆卸隶属度矩阵进行全局拆卸关系修正。通过n次传递闭包运算最终生成模糊等价矩阵。以此为依据可以得到λ-截矩动态聚类图。将大型拆卸系统划分成有限的拆卸模块。这样,以限位结构件单元和拆卸模块做为基本拆卸单元可以有效解降低拆卸建模的难度。EOL产品的损伤程度和拆卸过程中的不确定因素等都会直接影响到最优拆卸序列的决策。因此对产品和拆卸过程信息的管理和调用,是拆卸序列规划研究的重要内容。文章首先对拆卸知识进行分类,给出各类信息的存储形式,定义信息之间的关联和调用规则。重点对几种主要影响决策算法的随机变量进行了参数化定义。这些对拆卸路径优化算法的系统化提供有力的数据保证。最后,建立了基于知识的拆卸Petri网模型(KBDPN)全面清晰地表达可拆卸序列的嵌套关系和相关的所有信息。KBDPN中的基础Petri网结构表达了所有几何可拆卸路径,而对非几何拓扑信息如拆卸时间、回收成本等则以参数的形式独立地作用到对应的拆卸操作(触发)或零件(库所)中。从而可以同时实现多目标、多工况下各种EOL产品拆卸系统的路径优化。更多还原

【Abstract】 The increasing of end-of-life(EOL) produces brings unprecedented pressure to the ecological resources and environment all over the world. The technologies about disassembly and recycling are becoming research focus. Most electro-mechanical products cannot be recycled only because they cannot be disassembled well.Aiming at the EOL electric-mechanical product, this dissertation focuses on the research on disassembly modeling and the decision of the optimal disassembly sequence under multi-objective and multi-working- condition. The main contents are as following:Disassembly modeling, that is to express all disassembly sequences from the product to each component. As a NP-Complete problem, with the little quantity increasing of product’s components, combination explosion will be happen in common modeling algorithms. Two methods are proposed to solve above problem.Method 1: starting with the part structure, parts are reclassified and defined with the generating algorithm flows. In this means disassembly modeling is equal to finding the disassembly level relationships. And other types of part are compressed to the LSU list as its accessories, which can decrease the scale of disassembly system in modeling process greatly.Method 2: the quantity in disassembly system can be decreased by building the disassembly modules. In practice, based on the fuzzy clustering algorithm, at first, define the membership function of any two parts in disassembly system, using the connection strength functions in engineer. The disassembly membership matrix is then gained. Next, modify reference are defined based on the disassembly level relationship of position connection parts unit (PCU),with which the disassembly membership matrix is modified in global range. As result, the fuzzy equivalence matrix can be obtained after n-times of transitive closure operations. Judging by this, aλ-cut moment dynamic clustering graph is generated to divide disassembly system into several modules.The damage degree of EOL product and uncertain factors in operation can directly influence the decision of the optimal disassembly sequence. So it is important to manage and use information about the products and the disassembly process. Firstly, the knowledge are classified and be stored correspondingly. Secondly, it is also defined the relationship and the rules among these knowledge. Several kind of key random variables involved in decision algorithm are also defined. These data give a guarantee to the optimal algorithm for disassembly sequence.At last, A KBDPN (knowledge-based disassembly Petri net) model is built to express the nested relations of disassembly system with all topology information, in which the base Petri net model can express the geometrical information about disassembly system. As for the non-geometrical topology information (such as the disassembly time, recycle benefits et al.) is effected in form of references by the disassembly operation (trigger) or part (place) independently. In this way, the disassembly sequence planning of EOL products can be realized under multi-objective and multi-working-condition.更多还原