【作者】 吴淑芳；

【导师】 苏铁熊；

【作者基本信息】 中北大学 ， 机械设计及理论， 2014， 博士

【摘要】 机械结构设计过程主要包括概念设计、基本设计、详细设计三个阶段。每个阶段都存在着“设计-优化-再设计”的循环，计算机辅助设计（CAD，Computer Aided Design）和计算机辅助工程（CAE，Computer Aided Engineering）技术的不断发展，使机械结构设计的效率和质量有了较大提高。然而，随着产品的市场竞争日益加剧，机械产品设计要求不断提高，高效率、高精度、低成本的设计成为了设计者追求的目标。因此，机械产品设计的关键是要在短周期内实现低成本高质量的结构设计。本文在研究现有机械结构设计方法的基础上，研究了基于网格数据的CAD有限元模型到CAE模型的转换方法，建立了以参数化设计理论为基础，共享数据库为支承的双向相关模型转换机制，推导了双向渐进结构优化（BESO，The Bi-direcfionalEvolutionary Structural Optimization）法进行机械结构拓扑优化的数学模型，提出了基于精英保留策略的遗传双向渐进结构优化（GBESO，Genetic Bi-direcfional EvolutionaryStructural Optimization）法寻优策略，建立了基于特征和约束的优化结果模型重构方法，为实现高效率、高精度、低成本的机械结构设计提供了理论指导和方法依据。具体研究了以下内容：（1）在分析机械结构设计流程的基础上，研究了参数化设计方法，确定了基于特征的参数化CAD模型建立原则；探讨了数据传输的实现方法，建立了CAD模型及其相关数据的共享机制；通过分析CAD模型到CAE模型的转换机理，提出了一种有限元网格模型转换方法及其模型转换双向相关机制，实现了结构设计过程中快速、准确的CAD/CAE响应，为后续的优化设计奠定了基础。（2）分析了渐进结构拓扑优化方法的优化理念，推导了BESO法进行机械结构拓扑优化轻量化设计、应力优化设计、刚度最大化设计的数学模型，针对BESO法受其优化准则（Optimality Criteria，OC）的影响优化速度较慢，且优化结果不一定是原结构的真正最优解的问题，引入遗传算法（GA，Genetic Algorithm）的精英保留策略，提出了遗传双向渐进结构优化（GBESO）法寻优方式及其关键问题的解决方法，通过与现有拓扑优化方法寻优结果进行比较，验证了该寻优算法的稳定性和高效性。（3）分析了基于有限元法的拓扑优化普遍存在的棋盘格现象和锯齿状边界等数值问题，提出了基于特征和约束的拓扑优化结果几何模型的自动重构方法。通过提取工程数据库中拓扑优化结果数据，以材料去除区域的点云数据为基础，进行面向设计制造的数值处理；通过特征提取和约束识别技术进行特征拟合，构建材料去除区域的CAD模型，并与原模型进行布尔减操作，自动重构CAD模型。以弹槽隔板为例验证了该方法的可行性，解决了拓扑优化结果CAD模型手工绘制数据量大、时间长且精度无法保证的问题，为机械结构快速优化设计提供了方法依据。（4）利用所研究的基于CAD/CAE快速响应的结构拓扑优化理论和方法，对弹鼓结构三维模型进行参数化设计，根据关键件拓扑优化结果及弹鼓装配要求，实现了优化后弹鼓结构的快速参数化模型更新，通过调用数据库文件对比分析了优化前后的弹鼓力学性能，优化后的模型在刚强度满足设计要求的前提下，质量减少了17.9%。结果表明，将该方法应用与连续体的优化设计中，对于改进设计方案，降低结构重量，实现轻量化设计具有重要意义。更多还原

【Abstract】 Conceptual design,basic design and detailed design are three stage of the process ofmechanical structure design. Each of them contains the circulation of de-sign-optimize-redesign. With the continuous development of CAD and CAE technology,efficiency and quality of the mechanical structure design have a large increase.However,due to increasing competition in product markets,the requirements of mechanicalproduct design are constantly improved and the design of high efficiency,high precision andlow-cost design become the goal of the designers. Therefore,the key of mechanical productdesign is to achieve a low-cost and high-quality design in a short term.On the basis of the existing research on the mechanical structure design method in thispaper,the conversion method between CAD and CAE finite element model based on meshdata was studied,Bi-directional relevance model transformation mechanism based on thetheory of parametric design and shared data was established,mathematical model of themechanical structure topology optimization with BESO (Bi-directional EvolutionaryStructural Optimization) was deduced,the GBESO (Genetic Bi-directional EvolutionaryStructural Optimization) optimization strategy based on elitist preservation was putforward,reverse-engineered reconstruction method of optimized outcome model based onthe characteristics and constraints was set,and the theoretical guidance and method toachieve the high efficiency,high precision and low-cost mechanical structure design wereprovided. The specific content as follows:(1) Researching on the theory of parametric design, parametric CAD modeling principleswas built based on characteristics by the analysis of mechanical structure design process.Implementation method of data transmission was discussed and sharing mechanism of CADmodel and its associated data was established. By analyzing the transformation mechanism of CAD model to CAE model, conversion method of finite element model was found andBi-directional relevance model transformation mechanism was established. Fast and accurateCAD/CAE response in structural design process was achieved and the foundation forsubsequent optimization was laid.(2) By analyzing the optimization concept of evolutionary structural topologyoptimization, mathematical model of mechanical structure topology optimization with BSEOin lightweight design, stress optimized design and stiffness maximize design deduced. For theproblems that slower optimum speed of BESO under the influence of its optimization criteriaand optimal results were not necessarily the truly optimal solution of original structure. Byintroducing elitism strategy of GA, the optimize method and the solution on key issues ofGBESO were obtained. Through the comparison to optimal results with existing topologyoptimization methods, the stability and efficiency of the optimize method were verified.(3) By analyzing ubiquitous numerical problems in topology optimization based on thefinite element method such as checkerboard phenomenon and serrated border, introducingreverse engineering technology, extracting topology optimization results data in projectdatabase, and based on point-cloud data in material removal area, numerical processingmethods for design and manufacture were proposed.Feature Fitting was achieved with the technology of feature extraction and constraintrecognition. CAD model in material removal area was built for boolean subtraction withsource model. The automatic reconstruction of the geometric model with topologyoptimization results was achieved. Taking groove division as an example to verify thefeasibility of the method,the problems that huge data,much time consuming and uncertainprecision in hand-drawn CAD model of topology optimization results were solved,methodsbasis in fast optimal design of mechanical structures were provided.(4) Based on the structural topology optimization theory and methods of CAD/CAErapid response,3-D model parametric design of magazine structure was achieved. Accordingto the topology optimization results of the components and the assembly requirements ofmagazine, quick update of the parametric model of optimized magazine structure was realized.By calling the database file, magazine mechanical properties before and after optimizationwere comparative analyzed. Under the condition that meeting the design requirement ofstiffness and strength, optimized model has a17.9%reduction in quality. The results showthat this method applied to optimal design of continuum has a great significance on improving design scheme, reducing structural weight and realizing lightweight design.更多还原