【作者】 王彩红；

【导师】 查建中； 陆一平；

【作者基本信息】 北京交通大学 ， 载运工具运用工程， 2007， 博士

【摘要】 布局问题有着广泛的工程应用背景，尤其对一些复杂机械产品的设计，布局设计关系到它们主要性能指标的优劣乃至整个设计的成败。由于复杂产品的布局问题具有建模和求解的双重复杂性，特别是布局过程中还涉及大量难以用计算机模拟的人类专家的认知活动，目前工程实用性较低。依据智能工程理论，本文提出了布置设计的复合知识模型、基于元系统的分布式协同设计、人机协同和多机并行方法，并将其集成于基于分布式知识多机并行人机协同的智能布置设计系统。本研究旨在对复杂工程布置设计问题的求解进行探索，使布置设计更加贴近工程实际。首先，总结和分析了布局问题的国内外研究现状，指出了当前的困难与不足。由于复杂产品布置设计包括不同领域、不同阶段的知识很难用单一的知识模型进行表达，本文通过对布置设计知识的分类，结合人工智能的思想，采用复合知识模型表示方法对方案实例、规则知识、约束知识和求解知识进行表达。并且建立满足智能布置设计不同阶段以及多机并行运算的待布置部件简化模型、布置空间、布置实例以及约束等知识的复合知识表达模型，实现了信息的交换和共享。由于复杂产品布置设计过程中的知识资源是大量的、分散的，为保证布置设计的各个阶段的分布式协同，文中提出以元系统为集成核心建立分布式智能系统，该方法使系统的结构更加灵活。由于复杂产品布置设计过程还涉及大量无法用计算机语言来描述的专家的认知活动，文中提出分布式环境下人机协同的智能布置设计方法，利用人-算法-知识的协同进行求解，即充分发挥人和机器各自特长，建立了基于B／S和C／S混合结构的人机协同的分布式智能布置设计平台。为提高布置算法的效率，充分利用各种计算资源，提出将多机并行计算和模拟退火相结合，使并行的模拟退火算法能提高求解的效率、更贴近工程的要求。利用三维可视化的人机交互界面动态显示算法布置过程，并支持人工干预，增强了算法的实用性。最后，以坦克动力舱布置设计为例，对本文提出的方法进行工程实例验证。文末对全文的工作进行了总结并对以后的研究方向作了展望。更多还原

【Abstract】 Packing problems have broad applications. As to some complex mechanical products, packing design is heavily related to their main performance even can determine their final success and failure. Packing problems have high complexity in terms of modeling and computing, and especially a lot of expert cognition that is formed during the packing process is difficult to be simulated by computer. So, it needs researching process to solve project problems. Under the guidance of the intelligence engineering, compound knowledge model of layout, distributed cooperation based on meta system, human computer interaction and parallel computing are introduced, they are integrated in intelligent layout design system of human computer interaction based on distributed knowledge environment. Aim for searching a method to solve the project problem of complex packing and making the packing design close to project.First of all, difficulty and deficiency of existing packing methods were found out based on summarizing and analyzing current research status both inland and abroad.As there is much knowledge in packing problem of complex mechatronic, this knowledge involving in different domain and phase are difficult to express with single model. The layout design knowledge of complex products is divided into layout cases, rule knowledge and constrain knowledge, they are expressed with compound knowledge model. Compound knowledge of simplified part, layout space, case and constrain knowledge satisfied with different subsystem and parallel computing of designing the intelligent layout process. This knowledge can be exchanged and shared.Knowledge of layout is saved in separate subsystems. In order to distributed cooperation in different phase of layout design, distributed intelligent system based on meta system is built. Meta system makes structure of whole intelligent system flexible.A lot of expert cognition formed during the packing process is difficult to be simulated by computer. A human-computer cooperation method for packing problems is presented to make use the speciality of human and computer. A web-based layout design system structure of human computer interaction is represented also, which combines the merit of C/S and B/S.In order to improve efficiency of computer, parallel algorithm and simulated annealing algorithm are integrated in parallel cluster environment. Method of scheduling computing resource by the grid strategy make this algorithm can improve layout result and be used to engineering problem. User interface shows 3D packing process and result. Through the user interface layout result can be modified by designer. At last, intelligent layout design based on distributed knowledge for solving the layout problem of armor power cabin was introduced. In the end of the dissertation, conclusions are drawn and the future research directions in the field are suggested.更多还原